diff --git a/.gitignore b/.gitignore new file mode 100644 index 000000000..14203f278 --- /dev/null +++ b/.gitignore @@ -0,0 +1,18 @@ +# Exe +*.exe +*.out + +# Obj +*.o +*.obj + +# Editors +.vscode/ +*.code-workspace +*.swp + +# Git +.git/ + +# Build +build/ diff --git a/CMakeLists.txt b/CMakeLists.txt new file mode 100644 index 000000000..c4bac7c18 --- /dev/null +++ b/CMakeLists.txt @@ -0,0 +1,113 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +cmake_minimum_required (VERSION 3.13) + +project(oneMKL VERSION 0.1.0 LANGUAGES CXX) + +# Default options. Override by defining in command +## Build type +option(BUILD_DEBUG "" OFF) + +if(BUILD_DEBUG) + set(CMAKE_BUILD_TYPE "Debug") +else() + set(CMAKE_BUILD_TYPE "Release") +endif() + +option(BUILD_SHARED_LIBS "Build dynamic libraries" ON) +## Backends +option(ENABLE_MKLCPU_BACKEND "" ON) +option(ENABLE_MKLGPU_BACKEND "" ON) +if(ENABLE_MKLCPU_BACKEND) + option(ENABLE_MKLCPU_THREAD_TBB "" ON) +endif() +## Testing +option(BUILD_FUNCTIONAL_TESTS "" ON) +## Documentation +option(BUILD_DOC "" OFF) + +# Set compilation flags +set(CMAKE_CXX_STANDARD 11) + +set(CMAKE_CXX_REQUIRED ON) +set(CMAKE_CXX_EXTENSIONS OFF) +set(CMAKE_EXPORT_COMPILE_COMMANDS ON) + +# Set output directories for the project +set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin) +set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib) +set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/lib) + +# Add CMake Finders +add_subdirectory(cmake) + +# Include general cmake config files +list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake") + +# Find necessary packages +find_package(Compiler REQUIRED) + +# Add source directory and output to bin/ +add_subdirectory(src bin) + +# Functional Tests +if(BUILD_FUNCTIONAL_TESTS) + enable_testing() + add_subdirectory(tests) +endif() + +if(BUILD_DOC) + add_subdirectory(docs) +endif() + +install(DIRECTORY include/ + DESTINATION include + COMPONENT Devel +) + +include(CMakePackageConfigHelpers) +write_basic_package_version_file( + "${CMAKE_CURRENT_BINARY_DIR}/oneMKLConfigVersion.cmake" + VERSION ${PROJECT_VERSION} + COMPATIBILITY AnyNewerVersion +) + +export(EXPORT oneMKLTargets + FILE "${CMAKE_CURRENT_BINARY_DIR}/oneMKLTargets.cmake" + NAMESPACE ONEMKL:: +) +configure_file("${PROJECT_SOURCE_DIR}/cmake/oneMKLConfig.cmake" + "${CMAKE_CURRENT_BINARY_DIR}/oneMKLConfig.cmake" + COPYONLY +) + +set(config_package_location "lib/cmake/${PROJECT_NAME}") +install(EXPORT oneMKLTargets + FILE oneMKLTargets.cmake + NAMESPACE MKL:: + DESTINATION ${config_package_location} +) +install( + FILES + "${PROJECT_SOURCE_DIR}/cmake/oneMKLConfig.cmake" + "${CMAKE_CURRENT_BINARY_DIR}/oneMKLConfigVersion.cmake" + DESTINATION ${config_package_location} + COMPONENT Devel +) diff --git a/CODE_OF_CONDUCT.md b/CODE_OF_CONDUCT.md new file mode 100644 index 000000000..71148e594 --- /dev/null +++ b/CODE_OF_CONDUCT.md @@ -0,0 +1,84 @@ +# Contributor Covenant Code of Conduct + +## Our Pledge + +We as members, contributors, and leaders pledge to make participation in our community a harassment-free experience for everyone, regardless of age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, religion, or sexual identity and orientation. + +We pledge to act and interact in ways that contribute to an open, welcoming, diverse, inclusive, and healthy community. + +## Our Standards + +Examples of behavior that contributes to a positive environment for our community include: + +* Demonstrating empathy and kindness toward other people +* Being respectful of differing opinions, viewpoints, and experiences +* Giving and gracefully accepting constructive feedback +* Accepting responsibility and apologizing to those affected by our mistakes, and learning from the experience +* Focusing on what is best not just for us as individuals, but for the overall community + +Examples of unacceptable behavior include: + +* The use of sexualized language or imagery, and sexual attention or + advances of any kind +* Trolling, insulting or derogatory comments, and personal or political attacks +* Public or private harassment +* Publishing others' private information, such as a physical or email + address, without their explicit permission +* Other conduct which could reasonably be considered inappropriate in a + professional setting + +## Enforcement Responsibilities + +Community leaders are responsible for clarifying and enforcing our standards of acceptable behavior and will take appropriate and fair corrective action in response to any behavior that they deem inappropriate, threatening, offensive, or harmful. + +Community leaders have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, and will communicate reasons for moderation decisions when appropriate. + +## Scope + +This Code of Conduct applies within all community spaces, and also applies when an individual is officially representing the community in public spaces. Examples of representing our community include using an official e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event. + +## Enforcement + +Instances of abusive, harassing, or otherwise unacceptable behavior may be reported to the community leaders responsible for enforcement at onemkl.maintainers@intel.com. All complaints will be reviewed and investigated promptly and fairly. + +All community leaders are obligated to respect the privacy and security of the reporter of any incident. + +## Enforcement Guidelines + +Community leaders will follow these Community Impact Guidelines in determining the consequences for any action they deem in violation of this Code of Conduct: + +### 1. Correction + +**Community Impact**: Use of inappropriate language or other behavior deemed unprofessional or unwelcome in the community. + +**Consequence**: A private, written warning from community leaders, providing clarity around the nature of the violation and an explanation of why the behavior was inappropriate. A public apology may be requested. + +### 2. Warning + +**Community Impact**: A violation through a single incident or series of actions. + +**Consequence**: A warning with consequences for continued behavior. No interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, for a specified period of time. This includes avoiding interactions in community spaces as well as external channels like social media. Violating these terms may lead to a temporary or permanent ban. + +### 3. Temporary Ban + +**Community Impact**: A serious violation of community standards, including sustained inappropriate behavior. + +**Consequence**: A temporary ban from any sort of interaction or public communication with the community for a specified period of time. No public or private interaction with the people involved, including unsolicited interaction with those enforcing the Code of Conduct, is allowed during this period. Violating these terms may lead to a permanent ban. + +### 4. Permanent Ban + +**Community Impact**: Demonstrating a pattern of violation of community standards, including sustained inappropriate behavior, harassment of an individual, or aggression toward or disparagement of classes of individuals. + +**Consequence**: A permanent ban from any sort of public interaction within the community. + +## Attribution + +This Code of Conduct is adapted from the [Contributor Covenant][homepage], version 2.0, +available at https://www.contributor-covenant.org/version/2/0/code_of_conduct.html. + +Community Impact Guidelines were inspired by [Mozilla's code of conduct enforcement ladder](https://github.com/mozilla/diversity). + +[homepage]: https://www.contributor-covenant.org + +For answers to common questions about this code of conduct, see the FAQ at +https://www.contributor-covenant.org/faq. Translations are available at https://www.contributor-covenant.org/translations. diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md new file mode 100644 index 000000000..c9a7cae48 --- /dev/null +++ b/CONTRIBUTING.md @@ -0,0 +1,308 @@ +# Contributing Guidelines +If you have improvements, new libraries integrated under oneAPI Math Kernel Library (oneMKL) interfaces, or new interfaces to contribute to the oneMKL Specification, please send us your pull requests! For getting started, see GitHub [howto](https://help.github.com/en/articles/about-pull-requests). + +The current guidelines are a work in progress. + +## Pull Request Checklist + +Before sending your pull requests, ensure that you follow this checklist. + +* If you are contributing a new interface, refer to the [library functionality guidelines](CONTRIBUTING.md#library-functionality-guidelines). It is strongly advised that you first open a [RFC issue](CONTRIBUTING.md#RFC-issue) with a detailed explanation of the expected use cases. + +* Ensure that your code includes proper documentation. + +* Ensure that the changes are consistent with the [coding style](CONTRIBUTING.md#coding-style). + +* Ensure that [unit tests](CONTRIBUTING.md#unit-tests) pass. Include logs from tests as attachments to the pull request. + +## Library Functionality Guidelines + +oneMKL focuses on the following criteria: + +1. *Performance*: Functionality that has highly optimized and extensively parallelized routines for applications that require maximum performance. + + This means that for a new primitive you should demonstrate that it brings visible performance improvement to some applications. + +2. *Generality*: Functionality is useful in a wide range of applications. + + This implies that when introducing a new function, its API needs to be general enough to be integrated into multiple applications that have similar functionality and its interface can support multiple hardware (HW). + +3. *Complexity*: Functionality that is not trivial to implement directly or by combining existing primitives. + +For the new API to become a part of the open source project, it should be accepted as part of [oneMKL spec](https://spec.oneapi.com/versions/latest/elements/oneMKL/source/index.html). + + +### RFC Issue + +Open an Request For Comment (RFC) issue when contributing new interfaces. In the RFC, please provide the following details: + +* Describe how the new interface meets [library functionality guidelines](CONTRIBUTING.md#library-functionality-guidelines). + +* The definition of the function including the interface and semantics, and how this interface will be extendable for different HW implementations. + +* What exisitng libraries have implementation of this function and can be used under the oneMKL interface. + +## Coding Style + +The general principle is to follow the style of existing / surrounding code. If you are in doubt, use the `clang-format`: +```sh +clang-format -style=file -i foo.cpp +``` +This formats code using the `_clang_format` file found in the oneMKL top level directory. + + +### GN: General Naming +* **GN1:** Use snake_case for all type names: classes, structures, enums, template type arguments, type aliases. + +* **GN2:** Use snake_case for all variables (global, local, files, function parameters), global and local constants (including constexpr), functions (member, non-member) and enum values. + +* **GN3:** Use capitalized SNAKE_CASE only for +macros. + +### GF: General Formatting +* **GF1:** Each line of text in the code shall be at most 100 characters long. + +* **GF2:** Use only spaces, and indent 4 spaces at a time, never use tabs. + +* **GF3:** The open curly brace is always on the end of the last line of the statement (type, function, namespace declaration or control flow statement), not the start of the next line. +```c +int foo() { // <-- curly brace here + do_something(); +} + +if (condition) { // <-- curly brace here + do_something(); +} +else { // <-- curly brace here + do_something(); +} + +if (condition) { // <-- curly brace here + do_something(); +} else { // <-- Also possible + do_something(); +} +``` + +* **GF4:** There is never a space between the parentheses and the parameters in function declaration/invocation or control flow statements. + +```c +// Wrong +int foo( int arg_1, float arg_2 ); +if ( condition ); +call_foo( value_1, value_2 ); +for ( int i = 0; i < loop_count; i++ ); + +// Right +int foo(int arg_1, float arg_2); +if (condition); +call_foo(value_1, value_2); +for (int i = 0; i < loop_count; i++); +``` + +### FA: Files +* **FA1:** Filenames should be lowercase and can include underscores "_". + +* **FA2:** C++ header files exposed to the user should end in .hpp. + +* **FA3:** C++ source files should end in .cpp. + +* **FA4:** All header files shall start with `#pragma once` guards to prevent multiple inclusion, refer to [Structure of Header Files](CONTRIBUTING.md#structure-of-header-files) for more details. + +* **FA5:** Each header file shall contain items in the following order: + 1. Copyright + 2. Single blank line + 3. Preprocessor guard + 4. Single blank line + 5. Include statements (if there) + 6. Single blank line if include statements are present + 7. Global macros* (if any) + 8. Single blank line if macros statements are present + 9. Type/function declarations wrapped into namespaces + +Note: It is not necessary to put all macro definitions here. Sometimes it is convenient to have macros closer to the place where they are used. For example, sometimes it makes more sense to define macros inside the functions that use them (see Macros for more details). However, if the macro is used throughout the library, put it in header file between includes and the namespace declaration. + +* **FA6:** Each header file shall include other header +files in the following order: + + 1. C standard headers + 2. C++ standard headers + 3. Single blank line if C/C++ headers are present + 4. Third party libraries' header files (e.g., SYCL, TBB, OMP, etc.) + 5. Single blank line if third party headers are present + 6. Project's header files + +### NS: Namespaces +* **NS1:** Use snake_case: all lowercase, with underscores "_" between words for all namespaces. + +* **NS2:** The name of a top-level namespace must be the name of the project (oneMKL). + +* **NS3:** Do not indent content inside a namespace scope. + +```c +// Wrong! Do not indent +namespace onemkl { + + class table { }; + +} // namespace onemkl + +// Right +namespace onemkl { + +class table { }; + +} // namespace onemkl +``` + +* **NS4:** Put each namespace on its own line when declaring nested namespaces. + +```c +#include "onemkl/blas/path_to_some_header.hpp" + +namespace onemkl { +namespace blas { + +/* ... */ + +} // namespace blas +} // namespace onemkl +``` + + +### FU: Functions + +* **FU1:** Use snake_case: all lowercase, with +```c +// Underscores between words for all function names. +return_type class_name::function_name(type_1 arg_name_1, type_2 arg_name_2) { + do_something(); +} +``` + +* **FU2:** There is never a space between the function name (or operator) and the open brace. This rule applies to both function declaration/definitions and calls. + +Declaration +```c +// Wrong +void foo (type arg_name); +void operator() (type arg_name); +void operator bool (); + +// Right +void foo(type arg_name); +void operator()(type arg_name); +void operator bool(); +``` + +Call +```c +// Wrong +const auto x = foo (arg_1, arg_2); + +// Right +const auto x = foo(arg_1, arg_2); +``` + +* **FU3:** Do not put the function signature and the body on the same line. The only exception is an empty body, in that case place the curly braces at the same line (see rule FU4). +```c +// Wrong +std::int32_t get_something() const { return something_; } + +// Right +std::int32_t get_something() const { + return something_; +} +``` + +* **FU4:** Empty function body shall be at the same line as function signature. +```c +// Wrong +void empty_foo(type arg) { +} + +// Right +void empty_foo(type arg) {} +``` + + +### CS: Classes and Structures + +* **CS1:** Use snake_case: lower case and all words are separated with underscore character (_). +```c +class numeric_table; +class image; +struct params; +``` + +* **CS2:** The acceptable formats for initializer lists are when everything fits on one line: +```c +my_class::my_class(int var) : some_var_(var) { + do_something(); +} +``` +If the signature and initializer list are not all on one line, you must line wrap before the colon, indent 8 spaces, put each member on its own line, and align them: +```c +my_class::my_class(int var) + : some_var_(var), // <-- 8 space indent + some_other_var_(var + 1) { // lined up + do_something(); +} +``` +As with any other code block, the close curly brace can be on the same line as the open curly, if it fits: +```c +my_class::my_class(int var) + : some_var_(var), + another_var_(0) {} +``` + + +### VC: Variables and Constants + +* **VC1:** Use snake_case for all variables, function's arguments and constants. + +* **VC2:** Use variables and constant names followed by one underscore "_" for private and protected class-level variables. + +* **VC3:** The assignment operator "=" shall be surrounded by single whitespace. +```c +const auto val = get_some_value(); +``` + + +### ST: Statements + +* **ST1:** Each of the keywords +if/else/do/while/for/switch shall be followed by one space. An open curly brace after the condition shall be prepended with one space. +```c +while (condition) { // <-- one space after `while` and one space before `{` + do_something(); +} // <-- `;` is not required +``` + +* **ST2:** Each of the keywords if/else/do/while/for/switch shall always have accompanying curly braces even if they contain a single-line statement. +```c +// Wrong +if (my_const == my_var) + do_something(); + +// Right +if (my_const == my_var) { + do_something(); +} +``` + +* **ST3:** The statements within parentheses for operators if, for, while shall have no spaces adjacent to the open and close parentheses characters: +```c +// Wrong +for ( int i = 0; i < loop_size; i++ ) ...; + +// Right +for (int i = 0; i < loop_size; i++) ...; +``` + + +## Unit Tests + +oneMKL uses GoogleTest for functional testing. + +Be sure to extend the existing tests when fixing an issue, adding a new interface or new implementation under exisitng interfaces. diff --git a/LICENSE b/LICENSE new file mode 100644 index 000000000..580630b67 --- /dev/null +++ b/LICENSE @@ -0,0 +1,210 @@ + Apache License + Version 2.0, January 2004 + http://www.apache.org/licenses/ + + TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION + + 1. Definitions. + + "License" shall mean the terms and conditions for use, reproduction, + and distribution as defined by Sections 1 through 9 of this document. + + "Licensor" shall mean the copyright owner or entity authorized by + the copyright owner that is granting the License. + + "Legal Entity" shall mean the union of the acting entity and all + other entities that control, are controlled by, or are under common + control with that entity. For the purposes of this definition, + "control" means (i) the power, direct or indirect, to cause the + direction or management of such entity, whether by contract or + otherwise, or (ii) ownership of fifty percent (50%) or more of the + outstanding shares, or (iii) beneficial ownership of such entity. + + "You" (or "Your") shall mean an individual or Legal Entity + exercising permissions granted by this License. + + "Source" form shall mean the preferred form for making modifications, + including but not limited to software source code, documentation + source, and configuration files. + + "Object" form shall mean any form resulting from mechanical + transformation or translation of a Source form, including but + not limited to compiled object code, generated documentation, + and conversions to other media types. + + "Work" shall mean the work of authorship, whether in Source or + Object form, made available under the License, as indicated by a + copyright notice that is included in or attached to the work + (an example is provided in the Appendix below). + + "Derivative Works" shall mean any work, whether in Source or Object + form, that is based on (or derived from) the Work and for which the + editorial revisions, annotations, elaborations, or other modifications + represent, as a whole, an original work of authorship. For the purposes + of this License, Derivative Works shall not include works that remain + separable from, or merely link (or bind by name) to the interfaces of, + the Work and Derivative Works thereof. + + "Contribution" shall mean any work of authorship, including + the original version of the Work and any modifications or additions + to that Work or Derivative Works thereof, that is intentionally + submitted to Licensor for inclusion in the Work by the copyright owner + or by an individual or Legal Entity authorized to submit on behalf of + the copyright owner. For the purposes of this definition, "submitted" + means any form of electronic, verbal, or written communication sent + to the Licensor or its representatives, including but not limited to + communication on electronic mailing lists, source code control systems, + and issue tracking systems that are managed by, or on behalf of, the + Licensor for the purpose of discussing and improving the Work, but + excluding communication that is conspicuously marked or otherwise + designated in writing by the copyright owner as "Not a Contribution." + + "Contributor" shall mean Licensor and any individual or Legal Entity + on behalf of whom a Contribution has been received by Licensor and + subsequently incorporated within the Work. + + 2. Grant of Copyright License. Subject to the terms and conditions of + this License, each Contributor hereby grants to You a perpetual, + worldwide, non-exclusive, no-charge, royalty-free, irrevocable + copyright license to reproduce, prepare Derivative Works of, + publicly display, publicly perform, sublicense, and distribute the + Work and such Derivative Works in Source or Object form. + + 3. Grant of Patent License. Subject to the terms and conditions of + this License, each Contributor hereby grants to You a perpetual, + worldwide, non-exclusive, no-charge, royalty-free, irrevocable + (except as stated in this section) patent license to make, have made, + use, offer to sell, sell, import, and otherwise transfer the Work, + where such license applies only to those patent claims licensable + by such Contributor that are necessarily infringed by their + Contribution(s) alone or by combination of their Contribution(s) + with the Work to which such Contribution(s) was submitted. If You + institute patent litigation against any entity (including a + cross-claim or counterclaim in a lawsuit) alleging that the Work + or a Contribution incorporated within the Work constitutes direct + or contributory patent infringement, then any patent licenses + granted to You under this License for that Work shall terminate + as of the date such litigation is filed. + + 4. Redistribution. You may reproduce and distribute copies of the + Work or Derivative Works thereof in any medium, with or without + modifications, and in Source or Object form, provided that You + meet the following conditions: + + (a) You must give any other recipients of the Work or + Derivative Works a copy of this License; and + + (b) You must cause any modified files to carry prominent notices + stating that You changed the files; and + + (c) You must retain, in the Source form of any Derivative Works + that You distribute, all copyright, patent, trademark, and + attribution notices from the Source form of the Work, + excluding those notices that do not pertain to any part of + the Derivative Works; and + + (d) If the Work includes a "NOTICE" text file as part of its + distribution, then any Derivative Works that You distribute must + include a readable copy of the attribution notices contained + within such NOTICE file, excluding those notices that do not + pertain to any part of the Derivative Works, in at least one + of the following places: within a NOTICE text file distributed + as part of the Derivative Works; within the Source form or + documentation, if provided along with the Derivative Works; or, + within a display generated by the Derivative Works, if and + wherever such third-party notices normally appear. The contents + of the NOTICE file are for informational purposes only and + do not modify the License. You may add Your own attribution + notices within Derivative Works that You distribute, alongside + or as an addendum to the NOTICE text from the Work, provided + that such additional attribution notices cannot be construed + as modifying the License. + + You may add Your own copyright statement to Your modifications and + may provide additional or different license terms and conditions + for use, reproduction, or distribution of Your modifications, or + for any such Derivative Works as a whole, provided Your use, + reproduction, and distribution of the Work otherwise complies with + the conditions stated in this License. + + 5. Submission of Contributions. Unless You explicitly state otherwise, + any Contribution intentionally submitted for inclusion in the Work + by You to the Licensor shall be under the terms and conditions of + this License, without any additional terms or conditions. + Notwithstanding the above, nothing herein shall supersede or modify + the terms of any separate license agreement you may have executed + with Licensor regarding such Contributions. + + 6. Trademarks. This License does not grant permission to use the trade + names, trademarks, service marks, or product names of the Licensor, + except as required for reasonable and customary use in describing the + origin of the Work and reproducing the content of the NOTICE file. + + 7. Disclaimer of Warranty. Unless required by applicable law or + agreed to in writing, Licensor provides the Work (and each + Contributor provides its Contributions) on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or + implied, including, without limitation, any warranties or conditions + of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A + PARTICULAR PURPOSE. You are solely responsible for determining the + appropriateness of using or redistributing the Work and assume any + risks associated with Your exercise of permissions under this License. + + 8. Limitation of Liability. In no event and under no legal theory, + whether in tort (including negligence), contract, or otherwise, + unless required by applicable law (such as deliberate and grossly + negligent acts) or agreed to in writing, shall any Contributor be + liable to You for damages, including any direct, indirect, special, + incidental, or consequential damages of any character arising as a + result of this License or out of the use or inability to use the + Work (including but not limited to damages for loss of goodwill, + work stoppage, computer failure or malfunction, or any and all + other commercial damages or losses), even if such Contributor + has been advised of the possibility of such damages. + + 9. Accepting Warranty or Additional Liability. While redistributing + the Work or Derivative Works thereof, You may choose to offer, + and charge a fee for, acceptance of support, warranty, indemnity, + or other liability obligations and/or rights consistent with this + License. However, in accepting such obligations, You may act only + on Your own behalf and on Your sole responsibility, not on behalf + of any other Contributor, and only if You agree to indemnify, + defend, and hold each Contributor harmless for any liability + incurred by, or claims asserted against, such Contributor by reason + of your accepting any such warranty or additional liability. + + END OF TERMS AND CONDITIONS + + ============================================================================ + + APPENDIX: How to apply the Apache License to your work. + + To apply the Apache License to your work, attach the following + boilerplate notice, with the fields enclosed by brackets "[]" + replaced with your own identifying information. (Don't include + the brackets!) The text should be enclosed in the appropriate + comment syntax for the file format. We also recommend that a + file or class name and description of purpose be included on the + same "printed page" as the copyright notice for easier + identification within third-party archives. + + Copyright [yyyy] [name of copyright owner] + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. + +This distribution includes third party software ("third party programs"). This +third party software, even if included with the distribution of the oneMKL open +source project, may be governed by separate license terms, including without +limitation, third party license terms, and open source software license terms. +These separate license terms govern your use of the third party programs as set +forth in the "THIRD-PARTY-PROGRAMS" file. diff --git a/README.md b/README.md new file mode 100644 index 000000000..4fc357f16 --- /dev/null +++ b/README.md @@ -0,0 +1,273 @@ +# oneAPI Math Kernel Library (oneMKL) Interfaces + +## Contents + +- [Introduction](#introduction) +- [Support and Requirements](#support-and-requirements) +- [Build Setup](#build-setup) +- [Building with CMake](#building-with-cmake) +- [Project Cleanup](#project-cleanup) +- [Notice and Disclaimer](#notice-and-disclaimer) + +--- + +## Introduction + +oneMKL interfaces is an open-source implementation of oneMKL Data Parallel C++ (DPC++) interfaces according to [oneMKL specification](https://spec.oneapi.com/versions/latest/elements/oneMKL/source/index.html) that can work with multiple devices (backends) using device specific libraries underneath. + + + + + + + + + + + + + + + + + + + + + + +
User ApplicationoneMKL LayerThird-Party LibraryHardware Backend
oneMKL interfaceoneMKL selector[Intel(R) oneAPI Math Kernel Library](https://software.intel.com/en-us/oneapi/onemkl) for Intel CPUIntel CPU
[Intel(R) oneAPI Math Kernel Library](https://software.intel.com/en-us/oneapi/onemkl) for Intel GPUIntel GPU
+ +### Supported Usage Models: + +There are two oneMKL selector layer implementations: + +- **Run-time dispatching**: The application is linked with the onemkl library and the required backend is loaded at run-time based on device vendor (all libraries should be dynamic). + +Example of app.cpp with run-time dispatching: + +```cpp +include "onemkl/onemkl.hpp" + +... +cpu_dev = cl::sycl::device(cl::sycl::cpu_selector()); +gpu_dev = cl::sycl::device(cl::sycl::gpu_selector()); + +cl::sycl::queue cpu_queue(cpu_dev); +cl::sycl::queue gpu_queue(gpu_dev); + +onemkl::blas::gemm(cpu_queue, transA, transB, m, ...); +onemkl::blas::gemm(gpu_queue, transA, transB, m, ...); +``` +How to build an application with run-time dispatching: + +```cmd +$> clang++ -fsycl –I$ONEMKL/include app.cpp +$> clang++ -fsycl app.o –L$ONEMKL/lib –lonemkl +``` + +- **Compile-time dispatching**: The application uses a templated API where the template parameters specify the required backends and third-party libraries and the application is linked with required onemkl backend wrapper libraries (libraries can be static or dynamic). + +Example of app.cpp with compile-time dispatching: + +```cpp +include "onemkl/onemkl.hpp" + +... +cpu_dev = cl::sycl::device(cl::sycl::cpu_selector()); +gpu_dev = cl::sycl::device(cl::sycl::gpu_selector()); + +cl::sycl::queue cpu_queue(cpu_dev); +cl::sycl::queue gpu_queue(gpu_dev); + +onemkl::blas::gemm(cpu_queue, transA, transB, m, ...); +onemkl::blas::gemm(gpu_queue, transA, transB, m, ...); +``` +How to build an application with run-time dispatching: + +```cmd +$> clang++ -fsycl –I$ONEMKL/include app.cpp +$> clang++ -fsycl app.o –L$ONEMKL/lib –lonemkl_blas_mklcpu –lonemkl_blas_mklgpu +``` + +### Supported Configurations: + +Supported domains: BLAS + +#### Linux* + + Backend | Library | Supported Link Type + :------| :-------| :------------------ + Intel CPU | Intel(R) oneAPI Math Kernel Library | Dynamic, Static + Intel GPU | Intel(R) oneAPI Math Kernel Library | Dynamic, Static + +--- + +## Support and Requirements + +### Hardware Platform Support + +#### Linux* +- CPU + - Intel Atom(R) Processors + - Intel(R) Core(TM) Processor Family + - Intel(R) Xeon(R) Processor Family +- Accelerators + - Intel(R) Processor Graphics GEN9 + +--- +### Supported Operating Systems + +#### Linux* + +Operating System | CPU Host/Target | Integrated Graphics from Intel (Intel GPU) +:--- | :--- | :--- +Ubuntu | 18.04.3, 19.04 | 18.04.3, 19.10 +SUSE Linux Enterprise Server* | 15 | *Not supported* +Red Hat Enterprise Linux* (RHEL*) | 8 | *Not supported* +Linux* kernel | *N/A* | 4.11 or higher + +--- + +### Software Requirements + +**What should I download?** + +#### General: + + + + + + + + + + + + + + + + + + + + + + + + + + +
Functional Testing Build Only Documentation
CMake CMake CMake
Ninja (optional) Ninja (optional) Sphinx
GNU* FORTRAN Compiler
NETLIB LAPACK
+ + +#### Hardware and OS Specific: + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Operating SystemHardwareUsing CMake
Linux* Any GNU* GCC 5.1 or higher
Intel CPU Intel(R) oneAPI DPC++ Compiler
or
Intel project for LLVM* technology
Intel(R) oneAPI Math Kernel Library
Intel GPU Intel(R) oneAPI DPC++ Compiler
Intel GPU driver
Intel(R) oneAPI Math Kernel Library
+ +#### Product and Version Information: + +Product | Supported Version | License +:--- | :--- | :--- +Python | 3.6 or higher | [PSF](https://docs.python.org/3.6/license.html) +[CMake](https://cmake.org/download/) | 3.13 or higher | [The OSI-approved BSD 3-clause License](https://gitlab.kitware.com/cmake/cmake/raw/master/Copyright.txt) +[Ninja](https://ninja-build.org/) | 1.9.0 | [Apache License v2.0](https://github.com/ninja-build/ninja/blob/master/COPYING) +[GNU* FORTRAN Compiler](https://gcc.gnu.org/wiki/GFortran) | 7.4.0 or higher | [GNU General Public License v2](https://www.gnu.org/licenses/old-licenses/gpl-2.0.html) +[Intel(R) oneAPI DPC++ Compiler](https://software.intel.com/en-us/oneapi/dpc-compiler) | 2021.1-beta05 | [End User License Agreement for the Intel(R) Software Development Products](https://software.intel.com/en-us/license/eula-for-intel-software-development-products) +[Intel project for LLVM* technology binary for Intel CPU](https://github.com/intel/llvm/releases) | Daily builds (experimental) tested with [20200331](https://github.com/intel/llvm/releases/download/20200331/dpcpp-compiler.tar.gz) | [Apache License v2](https://github.com/intel/llvm/blob/sycl/sycl/LICENSE.TXT) +[Intel(R) oneAPI Math Kernel Library](https://software.intel.com/en-us/oneapi/onemkl) | 2021.1-beta05 | [Intel Simplified Software License](https://software.intel.com/en-us/license/intel-simplified-software-license) +[NETLIB LAPACK](https://github.com/Reference-LAPACK/lapack) | 3.7.1 | [BSD like license](http://www.netlib.org/lapack/LICENSE.txt) +[Sphinx](https://www.sphinx-doc.org/en/master/) | 2.4.4 | [BSD License](https://github.com/sphinx-doc/sphinx/blob/3.x/LICENSE) + +--- + +## Build Setup + +1. Install Intel(R) oneAPI DPC++ Compiler (select variant as per requirement). + +2. Clone this project to ``, where `` is the root directory of this repository. + +3. [Build with CMake](#building-with-cmake). + +--- + +## Building with CMake + +1. Make sure you have completed [Build Setup](#build-setup). + +2. Build and install all required [dependencies](#software-requirements). + +Then: + +- On Linux* +```bash +# Inside +mkdir build && cd build +export CXX=/bin/dpcpp; +cmake .. [-DMKL_ROOT=] \ # required only if enviroment variable MKLROOT is not set + [-DREF_BLAS_ROOT=] # required only for testing +cmake --build . +ctest +cmake --install . --prefix +``` + +### Build Options +You can specify build options using `-D=`. The following table provides the list of options supported by CMake. + +CMake Option | Supported Values | Default Value +:----------- | :--------------- | :--- +BUILD_SHARED_LIBS | True, False | True +ENABLE_MKLCPU_BACKEND | True, False | True +ENABLE_MKLGPU_BACKEND | True, False | True +ENABLE_MKLCPU_THREAD_TBB | True, False | True +BUILD_FUNCTIONAL_TESTS | True, False | True +BUILD_DOC | True, False | False + +--- + +## Project Cleanup + +Most use-cases involve building the project without the need to cleanup the build directory. However, if you wish to cleanup the build directory, you can delete the `build` folder and create a new one. If you wish to cleanup the build files but retain the build configuration, following commands will help you do so. + +```sh +# If you use "GNU/Unix Makefiles" for building, +make clean + +# If you use "Ninja" for building +ninja -t clean +``` + +--- + +[Legal information](legal_information.md) diff --git a/THIRD-PARTY-PROGRAMS b/THIRD-PARTY-PROGRAMS new file mode 100644 index 000000000..6a560a4ae --- /dev/null +++ b/THIRD-PARTY-PROGRAMS @@ -0,0 +1,51 @@ +oneMKL Open Source Third Party Programs File + +This file contains the list of third party software ("third party programs") +contained in the oneMKL open source project and their required notices and/or +license terms. This third party software, even if included with the distribution +of the oneMKL open source project, may be governed by separate license terms, +including without limitation, third party license terms, and open source +software license terms. These separate license terms govern your use of the +third party programs as set forth in the "THIRD-PARTY-PROGRAMS" file. + +Third party programs and their corresponding required notices and/or license +terms are listed below. + +1. googletest (https://github.com/google/googletest) + +%% googletest NOTICES, INFORMATION, AND LICENSE BEGINE HERE +------------------------------------------------------------------------------- +BSD-Like License + +Copyright 2008, Google Inc. +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are +met: + + * Redistributions of source code must retain the above copyright +notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above +copyright notice, this list of conditions and the following disclaimer +in the documentation and/or other materials provided with the +distribution. + * Neither the name of Google Inc. nor the names of its +contributors may be used to endorse or promote products derived from +this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +------------------------------------------------------------------------------- +END OF googletest NOTICES, INFORMATION, AND LICENSE + + diff --git a/_clang-format b/_clang-format new file mode 100644 index 000000000..85eed3284 --- /dev/null +++ b/_clang-format @@ -0,0 +1,164 @@ +#=============================================================================== +# Copyright 2016-2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +--- +Language: Cpp +AccessModifierOffset: -4 +AlignAfterOpenBracket: Align +AlignConsecutiveAssignments: true +AlignConsecutiveDeclarations: false +AlignConsecutiveMacros: true +AlignEscapedNewlines: Left +AlignOperands: true +AlignTrailingComments: false +AllowAllArgumentsOnNextLine: false +AllowAllConstructorInitializersOnNextLine: false +AllowAllParametersOfDeclarationOnNextLine: false +AllowShortBlocksOnASingleLine: false +AllowShortCaseLabelsOnASingleLine: false +AllowShortFunctionsOnASingleLine: Empty +AllowShortIfStatementsOnASingleLine: Never +AllowShortLambdasOnASingleLine: None +AllowShortLoopsOnASingleLine: false +AlwaysBreakAfterDefinitionReturnType: None +AlwaysBreakAfterReturnType: None +AlwaysBreakBeforeMultilineStrings: false +AlwaysBreakTemplateDeclarations: Yes +BinPackArguments: true +BinPackParameters: true +BraceWrapping: + AfterCaseLabel: false + AfterClass: false + AfterControlStatement: false + AfterEnum: false + AfterFunction: false + AfterNamespace: false + AfterObjCDeclaration: false + AfterStruct: false + AfterUnion: false + AfterExternBlock: false + BeforeCatch: true + BeforeElse: true + IndentBraces: false + SplitEmptyFunction: false + SplitEmptyRecord: false + SplitEmptyNamespace: false +BreakBeforeBinaryOperators: None +BreakBeforeBraces: Custom +BreakBeforeTernaryOperators: true +BreakConstructorInitializers: BeforeColon +BreakInheritanceList: BeforeColon +BreakStringLiterals: false +ColumnLimit: 100 +CommentPragmas: '^ IWYU pragma:' +CompactNamespaces: false +ConstructorInitializerAllOnOneLineOrOnePerLine: true +ConstructorInitializerIndentWidth: 8 +ContinuationIndentWidth: 4 +Cpp11BracedListStyle: false +DerivePointerAlignment: true +DisableFormat: false +FixNamespaceComments: true +ForEachMacros: + - foreach + - Q_FOREACH + - BOOST_FOREACH +IncludeBlocks: Preserve +IncludeCategories: + - Regex: '^' + Priority: 2 + - Regex: '^<.*\.h>' + Priority: 1 + - Regex: '^<.*' + Priority: 2 + - Regex: '.*' + Priority: 3 +IncludeIsMainRegex: '([-_](test|unittest))?$' +IndentCaseLabels: true +IndentPPDirectives: BeforeHash +IndentWidth: 4 +IndentWrappedFunctionNames: false +KeepEmptyLinesAtTheStartOfBlocks: false +MacroBlockBegin: '' +MacroBlockEnd: '' +MaxEmptyLinesToKeep: 1 +NamespaceIndentation: None +PenaltyBreakAssignment: 2 +PenaltyBreakBeforeFirstCallParameter: 1 +PenaltyBreakComment: 300 +PenaltyBreakFirstLessLess: 120 +PenaltyBreakString: 1000 +PenaltyBreakTemplateDeclaration: 10 +PenaltyExcessCharacter: 1000000 +PenaltyReturnTypeOnItsOwnLine: 200 +PointerAlignment: Left +RawStringFormats: + - Language: Cpp + Delimiters: + - cc + - CC + - cpp + - Cpp + - CPP + - 'c++' + - 'C++' + CanonicalDelimiter: '' + BasedOnStyle: google + - Language: TextProto + Delimiters: + - pb + - PB + - proto + - PROTO + EnclosingFunctions: + - EqualsProto + - EquivToProto + - PARSE_PARTIAL_TEXT_PROTO + - PARSE_TEST_PROTO + - PARSE_TEXT_PROTO + - ParseTextOrDie + - ParseTextProtoOrDie + CanonicalDelimiter: '' + BasedOnStyle: google +ReflowComments: false +SortIncludes: true +SortUsingDeclarations: false +SpaceAfterCStyleCast: false +SpaceAfterLogicalNot: false +SpaceAfterTemplateKeyword: true +SpaceBeforeAssignmentOperators: true +SpaceBeforeCpp11BracedList: false +SpaceBeforeCtorInitializerColon: true +SpaceBeforeInheritanceColon: true +SpaceBeforeParens: ControlStatements +SpaceBeforeRangeBasedForLoopColon: true +SpaceInEmptyParentheses: false +SpacesBeforeTrailingComments: 1 +SpacesInAngles: false +SpacesInContainerLiterals: false +SpacesInCStyleCastParentheses: false +SpacesInParentheses: false +SpacesInSquareBrackets: false +Standard: Cpp11 +StatementMacros: + - Q_UNUSED + - QT_REQUIRE_VERSION +TabWidth: 1 +UseTab: Never +... diff --git a/cmake/CMakeLists.txt b/cmake/CMakeLists.txt new file mode 100644 index 000000000..e221500c9 --- /dev/null +++ b/cmake/CMakeLists.txt @@ -0,0 +1,23 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +install(FILES FindMKL.cmake + FindCompiler.cmake + DESTINATION "lib/cmake/${PROJECT_NAME}" +) diff --git a/cmake/FindCBLAS.cmake b/cmake/FindCBLAS.cmake new file mode 100644 index 000000000..466150998 --- /dev/null +++ b/cmake/FindCBLAS.cmake @@ -0,0 +1,50 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +include_guard() + +list(APPEND BLAS_LIBS cblas) +list(APPEND BLAS_LIBS blas) + +foreach(lib ${BLAS_LIBS}) + find_library(${lib}_file NAMES ${lib} HINTS ${REF_BLAS_ROOT} PATH_SUFFIXES lib lib64) + find_package_handle_standard_args(CBLAS REQUIRED_VARS ${lib}_file) +endforeach() + +if(UNIX) + set(CBLAS_LIBNAME libcblas.so) +else() + set(CBLAS_LIBNAME cblas.lib) +endif() + +find_path(CBLAS_LIB_DIR ${CBLAS_LIBNAME} HINTS ${REF_BLAS_ROOT} PATH_SUFFIXES lib lib64) + +find_path(CBLAS_INCLUDE cblas.h HINTS ${REF_BLAS_ROOT} PATH_SUFFIXES include) + + +if(UNIX) + list(APPEND CBLAS_LINK "-Wl,-rpath,${CBLAS_LIB_DIR}") + list(APPEND CBLAS_LINK "-L${CBLAS_LIB_DIR}") + foreach(lib ${BLAS_LIBS}) + list(APPEND CBLAS_LINK -l${lib}) + endforeach() +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(CBLAS REQUIRED_VARS CBLAS_INCLUDE CBLAS_LINK) diff --git a/cmake/FindCompiler.cmake b/cmake/FindCompiler.cmake new file mode 100644 index 000000000..e9125af8f --- /dev/null +++ b/cmake/FindCompiler.cmake @@ -0,0 +1,40 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +include_guard() + +include(CheckCXXCompilerFlag) +include(FindPackageHandleStandardArgs) + +check_cxx_compiler_flag("-fsycl" is_dpcpp) + +if(is_dpcpp) + find_package_handle_standard_args(SYCL DEFAULT_MSG) +endif() + +if(SYCL_FOUND AND is_dpcpp) + # Workaround for internal compiler error during linking if -fsycl is used + get_filename_component(SYCL_BINARY_DIR ${CMAKE_CXX_COMPILER} DIRECTORY) + find_library(SYCL_LIBRARY NAMES sycl PATHS "${SYCL_BINARY_DIR}/../lib") + + add_library(ONEMKL::SYCL::SYCL INTERFACE IMPORTED) + set_target_properties(ONEMKL::SYCL::SYCL PROPERTIES + INTERFACE_COMPILE_OPTIONS "-fsycl" + INTERFACE_LINK_LIBRARIES ${SYCL_LIBRARY}) +endif() diff --git a/cmake/FindMKL.cmake b/cmake/FindMKL.cmake new file mode 100644 index 000000000..c44a46b2f --- /dev/null +++ b/cmake/FindMKL.cmake @@ -0,0 +1,97 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +include_guard() +set(MKL_SYCL mkl_sycl) +set(MKL_IFACE mkl_intel_ilp64) +set(MKL_SEQ mkl_sequential) +set(MKL_TBB mkl_tbb_thread) +set(MKL_CORE mkl_core) + +set(MKL_C ${MKL_IFACE}) + +if(ENABLE_MKLCPU_THREAD_TBB) + find_package(TBB REQUIRED) + list(APPEND MKL_C ${MKL_TBB}) +else() + list(APPEND MKL_C ${MKL_SEQ}) +endif() + +list(APPEND MKL_C ${MKL_CORE}) + + +if (ENABLE_MKLCPU_BACKEND OR ENABLE_MKLGPU_BACKEND) + if(ENABLE_MKLGPU_BACKEND) + list(APPEND MKL_LIBRARIES ${MKL_SYCL}) + endif() + list(APPEND MKL_LIBRARIES ${MKL_C}) +endif() + +foreach(lib ${MKL_LIBRARIES}) + find_library(${lib}_file NAMES ${lib} + HINTS $ENV{MKLROOT} ${MKL_ROOT} + PATH_SUFFIXES lib/intel64) + find_package_handle_standard_args(MKL REQUIRED_VARS ${lib}_file) +endforeach() + +if(UNIX) + set(MKL_CORE_LIBNAME libmkl_core.so) +else() + set(MKL_CORE_LIBNAME mkl_core.lib) +endif() + +find_path(MKL_LIB_DIR ${MKL_CORE_LIBNAME} + HINTS $ENV{MKLROOT} ${MKL_ROOT} + PATH_SUFFIXES lib/intel64) + +find_path(MKL_INCLUDE mkl.h + HINTS $ENV{MKLROOT} ${MKL_ROOT} + PATH_SUFFIXES include) + +if(${CMAKE_SIZEOF_VOID_P} EQUAL 8) + set(MKL_COPT "-DMKL_ILP64") +else() + set(MKL_COPT "") +endif() + +#Workaround for soname problem +if(UNIX) + list(APPEND MKL_LINK_PREFIX "-Wl,-rpath,${MKL_LIB_DIR}") + list(APPEND MKL_LINK_PREFIX "-L${MKL_LIB_DIR}") + if (ENABLE_MKLCPU_BACKEND OR ENABLE_MKLGPU_BACKEND) + set(MKL_LINK_C ${MKL_LINK_PREFIX}) + foreach(lib ${MKL_C}) + list(APPEND MKL_LINK_C -l${lib}) + endforeach() + if(ENABLE_MKLCPU_THREAD_TBB) + list(APPEND MKL_LINK_C ${TBB_LINK}) + endif() + if(ENABLE_MKLGPU_BACKEND) + set(MKL_LINK_SYCL ${MKL_LINK_PREFIX} -l${MKL_SYCL} ${MKL_LINK_C} -lOpenCL) + endif() + endif() +endif() + + +include(FindPackageHandleStandardArgs) +if (ENABLE_MKLCPU_BACKEND) + find_package_handle_standard_args(MKL REQUIRED_VARS MKL_INCLUDE MKL_COPT MKL_LINK_C) +else(ENABLE_MKLGPU_BACKEND) + find_package_handle_standard_args(MKL REQUIRED_VARS MKL_INCLUDE MKL_COPT MKL_LINK_SYCL) +endif() diff --git a/cmake/FindSphinx.cmake b/cmake/FindSphinx.cmake new file mode 100644 index 000000000..d0ae742d5 --- /dev/null +++ b/cmake/FindSphinx.cmake @@ -0,0 +1,31 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +find_program(SPHINX_EXECUTABLE + NAMES sphinx-build + HINTS $ENV{SPHINXROOT} ${SPHINX_ROOT} + PATH_SUFFIXES bin + DOC "Sphinx Documentation generator" +) + +include(FindPackageHandleStandardArgs) + +find_package_handle_standard_args(Sphinx DEFAULT_MSG SPHINX_EXECUTABLE) + +mark_as_advanced(SPHINX_EXECUTABLE) diff --git a/cmake/FindTBB.cmake b/cmake/FindTBB.cmake new file mode 100644 index 000000000..47ea5ad2e --- /dev/null +++ b/cmake/FindTBB.cmake @@ -0,0 +1,50 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +include_guard() + +#Workaround for soname problem +if(UNIX) + set(TBB_LIBNAME libtbb.so) +else() + set(TBB_LIBNAME tbb.lib) +endif() + +find_path(TBB_LIB_DIR ${TBB_LIBNAME} + HINTS $ENV{TBBROOT} $ENV{MKLROOT} ${MKL_ROOT} ${TBB_ROOT} + PATH_SUFFIXES "lib" "lib/intel64/gcc4.4" "lib/intel64/gcc4.8" + "../tbb/lib/intel64/gcc4.4" "../tbb/lib/intel64/gcc4.8" + "../../tbb/latest/lib/intel64/gcc4.8" +) + +find_library(TBB_LIBRARIES NAMES tbb + HINTS $ENV{TBBROOT} $ENV{MKLROOT} ${MKL_ROOT} ${TBB_ROOT} + PATH_SUFFIXES "lib" "lib/intel64/gcc4.4" "lib/intel64/gcc4.8" + "../tbb/lib/intel64/gcc4.4" "../tbb/lib/intel64/gcc4.8" + "../../tbb/latest/lib/intel64/gcc4.8" +) + +#Workaround for ref problem +if(UNIX) + set(TBB_LINK "-Wl,-rpath,${TBB_LIB_DIR} -L${TBB_LIB_DIR} -ltbb") +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(TBB REQUIRED_VARS TBB_LIBRARIES TBB_LINK) + diff --git a/cmake/oneMKLConfig.cmake b/cmake/oneMKLConfig.cmake new file mode 100644 index 000000000..8e03a2ada --- /dev/null +++ b/cmake/oneMKLConfig.cmake @@ -0,0 +1,26 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +list(APPEND CMAKE_MODULE_PATH ${CMAKE_CURRENT_LIST_DIR}) +include(CMakeFindDependencyMacro) + +#find_dependency(MKL REQUIRED) +find_dependency(SYCL REQUIRED) + +include("${CMAKE_CURRENT_LIST_DIR}/oneMKLTargets.cmake") diff --git a/cmake/toolchain/intel_clang-toolchain.cmake b/cmake/toolchain/intel_clang-toolchain.cmake new file mode 100644 index 000000000..37ad585b3 --- /dev/null +++ b/cmake/toolchain/intel_clang-toolchain.cmake @@ -0,0 +1,54 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# Clang toolchain for Intel(R) oneAPI DPC++ Compiler +# Todo: Generalize location of compiler variables +set(CMAKE_SYSTEM_NAME Linux) +set(CMAKE_SYSTEM_PROCESSOR x86_64) + +find_program(CMAKE_C_COMPILER clang) +find_program(CMAKE_CXX_COMPILER clang++) + +if(NOT CMAKE_C_COMPILER) + message(ERROR "clang not found") +endif() + +if(NOT CMAKE_CXX_COMPILER) + message(ERROR "clang++ not found") +endif() + +#EXECUTE_PROCESS(COMMAND clang++ --version CLANG_VERSION) + +set( + CMAKE_C_COMPILER + "${CMAKE_C_COMPILER}" + CACHE + STRING + "C compiler" + FORCE +) + +set( + CMAKE_CXX_COMPILER + "${CMAKE_CXX_COMPILER}" + CACHE + STRING + "C++ compiler" + FORCE +) diff --git a/deps/googletest/CMakeLists.txt b/deps/googletest/CMakeLists.txt new file mode 100644 index 000000000..f6a528753 --- /dev/null +++ b/deps/googletest/CMakeLists.txt @@ -0,0 +1,325 @@ +######################################################################## +# CMake build script for Google Test. +# +# To run the tests for Google Test itself on Linux, use 'make test' or +# ctest. You can select which tests to run using 'ctest -R regex'. +# For more options, run 'ctest --help'. + +# When other libraries are using a shared version of runtime libraries, +# Google Test also has to use one. +option( + gtest_force_shared_crt + "Use shared (DLL) run-time lib even when Google Test is built as static lib." + OFF) + +option(gtest_build_tests "Build all of gtest's own tests." OFF) + +option(gtest_build_samples "Build gtest's sample programs." OFF) + +option(gtest_disable_pthreads "Disable uses of pthreads in gtest." OFF) + +option( + gtest_hide_internal_symbols + "Build gtest with internal symbols hidden in shared libraries." + OFF) + +# Defines pre_project_set_up_hermetic_build() and set_up_hermetic_build(). +include(cmake/hermetic_build.cmake OPTIONAL) + +if (COMMAND pre_project_set_up_hermetic_build) + pre_project_set_up_hermetic_build() +endif() + +######################################################################## +# +# Project-wide settings + +# Name of the project. +# +# CMake files in this project can refer to the root source directory +# as ${gtest_SOURCE_DIR} and to the root binary directory as +# ${gtest_BINARY_DIR}. +# Language "C" is required for find_package(Threads). + +# Project version: + +if (CMAKE_VERSION VERSION_LESS 3.0) + project(gtest CXX C) + set(PROJECT_VERSION ${GOOGLETEST_VERSION}) +else() + cmake_policy(SET CMP0048 NEW) + project(gtest VERSION 1.8.1 LANGUAGES CXX C) +endif() +cmake_minimum_required(VERSION 2.6.4) + +if (POLICY CMP0063) # Visibility + cmake_policy(SET CMP0063 NEW) +endif (POLICY CMP0063) + +if (COMMAND set_up_hermetic_build) + set_up_hermetic_build() +endif() + +# These commands only run if this is the main project +if(CMAKE_PROJECT_NAME STREQUAL "gtest" OR CMAKE_PROJECT_NAME STREQUAL "googletest-distribution") + + # BUILD_SHARED_LIBS is a standard CMake variable, but we declare it here to + # make it prominent in the GUI. + option(BUILD_SHARED_LIBS "Build shared libraries (DLLs)." OFF) + +else() + + mark_as_advanced( + gtest_force_shared_crt + gtest_build_tests + gtest_build_samples + gtest_disable_pthreads + gtest_hide_internal_symbols) + +endif() + + +if (gtest_hide_internal_symbols) + set(CMAKE_CXX_VISIBILITY_PRESET hidden) + set(CMAKE_VISIBILITY_INLINES_HIDDEN 1) +endif() + +# Define helper functions and macros used by Google Test. +include(cmake/internal_utils.cmake) + +config_compiler_and_linker() # Defined in internal_utils.cmake. + +# Create the CMake package file descriptors. +if (INSTALL_GTEST) + include(CMakePackageConfigHelpers) + set(cmake_package_name GTest) + set(targets_export_name ${cmake_package_name}Targets CACHE INTERNAL "") + set(generated_dir "${CMAKE_CURRENT_BINARY_DIR}/generated" CACHE INTERNAL "") + set(cmake_files_install_dir "${CMAKE_INSTALL_LIBDIR}/cmake/${cmake_package_name}") + set(version_file "${generated_dir}/${cmake_package_name}ConfigVersion.cmake") + write_basic_package_version_file(${version_file} COMPATIBILITY AnyNewerVersion) + install(EXPORT ${targets_export_name} + NAMESPACE ${cmake_package_name}:: + DESTINATION ${cmake_files_install_dir}) + set(config_file "${generated_dir}/${cmake_package_name}Config.cmake") + configure_package_config_file("${gtest_SOURCE_DIR}/cmake/Config.cmake.in" + "${config_file}" INSTALL_DESTINATION ${cmake_files_install_dir}) + install(FILES ${version_file} ${config_file} + DESTINATION ${cmake_files_install_dir}) +endif() + +# Where Google Test's .h files can be found. +set(gtest_build_include_dirs + "${gtest_SOURCE_DIR}/include" + "${gtest_SOURCE_DIR}") +include_directories(${gtest_build_include_dirs}) + +######################################################################## +# +# Defines the gtest & gtest_main libraries. User tests should link +# with one of them. + +# Google Test libraries. We build them using more strict warnings than what +# are used for other targets, to ensure that gtest can be compiled by a user +# aggressive about warnings. +cxx_library(gtest "${cxx_strict}" src/gtest-all.cc) +cxx_library(gtest_main "${cxx_strict}" src/gtest_main.cc) +# If the CMake version supports it, attach header directory information +# to the targets for when we are part of a parent build (ie being pulled +# in via add_subdirectory() rather than being a standalone build). +if (DEFINED CMAKE_VERSION AND NOT "${CMAKE_VERSION}" VERSION_LESS "2.8.11") + target_include_directories(gtest SYSTEM INTERFACE + "$" + "$/${CMAKE_INSTALL_INCLUDEDIR}>") + target_include_directories(gtest_main SYSTEM INTERFACE + "$" + "$/${CMAKE_INSTALL_INCLUDEDIR}>") +endif() +target_link_libraries(gtest_main PUBLIC gtest) + +######################################################################## +# +# Install rules +install_project(gtest gtest_main) + +######################################################################## +# +# Samples on how to link user tests with gtest or gtest_main. +# +# They are not built by default. To build them, set the +# gtest_build_samples option to ON. You can do it by running ccmake +# or specifying the -Dgtest_build_samples=ON flag when running cmake. + +if (gtest_build_samples) + cxx_executable(sample1_unittest samples gtest_main samples/sample1.cc) + cxx_executable(sample2_unittest samples gtest_main samples/sample2.cc) + cxx_executable(sample3_unittest samples gtest_main) + cxx_executable(sample4_unittest samples gtest_main samples/sample4.cc) + cxx_executable(sample5_unittest samples gtest_main samples/sample1.cc) + cxx_executable(sample6_unittest samples gtest_main) + cxx_executable(sample7_unittest samples gtest_main) + cxx_executable(sample8_unittest samples gtest_main) + cxx_executable(sample9_unittest samples gtest) + cxx_executable(sample10_unittest samples gtest) +endif() + +######################################################################## +# +# Google Test's own tests. +# +# You can skip this section if you aren't interested in testing +# Google Test itself. +# +# The tests are not built by default. To build them, set the +# gtest_build_tests option to ON. You can do it by running ccmake +# or specifying the -Dgtest_build_tests=ON flag when running cmake. + +if (gtest_build_tests) + # This must be set in the root directory for the tests to be run by + # 'make test' or ctest. + enable_testing() + + if (WIN32) + file(GENERATE OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/$/RunTest.ps1" + CONTENT +"$project_bin = \"${CMAKE_BINARY_DIR}/bin/$\" +$env:Path = \"$project_bin;$env:Path\" +& $args") + elseif (MINGW) + file(GENERATE OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/RunTest.ps1" + CONTENT +"$project_bin = (cygpath --windows ${CMAKE_BINARY_DIR}/bin) +$env:Path = \"$project_bin;$env:Path\" +& $args") + endif() + + ############################################################ + # C++ tests built with standard compiler flags. + + cxx_test(googletest-death-test-test gtest_main) + cxx_test(gtest_environment_test gtest) + cxx_test(googletest-filepath-test gtest_main) + cxx_test(googletest-listener-test gtest_main) + cxx_test(gtest_main_unittest gtest_main) + cxx_test(googletest-message-test gtest_main) + cxx_test(gtest_no_test_unittest gtest) + cxx_test(googletest-options-test gtest_main) + cxx_test(googletest-param-test-test gtest + test/googletest-param-test2-test.cc) + cxx_test(googletest-port-test gtest_main) + cxx_test(gtest_pred_impl_unittest gtest_main) + cxx_test(gtest_premature_exit_test gtest + test/gtest_premature_exit_test.cc) + cxx_test(googletest-printers-test gtest_main) + cxx_test(gtest_prod_test gtest_main + test/production.cc) + cxx_test(gtest_repeat_test gtest) + cxx_test(gtest_sole_header_test gtest_main) + cxx_test(gtest_stress_test gtest) + cxx_test(googletest-test-part-test gtest_main) + cxx_test(gtest_throw_on_failure_ex_test gtest) + cxx_test(gtest-typed-test_test gtest_main + test/gtest-typed-test2_test.cc) + cxx_test(gtest_unittest gtest_main) + cxx_test(gtest-unittest-api_test gtest) + cxx_test(gtest_skip_in_environment_setup_test gtest_main) + cxx_test(gtest_skip_test gtest_main) + + ############################################################ + # C++ tests built with non-standard compiler flags. + + # MSVC 7.1 does not support STL with exceptions disabled. + if (NOT MSVC OR MSVC_VERSION GREATER 1310) + cxx_library(gtest_no_exception "${cxx_no_exception}" + src/gtest-all.cc) + cxx_library(gtest_main_no_exception "${cxx_no_exception}" + src/gtest-all.cc src/gtest_main.cc) + endif() + cxx_library(gtest_main_no_rtti "${cxx_no_rtti}" + src/gtest-all.cc src/gtest_main.cc) + + cxx_test_with_flags(gtest-death-test_ex_nocatch_test + "${cxx_exception} -DGTEST_ENABLE_CATCH_EXCEPTIONS_=0" + gtest test/googletest-death-test_ex_test.cc) + cxx_test_with_flags(gtest-death-test_ex_catch_test + "${cxx_exception} -DGTEST_ENABLE_CATCH_EXCEPTIONS_=1" + gtest test/googletest-death-test_ex_test.cc) + + cxx_test_with_flags(gtest_no_rtti_unittest "${cxx_no_rtti}" + gtest_main_no_rtti test/gtest_unittest.cc) + + cxx_shared_library(gtest_dll "${cxx_default}" + src/gtest-all.cc src/gtest_main.cc) + + cxx_executable_with_flags(gtest_dll_test_ "${cxx_default}" + gtest_dll test/gtest_all_test.cc) + set_target_properties(gtest_dll_test_ + PROPERTIES + COMPILE_DEFINITIONS "GTEST_LINKED_AS_SHARED_LIBRARY=1") + + ############################################################ + # Python tests. + + cxx_executable(googletest-break-on-failure-unittest_ test gtest) + py_test(googletest-break-on-failure-unittest) + + py_test(gtest_skip_environment_check_output_test) + + # Visual Studio .NET 2003 does not support STL with exceptions disabled. + if (NOT MSVC OR MSVC_VERSION GREATER 1310) # 1310 is Visual Studio .NET 2003 + cxx_executable_with_flags( + googletest-catch-exceptions-no-ex-test_ + "${cxx_no_exception}" + gtest_main_no_exception + test/googletest-catch-exceptions-test_.cc) + endif() + + cxx_executable_with_flags( + googletest-catch-exceptions-ex-test_ + "${cxx_exception}" + gtest_main + test/googletest-catch-exceptions-test_.cc) + py_test(googletest-catch-exceptions-test) + + cxx_executable(googletest-color-test_ test gtest) + py_test(googletest-color-test) + + cxx_executable(googletest-env-var-test_ test gtest) + py_test(googletest-env-var-test) + + cxx_executable(googletest-filter-unittest_ test gtest) + py_test(googletest-filter-unittest) + + cxx_executable(gtest_help_test_ test gtest_main) + py_test(gtest_help_test) + + cxx_executable(googletest-list-tests-unittest_ test gtest) + py_test(googletest-list-tests-unittest) + + cxx_executable(googletest-output-test_ test gtest) + py_test(googletest-output-test --no_stacktrace_support) + + cxx_executable(googletest-shuffle-test_ test gtest) + py_test(googletest-shuffle-test) + + # MSVC 7.1 does not support STL with exceptions disabled. + if (NOT MSVC OR MSVC_VERSION GREATER 1310) + cxx_executable(googletest-throw-on-failure-test_ test gtest_no_exception) + set_target_properties(googletest-throw-on-failure-test_ + PROPERTIES + COMPILE_FLAGS "${cxx_no_exception}") + py_test(googletest-throw-on-failure-test) + endif() + + cxx_executable(googletest-uninitialized-test_ test gtest) + py_test(googletest-uninitialized-test) + + cxx_executable(gtest_xml_outfile1_test_ test gtest_main) + cxx_executable(gtest_xml_outfile2_test_ test gtest_main) + py_test(gtest_xml_outfiles_test) + py_test(googletest-json-outfiles-test) + + cxx_executable(gtest_xml_output_unittest_ test gtest) + py_test(gtest_xml_output_unittest --no_stacktrace_support) + py_test(googletest-json-output-unittest --no_stacktrace_support) +endif() diff --git a/deps/googletest/CONTRIBUTORS b/deps/googletest/CONTRIBUTORS new file mode 100644 index 000000000..feae2fc04 --- /dev/null +++ b/deps/googletest/CONTRIBUTORS @@ -0,0 +1,37 @@ +# This file contains a list of people who've made non-trivial +# contribution to the Google C++ Testing Framework project. People +# who commit code to the project are encouraged to add their names +# here. Please keep the list sorted by first names. + +Ajay Joshi +Balázs Dán +Bharat Mediratta +Chandler Carruth +Chris Prince +Chris Taylor +Dan Egnor +Eric Roman +Hady Zalek +Jeffrey Yasskin +Jói Sigurðsson +Keir Mierle +Keith Ray +Kenton Varda +Manuel Klimek +Markus Heule +Mika Raento +Miklós Fazekas +Pasi Valminen +Patrick Hanna +Patrick Riley +Peter Kaminski +Preston Jackson +Rainer Klaffenboeck +Russ Cox +Russ Rufer +Sean Mcafee +Sigurður Ásgeirsson +Tracy Bialik +Vadim Berman +Vlad Losev +Zhanyong Wan diff --git a/deps/googletest/LICENSE b/deps/googletest/LICENSE new file mode 100644 index 000000000..1941a11f8 --- /dev/null +++ b/deps/googletest/LICENSE @@ -0,0 +1,28 @@ +Copyright 2008, Google Inc. +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are +met: + + * Redistributions of source code must retain the above copyright +notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above +copyright notice, this list of conditions and the following disclaimer +in the documentation and/or other materials provided with the +distribution. + * Neither the name of Google Inc. nor the names of its +contributors may be used to endorse or promote products derived from +this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. diff --git a/deps/googletest/cmake/Config.cmake.in b/deps/googletest/cmake/Config.cmake.in new file mode 100644 index 000000000..12be4498b --- /dev/null +++ b/deps/googletest/cmake/Config.cmake.in @@ -0,0 +1,9 @@ +@PACKAGE_INIT@ +include(CMakeFindDependencyMacro) +if (@GTEST_HAS_PTHREAD@) + set(THREADS_PREFER_PTHREAD_FLAG @THREADS_PREFER_PTHREAD_FLAG@) + find_dependency(Threads) +endif() + +include("${CMAKE_CURRENT_LIST_DIR}/@targets_export_name@.cmake") +check_required_components("@project_name@") diff --git a/deps/googletest/cmake/gtest.pc.in b/deps/googletest/cmake/gtest.pc.in new file mode 100644 index 000000000..9aae29e26 --- /dev/null +++ b/deps/googletest/cmake/gtest.pc.in @@ -0,0 +1,10 @@ +prefix=${pcfiledir}/../.. +libdir=${prefix}/@CMAKE_INSTALL_LIBDIR@ +includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@ + +Name: gtest +Description: GoogleTest (without main() function) +Version: @PROJECT_VERSION@ +URL: https://github.com/google/googletest +Libs: -L${libdir} -lgtest @CMAKE_THREAD_LIBS_INIT@ +Cflags: -I${includedir} @GTEST_HAS_PTHREAD_MACRO@ @CMAKE_THREAD_LIBS_INIT@ diff --git a/deps/googletest/cmake/gtest_main.pc.in b/deps/googletest/cmake/gtest_main.pc.in new file mode 100644 index 000000000..915f2973a --- /dev/null +++ b/deps/googletest/cmake/gtest_main.pc.in @@ -0,0 +1,11 @@ +prefix=${pcfiledir}/../.. +libdir=${prefix}/@CMAKE_INSTALL_LIBDIR@ +includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@ + +Name: gtest_main +Description: GoogleTest (with main() function) +Version: @PROJECT_VERSION@ +URL: https://github.com/google/googletest +Requires: gtest +Libs: -L${libdir} -lgtest_main @CMAKE_THREAD_LIBS_INIT@ +Cflags: -I${includedir} @GTEST_HAS_PTHREAD_MACRO@ @CMAKE_THREAD_LIBS_INIT@ diff --git a/deps/googletest/cmake/internal_utils.cmake b/deps/googletest/cmake/internal_utils.cmake new file mode 100644 index 000000000..bc49a0b25 --- /dev/null +++ b/deps/googletest/cmake/internal_utils.cmake @@ -0,0 +1,356 @@ +# Defines functions and macros useful for building Google Test and +# Google Mock. +# +# Note: +# +# - This file will be run twice when building Google Mock (once via +# Google Test's CMakeLists.txt, and once via Google Mock's). +# Therefore it shouldn't have any side effects other than defining +# the functions and macros. +# +# - The functions/macros defined in this file may depend on Google +# Test and Google Mock's option() definitions, and thus must be +# called *after* the options have been defined. + +# Tweaks CMake's default compiler/linker settings to suit Google Test's needs. +# +# This must be a macro(), as inside a function string() can only +# update variables in the function scope. +macro(fix_default_compiler_settings_) + if (MSVC) + # For MSVC, CMake sets certain flags to defaults we want to override. + # This replacement code is taken from sample in the CMake Wiki at + # https://gitlab.kitware.com/cmake/community/wikis/FAQ#dynamic-replace. + foreach (flag_var + CMAKE_C_FLAGS CMAKE_C_FLAGS_DEBUG CMAKE_C_FLAGS_RELEASE + CMAKE_C_FLAGS_MINSIZEREL CMAKE_C_FLAGS_RELWITHDEBINFO + CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_DEBUG CMAKE_CXX_FLAGS_RELEASE + CMAKE_CXX_FLAGS_MINSIZEREL CMAKE_CXX_FLAGS_RELWITHDEBINFO) + if (NOT BUILD_SHARED_LIBS AND NOT gtest_force_shared_crt) + # When Google Test is built as a shared library, it should also use + # shared runtime libraries. Otherwise, it may end up with multiple + # copies of runtime library data in different modules, resulting in + # hard-to-find crashes. When it is built as a static library, it is + # preferable to use CRT as static libraries, as we don't have to rely + # on CRT DLLs being available. CMake always defaults to using shared + # CRT libraries, so we override that default here. + string(REPLACE "/MD" "-MT" ${flag_var} "${${flag_var}}") + endif() + + # We prefer more strict warning checking for building Google Test. + # Replaces /W3 with /W4 in defaults. + string(REPLACE "/W3" "/W4" ${flag_var} "${${flag_var}}") + + # Prevent D9025 warning for targets that have exception handling + # turned off (/EHs-c- flag). Where required, exceptions are explicitly + # re-enabled using the cxx_exception_flags variable. + string(REPLACE "/EHsc" "" ${flag_var} "${${flag_var}}") + endforeach() + endif() +endmacro() + +# Defines the compiler/linker flags used to build Google Test and +# Google Mock. You can tweak these definitions to suit your need. A +# variable's value is empty before it's explicitly assigned to. +macro(config_compiler_and_linker) + # Note: pthreads on MinGW is not supported, even if available + # instead, we use windows threading primitives + unset(GTEST_HAS_PTHREAD) + if (NOT gtest_disable_pthreads AND NOT MINGW) + # Defines CMAKE_USE_PTHREADS_INIT and CMAKE_THREAD_LIBS_INIT. + find_package(Threads) + if (CMAKE_USE_PTHREADS_INIT) + set(GTEST_HAS_PTHREAD ON) + endif() + endif() + + fix_default_compiler_settings_() + if (MSVC) + # Newlines inside flags variables break CMake's NMake generator. + # TODO(vladl@google.com): Add -RTCs and -RTCu to debug builds. + set(cxx_base_flags "-GS -W4 -WX -wd4251 -wd4275 -nologo -J -Zi") + set(cxx_base_flags "${cxx_base_flags} -D_UNICODE -DUNICODE -DWIN32 -D_WIN32") + set(cxx_base_flags "${cxx_base_flags} -DSTRICT -DWIN32_LEAN_AND_MEAN") + set(cxx_exception_flags "-EHsc -D_HAS_EXCEPTIONS=1") + set(cxx_no_exception_flags "-EHs-c- -D_HAS_EXCEPTIONS=0") + set(cxx_no_rtti_flags "-GR-") + # Suppress "unreachable code" warning + # http://stackoverflow.com/questions/3232669 explains the issue. + set(cxx_base_flags "${cxx_base_flags} -wd4702") + elseif (CMAKE_CXX_COMPILER_ID STREQUAL "Clang") +# TODO: add Werror back when LLVM compiler stop generate warnings on CUDA 10.2 +# set(cxx_base_flags "-Wall -Wshadow -Werror -Wconversion") + set(cxx_base_flags "-Wall -Wshadow -Wconversion") + set(cxx_exception_flags "-fexceptions") + set(cxx_no_exception_flags "-fno-exceptions") + set(cxx_strict_flags "-W -Wpointer-arith -Wreturn-type -Wcast-qual -Wwrite-strings -Wswitch -Wunused-parameter -Wcast-align -Wchar-subscripts -Winline -Wredundant-decls") + set(cxx_no_rtti_flags "-fno-rtti") + elseif (CMAKE_COMPILER_IS_GNUCXX) + set(cxx_base_flags "-Wall -Wshadow -Werror") + if(NOT CMAKE_CXX_COMPILER_VERSION VERSION_LESS 7.0.0) + set(cxx_base_flags "${cxx_base_flags} -Wno-error=dangling-else") + endif() + set(cxx_exception_flags "-fexceptions") + set(cxx_no_exception_flags "-fno-exceptions") + # Until version 4.3.2, GCC doesn't define a macro to indicate + # whether RTTI is enabled. Therefore we define GTEST_HAS_RTTI + # explicitly. + set(cxx_no_rtti_flags "-fno-rtti -DGTEST_HAS_RTTI=0") + set(cxx_strict_flags + "-Wextra -Wno-unused-parameter -Wno-missing-field-initializers") + elseif (CMAKE_CXX_COMPILER_ID STREQUAL "SunPro") + set(cxx_exception_flags "-features=except") + # Sun Pro doesn't provide macros to indicate whether exceptions and + # RTTI are enabled, so we define GTEST_HAS_* explicitly. + set(cxx_no_exception_flags "-features=no%except -DGTEST_HAS_EXCEPTIONS=0") + set(cxx_no_rtti_flags "-features=no%rtti -DGTEST_HAS_RTTI=0") + elseif (CMAKE_CXX_COMPILER_ID STREQUAL "VisualAge" OR + CMAKE_CXX_COMPILER_ID STREQUAL "XL") + # CMake 2.8 changes Visual Age's compiler ID to "XL". + set(cxx_exception_flags "-qeh") + set(cxx_no_exception_flags "-qnoeh") + # Until version 9.0, Visual Age doesn't define a macro to indicate + # whether RTTI is enabled. Therefore we define GTEST_HAS_RTTI + # explicitly. + set(cxx_no_rtti_flags "-qnortti -DGTEST_HAS_RTTI=0") + elseif (CMAKE_CXX_COMPILER_ID STREQUAL "HP") + set(cxx_base_flags "-AA -mt") + set(cxx_exception_flags "-DGTEST_HAS_EXCEPTIONS=1") + set(cxx_no_exception_flags "+noeh -DGTEST_HAS_EXCEPTIONS=0") + # RTTI can not be disabled in HP aCC compiler. + set(cxx_no_rtti_flags "") + endif() + + # The pthreads library is available and allowed? + if (DEFINED GTEST_HAS_PTHREAD) + set(GTEST_HAS_PTHREAD_MACRO "-DGTEST_HAS_PTHREAD=1") + else() + set(GTEST_HAS_PTHREAD_MACRO "-DGTEST_HAS_PTHREAD=0") + endif() + set(cxx_base_flags "${cxx_base_flags} ${GTEST_HAS_PTHREAD_MACRO}") + + # For building gtest's own tests and samples. + set(cxx_exception "${cxx_base_flags} ${cxx_exception_flags}") + set(cxx_no_exception + "${CMAKE_CXX_FLAGS} ${cxx_base_flags} ${cxx_no_exception_flags}") + set(cxx_default "${cxx_exception}") + set(cxx_no_rtti "${cxx_default} ${cxx_no_rtti_flags}") + + # For building the gtest libraries. + set(cxx_strict "${cxx_default} ${cxx_strict_flags}") +endmacro() + +# Defines the gtest & gtest_main libraries. User tests should link +# with one of them. +function(cxx_library_with_type name type cxx_flags) + # type can be either STATIC or SHARED to denote a static or shared library. + # ARGN refers to additional arguments after 'cxx_flags'. + add_library(${name} ${type} ${ARGN}) + set_target_properties(${name} + PROPERTIES + COMPILE_FLAGS "${cxx_flags}") + # Generate debug library name with a postfix. + set_target_properties(${name} + PROPERTIES + DEBUG_POSTFIX "d") + # Set the output directory for build artifacts + set_target_properties(${name} + PROPERTIES + RUNTIME_OUTPUT_DIRECTORY "${CMAKE_BINARY_DIR}/bin" + LIBRARY_OUTPUT_DIRECTORY "${CMAKE_BINARY_DIR}/lib" + ARCHIVE_OUTPUT_DIRECTORY "${CMAKE_BINARY_DIR}/lib" + PDB_OUTPUT_DIRECTORY "${CMAKE_BINARY_DIR}/bin") + # make PDBs match library name + get_target_property(pdb_debug_postfix ${name} DEBUG_POSTFIX) + set_target_properties(${name} + PROPERTIES + PDB_NAME "${name}" + PDB_NAME_DEBUG "${name}${pdb_debug_postfix}" + COMPILE_PDB_NAME "${name}" + COMPILE_PDB_NAME_DEBUG "${name}${pdb_debug_postfix}") + + if (BUILD_SHARED_LIBS OR type STREQUAL "SHARED") + set_target_properties(${name} + PROPERTIES + COMPILE_DEFINITIONS "GTEST_CREATE_SHARED_LIBRARY=1") + if (NOT "${CMAKE_VERSION}" VERSION_LESS "2.8.11") + target_compile_definitions(${name} INTERFACE + $) + endif() + endif() + if (DEFINED GTEST_HAS_PTHREAD) + if ("${CMAKE_VERSION}" VERSION_LESS "3.1.0") + set(threads_spec ${CMAKE_THREAD_LIBS_INIT}) + else() + set(threads_spec Threads::Threads) + endif() + target_link_libraries(${name} PUBLIC ${threads_spec}) + endif() +endfunction() + +######################################################################## +# +# Helper functions for creating build targets. + +function(cxx_shared_library name cxx_flags) + cxx_library_with_type(${name} SHARED "${cxx_flags}" ${ARGN}) +endfunction() + +function(cxx_library name cxx_flags) + cxx_library_with_type(${name} "" "${cxx_flags}" ${ARGN}) +endfunction() + +# cxx_executable_with_flags(name cxx_flags libs srcs...) +# +# creates a named C++ executable that depends on the given libraries and +# is built from the given source files with the given compiler flags. +function(cxx_executable_with_flags name cxx_flags libs) + add_executable(${name} ${ARGN}) + if (MSVC) + # BigObj required for tests. + set(cxx_flags "${cxx_flags} -bigobj") + endif() + if (cxx_flags) + set_target_properties(${name} + PROPERTIES + COMPILE_FLAGS "${cxx_flags}") + endif() + if (BUILD_SHARED_LIBS) + set_target_properties(${name} + PROPERTIES + COMPILE_DEFINITIONS "GTEST_LINKED_AS_SHARED_LIBRARY=1") + endif() + # To support mixing linking in static and dynamic libraries, link each + # library in with an extra call to target_link_libraries. + foreach (lib "${libs}") + target_link_libraries(${name} ${lib}) + endforeach() +endfunction() + +# cxx_executable(name dir lib srcs...) +# +# creates a named target that depends on the given libs and is built +# from the given source files. dir/name.cc is implicitly included in +# the source file list. +function(cxx_executable name dir libs) + cxx_executable_with_flags( + ${name} "${cxx_default}" "${libs}" "${dir}/${name}.cc" ${ARGN}) +endfunction() + +# Sets PYTHONINTERP_FOUND and PYTHON_EXECUTABLE. +find_package(PythonInterp) + +# cxx_test_with_flags(name cxx_flags libs srcs...) +# +# creates a named C++ test that depends on the given libs and is built +# from the given source files with the given compiler flags. +function(cxx_test_with_flags name cxx_flags libs) + cxx_executable_with_flags(${name} "${cxx_flags}" "${libs}" ${ARGN}) + if (WIN32 OR MINGW) + add_test(NAME ${name} + COMMAND "powershell" "-Command" "${CMAKE_CURRENT_BINARY_DIR}/$/RunTest.ps1" "$") + else() + add_test(NAME ${name} + COMMAND "$") + endif() +endfunction() + +# cxx_test(name libs srcs...) +# +# creates a named test target that depends on the given libs and is +# built from the given source files. Unlike cxx_test_with_flags, +# test/name.cc is already implicitly included in the source file list. +function(cxx_test name libs) + cxx_test_with_flags("${name}" "${cxx_default}" "${libs}" + "test/${name}.cc" ${ARGN}) +endfunction() + +# py_test(name) +# +# creates a Python test with the given name whose main module is in +# test/name.py. It does nothing if Python is not installed. +function(py_test name) + if (PYTHONINTERP_FOUND) + if ("${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION}" VERSION_GREATER 3.1) + if (CMAKE_CONFIGURATION_TYPES) + # Multi-configuration build generators as for Visual Studio save + # output in a subdirectory of CMAKE_CURRENT_BINARY_DIR (Debug, + # Release etc.), so we have to provide it here. + if (WIN32 OR MINGW) + add_test(NAME ${name} + COMMAND powershell -Command ${CMAKE_CURRENT_BINARY_DIR}/$/RunTest.ps1 + ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py + --build_dir=${CMAKE_CURRENT_BINARY_DIR}/$ ${ARGN}) + else() + add_test(NAME ${name} + COMMAND ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py + --build_dir=${CMAKE_CURRENT_BINARY_DIR}/$ ${ARGN}) + endif() + else (CMAKE_CONFIGURATION_TYPES) + # Single-configuration build generators like Makefile generators + # don't have subdirs below CMAKE_CURRENT_BINARY_DIR. + if (WIN32 OR MINGW) + add_test(NAME ${name} + COMMAND powershell -Command ${CMAKE_CURRENT_BINARY_DIR}/RunTest.ps1 + ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py + --build_dir=${CMAKE_CURRENT_BINARY_DIR} ${ARGN}) + else() + add_test(NAME ${name} + COMMAND ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py + --build_dir=${CMAKE_CURRENT_BINARY_DIR} ${ARGN}) + endif() + endif (CMAKE_CONFIGURATION_TYPES) + else() + # ${CMAKE_CURRENT_BINARY_DIR} is known at configuration time, so we can + # directly bind it from cmake. ${CTEST_CONFIGURATION_TYPE} is known + # only at ctest runtime (by calling ctest -c ), so + # we have to escape $ to delay variable substitution here. + if (WIN32 OR MINGW) + add_test(NAME ${name} + COMMAND powershell -Command ${CMAKE_CURRENT_BINARY_DIR}/RunTest.ps1 + ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py + --build_dir=${CMAKE_CURRENT_BINARY_DIR}/\${CTEST_CONFIGURATION_TYPE} ${ARGN}) + else() + add_test(NAME ${name} + COMMAND ${PYTHON_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}/test/${name}.py + --build_dir=${CMAKE_CURRENT_BINARY_DIR}/\${CTEST_CONFIGURATION_TYPE} ${ARGN}) + endif() + endif() + endif(PYTHONINTERP_FOUND) +endfunction() + +# install_project(targets...) +# +# Installs the specified targets and configures the associated pkgconfig files. +function(install_project) + if(INSTALL_GTEST) + install(DIRECTORY "${PROJECT_SOURCE_DIR}/include/" + DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}") + # Install the project targets. + install(TARGETS ${ARGN} + EXPORT ${targets_export_name} + RUNTIME DESTINATION "${CMAKE_INSTALL_BINDIR}" + ARCHIVE DESTINATION "${CMAKE_INSTALL_LIBDIR}" + LIBRARY DESTINATION "${CMAKE_INSTALL_LIBDIR}") + if(CMAKE_CXX_COMPILER_ID MATCHES "MSVC") + # Install PDBs + foreach(t ${ARGN}) + get_target_property(t_pdb_name ${t} COMPILE_PDB_NAME) + get_target_property(t_pdb_name_debug ${t} COMPILE_PDB_NAME_DEBUG) + get_target_property(t_pdb_output_directory ${t} PDB_OUTPUT_DIRECTORY) + install(FILES + "${t_pdb_output_directory}/\${CMAKE_INSTALL_CONFIG_NAME}/$<$:${t_pdb_name_debug}>$<$>:${t_pdb_name}>.pdb" + DESTINATION ${CMAKE_INSTALL_LIBDIR} + OPTIONAL) + endforeach() + endif() + # Configure and install pkgconfig files. + foreach(t ${ARGN}) + set(configured_pc "${generated_dir}/${t}.pc") + configure_file("${PROJECT_SOURCE_DIR}/cmake/${t}.pc.in" + "${configured_pc}" @ONLY) + install(FILES "${configured_pc}" + DESTINATION "${CMAKE_INSTALL_LIBDIR}/pkgconfig") + endforeach() + endif() +endfunction() diff --git a/deps/googletest/cmake/libgtest.la.in b/deps/googletest/cmake/libgtest.la.in new file mode 100644 index 000000000..840c83885 --- /dev/null +++ b/deps/googletest/cmake/libgtest.la.in @@ -0,0 +1,21 @@ +# libgtest.la - a libtool library file +# Generated by libtool (GNU libtool) 2.4.6 + +# Please DO NOT delete this file! +# It is necessary for linking the library. + +# Names of this library. +library_names='libgtest.so' + +# Is this an already installed library? +installed=yes + +# Should we warn about portability when linking against -modules? +shouldnotlink=no + +# Files to dlopen/dlpreopen +dlopen='' +dlpreopen='' + +# Directory that this library needs to be installed in: +libdir='@CMAKE_INSTALL_FULL_LIBDIR@' diff --git a/deps/googletest/include/gtest/gtest-death-test.h b/deps/googletest/include/gtest/gtest-death-test.h new file mode 100644 index 000000000..0eb5b279f --- /dev/null +++ b/deps/googletest/include/gtest/gtest-death-test.h @@ -0,0 +1,343 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file defines the public API for death tests. It is +// #included by gtest.h so a user doesn't need to include this +// directly. +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_ +#define GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_ + +#include "gtest/internal/gtest-death-test-internal.h" + +namespace testing { + +// This flag controls the style of death tests. Valid values are "threadsafe", +// meaning that the death test child process will re-execute the test binary +// from the start, running only a single death test, or "fast", +// meaning that the child process will execute the test logic immediately +// after forking. +GTEST_DECLARE_string_(death_test_style); + +#if GTEST_HAS_DEATH_TEST + +namespace internal { + +// Returns a Boolean value indicating whether the caller is currently +// executing in the context of the death test child process. Tools such as +// Valgrind heap checkers may need this to modify their behavior in death +// tests. IMPORTANT: This is an internal utility. Using it may break the +// implementation of death tests. User code MUST NOT use it. +GTEST_API_ bool InDeathTestChild(); + +} // namespace internal + +// The following macros are useful for writing death tests. + +// Here's what happens when an ASSERT_DEATH* or EXPECT_DEATH* is +// executed: +// +// 1. It generates a warning if there is more than one active +// thread. This is because it's safe to fork() or clone() only +// when there is a single thread. +// +// 2. The parent process clone()s a sub-process and runs the death +// test in it; the sub-process exits with code 0 at the end of the +// death test, if it hasn't exited already. +// +// 3. The parent process waits for the sub-process to terminate. +// +// 4. The parent process checks the exit code and error message of +// the sub-process. +// +// Examples: +// +// ASSERT_DEATH(server.SendMessage(56, "Hello"), "Invalid port number"); +// for (int i = 0; i < 5; i++) { +// EXPECT_DEATH(server.ProcessRequest(i), +// "Invalid request .* in ProcessRequest()") +// << "Failed to die on request " << i; +// } +// +// ASSERT_EXIT(server.ExitNow(), ::testing::ExitedWithCode(0), "Exiting"); +// +// bool KilledBySIGHUP(int exit_code) { +// return WIFSIGNALED(exit_code) && WTERMSIG(exit_code) == SIGHUP; +// } +// +// ASSERT_EXIT(client.HangUpServer(), KilledBySIGHUP, "Hanging up!"); +// +// On the regular expressions used in death tests: +// +// GOOGLETEST_CM0005 DO NOT DELETE +// On POSIX-compliant systems (*nix), we use the library, +// which uses the POSIX extended regex syntax. +// +// On other platforms (e.g. Windows or Mac), we only support a simple regex +// syntax implemented as part of Google Test. This limited +// implementation should be enough most of the time when writing +// death tests; though it lacks many features you can find in PCRE +// or POSIX extended regex syntax. For example, we don't support +// union ("x|y"), grouping ("(xy)"), brackets ("[xy]"), and +// repetition count ("x{5,7}"), among others. +// +// Below is the syntax that we do support. We chose it to be a +// subset of both PCRE and POSIX extended regex, so it's easy to +// learn wherever you come from. In the following: 'A' denotes a +// literal character, period (.), or a single \\ escape sequence; +// 'x' and 'y' denote regular expressions; 'm' and 'n' are for +// natural numbers. +// +// c matches any literal character c +// \\d matches any decimal digit +// \\D matches any character that's not a decimal digit +// \\f matches \f +// \\n matches \n +// \\r matches \r +// \\s matches any ASCII whitespace, including \n +// \\S matches any character that's not a whitespace +// \\t matches \t +// \\v matches \v +// \\w matches any letter, _, or decimal digit +// \\W matches any character that \\w doesn't match +// \\c matches any literal character c, which must be a punctuation +// . matches any single character except \n +// A? matches 0 or 1 occurrences of A +// A* matches 0 or many occurrences of A +// A+ matches 1 or many occurrences of A +// ^ matches the beginning of a string (not that of each line) +// $ matches the end of a string (not that of each line) +// xy matches x followed by y +// +// If you accidentally use PCRE or POSIX extended regex features +// not implemented by us, you will get a run-time failure. In that +// case, please try to rewrite your regular expression within the +// above syntax. +// +// This implementation is *not* meant to be as highly tuned or robust +// as a compiled regex library, but should perform well enough for a +// death test, which already incurs significant overhead by launching +// a child process. +// +// Known caveats: +// +// A "threadsafe" style death test obtains the path to the test +// program from argv[0] and re-executes it in the sub-process. For +// simplicity, the current implementation doesn't search the PATH +// when launching the sub-process. This means that the user must +// invoke the test program via a path that contains at least one +// path separator (e.g. path/to/foo_test and +// /absolute/path/to/bar_test are fine, but foo_test is not). This +// is rarely a problem as people usually don't put the test binary +// directory in PATH. +// + +// Asserts that a given statement causes the program to exit, with an +// integer exit status that satisfies predicate, and emitting error output +// that matches regex. +# define ASSERT_EXIT(statement, predicate, regex) \ + GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_FATAL_FAILURE_) + +// Like ASSERT_EXIT, but continues on to successive tests in the +// test suite, if any: +# define EXPECT_EXIT(statement, predicate, regex) \ + GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_NONFATAL_FAILURE_) + +// Asserts that a given statement causes the program to exit, either by +// explicitly exiting with a nonzero exit code or being killed by a +// signal, and emitting error output that matches regex. +# define ASSERT_DEATH(statement, regex) \ + ASSERT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex) + +// Like ASSERT_DEATH, but continues on to successive tests in the +// test suite, if any: +# define EXPECT_DEATH(statement, regex) \ + EXPECT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex) + +// Two predicate classes that can be used in {ASSERT,EXPECT}_EXIT*: + +// Tests that an exit code describes a normal exit with a given exit code. +class GTEST_API_ ExitedWithCode { + public: + explicit ExitedWithCode(int exit_code); + bool operator()(int exit_status) const; + private: + // No implementation - assignment is unsupported. + void operator=(const ExitedWithCode& other); + + const int exit_code_; +}; + +# if !GTEST_OS_WINDOWS && !GTEST_OS_FUCHSIA +// Tests that an exit code describes an exit due to termination by a +// given signal. +// GOOGLETEST_CM0006 DO NOT DELETE +class GTEST_API_ KilledBySignal { + public: + explicit KilledBySignal(int signum); + bool operator()(int exit_status) const; + private: + const int signum_; +}; +# endif // !GTEST_OS_WINDOWS + +// EXPECT_DEBUG_DEATH asserts that the given statements die in debug mode. +// The death testing framework causes this to have interesting semantics, +// since the sideeffects of the call are only visible in opt mode, and not +// in debug mode. +// +// In practice, this can be used to test functions that utilize the +// LOG(DFATAL) macro using the following style: +// +// int DieInDebugOr12(int* sideeffect) { +// if (sideeffect) { +// *sideeffect = 12; +// } +// LOG(DFATAL) << "death"; +// return 12; +// } +// +// TEST(TestSuite, TestDieOr12WorksInDgbAndOpt) { +// int sideeffect = 0; +// // Only asserts in dbg. +// EXPECT_DEBUG_DEATH(DieInDebugOr12(&sideeffect), "death"); +// +// #ifdef NDEBUG +// // opt-mode has sideeffect visible. +// EXPECT_EQ(12, sideeffect); +// #else +// // dbg-mode no visible sideeffect. +// EXPECT_EQ(0, sideeffect); +// #endif +// } +// +// This will assert that DieInDebugReturn12InOpt() crashes in debug +// mode, usually due to a DCHECK or LOG(DFATAL), but returns the +// appropriate fallback value (12 in this case) in opt mode. If you +// need to test that a function has appropriate side-effects in opt +// mode, include assertions against the side-effects. A general +// pattern for this is: +// +// EXPECT_DEBUG_DEATH({ +// // Side-effects here will have an effect after this statement in +// // opt mode, but none in debug mode. +// EXPECT_EQ(12, DieInDebugOr12(&sideeffect)); +// }, "death"); +// +# ifdef NDEBUG + +# define EXPECT_DEBUG_DEATH(statement, regex) \ + GTEST_EXECUTE_STATEMENT_(statement, regex) + +# define ASSERT_DEBUG_DEATH(statement, regex) \ + GTEST_EXECUTE_STATEMENT_(statement, regex) + +# else + +# define EXPECT_DEBUG_DEATH(statement, regex) \ + EXPECT_DEATH(statement, regex) + +# define ASSERT_DEBUG_DEATH(statement, regex) \ + ASSERT_DEATH(statement, regex) + +# endif // NDEBUG for EXPECT_DEBUG_DEATH +#endif // GTEST_HAS_DEATH_TEST + +// This macro is used for implementing macros such as +// EXPECT_DEATH_IF_SUPPORTED and ASSERT_DEATH_IF_SUPPORTED on systems where +// death tests are not supported. Those macros must compile on such systems +// iff EXPECT_DEATH and ASSERT_DEATH compile with the same parameters on +// systems that support death tests. This allows one to write such a macro +// on a system that does not support death tests and be sure that it will +// compile on a death-test supporting system. It is exposed publicly so that +// systems that have death-tests with stricter requirements than +// GTEST_HAS_DEATH_TEST can write their own equivalent of +// EXPECT_DEATH_IF_SUPPORTED and ASSERT_DEATH_IF_SUPPORTED. +// +// Parameters: +// statement - A statement that a macro such as EXPECT_DEATH would test +// for program termination. This macro has to make sure this +// statement is compiled but not executed, to ensure that +// EXPECT_DEATH_IF_SUPPORTED compiles with a certain +// parameter iff EXPECT_DEATH compiles with it. +// regex - A regex that a macro such as EXPECT_DEATH would use to test +// the output of statement. This parameter has to be +// compiled but not evaluated by this macro, to ensure that +// this macro only accepts expressions that a macro such as +// EXPECT_DEATH would accept. +// terminator - Must be an empty statement for EXPECT_DEATH_IF_SUPPORTED +// and a return statement for ASSERT_DEATH_IF_SUPPORTED. +// This ensures that ASSERT_DEATH_IF_SUPPORTED will not +// compile inside functions where ASSERT_DEATH doesn't +// compile. +// +// The branch that has an always false condition is used to ensure that +// statement and regex are compiled (and thus syntactically correct) but +// never executed. The unreachable code macro protects the terminator +// statement from generating an 'unreachable code' warning in case +// statement unconditionally returns or throws. The Message constructor at +// the end allows the syntax of streaming additional messages into the +// macro, for compilational compatibility with EXPECT_DEATH/ASSERT_DEATH. +# define GTEST_UNSUPPORTED_DEATH_TEST(statement, regex, terminator) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::AlwaysTrue()) { \ + GTEST_LOG_(WARNING) \ + << "Death tests are not supported on this platform.\n" \ + << "Statement '" #statement "' cannot be verified."; \ + } else if (::testing::internal::AlwaysFalse()) { \ + ::testing::internal::RE::PartialMatch(".*", (regex)); \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + terminator; \ + } else \ + ::testing::Message() + +// EXPECT_DEATH_IF_SUPPORTED(statement, regex) and +// ASSERT_DEATH_IF_SUPPORTED(statement, regex) expand to real death tests if +// death tests are supported; otherwise they just issue a warning. This is +// useful when you are combining death test assertions with normal test +// assertions in one test. +#if GTEST_HAS_DEATH_TEST +# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \ + EXPECT_DEATH(statement, regex) +# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \ + ASSERT_DEATH(statement, regex) +#else +# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \ + GTEST_UNSUPPORTED_DEATH_TEST(statement, regex, ) +# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \ + GTEST_UNSUPPORTED_DEATH_TEST(statement, regex, return) +#endif + +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_ diff --git a/deps/googletest/include/gtest/gtest-matchers.h b/deps/googletest/include/gtest/gtest-matchers.h new file mode 100644 index 000000000..6e73ba143 --- /dev/null +++ b/deps/googletest/include/gtest/gtest-matchers.h @@ -0,0 +1,748 @@ +// Copyright 2007, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This file implements just enough of the matcher interface to allow +// EXPECT_DEATH and friends to accept a matcher argument. + +// IWYU pragma: private, include "testing/base/public/gunit.h" +// IWYU pragma: friend third_party/googletest/googlemock/.* +// IWYU pragma: friend third_party/googletest/googletest/.* + +#ifndef GTEST_INCLUDE_GTEST_GTEST_MATCHERS_H_ +#define GTEST_INCLUDE_GTEST_GTEST_MATCHERS_H_ + +#include +#include +#include + +#include "gtest/gtest-printers.h" +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-port.h" + +// MSVC warning C5046 is new as of VS2017 version 15.8. +#if defined(_MSC_VER) && _MSC_VER >= 1915 +#define GTEST_MAYBE_5046_ 5046 +#else +#define GTEST_MAYBE_5046_ +#endif + +GTEST_DISABLE_MSC_WARNINGS_PUSH_( + 4251 GTEST_MAYBE_5046_ /* class A needs to have dll-interface to be used by + clients of class B */ + /* Symbol involving type with internal linkage not defined */) + +namespace testing { + +// To implement a matcher Foo for type T, define: +// 1. a class FooMatcherImpl that implements the +// MatcherInterface interface, and +// 2. a factory function that creates a Matcher object from a +// FooMatcherImpl*. +// +// The two-level delegation design makes it possible to allow a user +// to write "v" instead of "Eq(v)" where a Matcher is expected, which +// is impossible if we pass matchers by pointers. It also eases +// ownership management as Matcher objects can now be copied like +// plain values. + +// MatchResultListener is an abstract class. Its << operator can be +// used by a matcher to explain why a value matches or doesn't match. +// +class MatchResultListener { + public: + // Creates a listener object with the given underlying ostream. The + // listener does not own the ostream, and does not dereference it + // in the constructor or destructor. + explicit MatchResultListener(::std::ostream* os) : stream_(os) {} + virtual ~MatchResultListener() = 0; // Makes this class abstract. + + // Streams x to the underlying ostream; does nothing if the ostream + // is NULL. + template + MatchResultListener& operator<<(const T& x) { + if (stream_ != nullptr) *stream_ << x; + return *this; + } + + // Returns the underlying ostream. + ::std::ostream* stream() { return stream_; } + + // Returns true iff the listener is interested in an explanation of + // the match result. A matcher's MatchAndExplain() method can use + // this information to avoid generating the explanation when no one + // intends to hear it. + bool IsInterested() const { return stream_ != nullptr; } + + private: + ::std::ostream* const stream_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(MatchResultListener); +}; + +inline MatchResultListener::~MatchResultListener() { +} + +// An instance of a subclass of this knows how to describe itself as a +// matcher. +class MatcherDescriberInterface { + public: + virtual ~MatcherDescriberInterface() {} + + // Describes this matcher to an ostream. The function should print + // a verb phrase that describes the property a value matching this + // matcher should have. The subject of the verb phrase is the value + // being matched. For example, the DescribeTo() method of the Gt(7) + // matcher prints "is greater than 7". + virtual void DescribeTo(::std::ostream* os) const = 0; + + // Describes the negation of this matcher to an ostream. For + // example, if the description of this matcher is "is greater than + // 7", the negated description could be "is not greater than 7". + // You are not required to override this when implementing + // MatcherInterface, but it is highly advised so that your matcher + // can produce good error messages. + virtual void DescribeNegationTo(::std::ostream* os) const { + *os << "not ("; + DescribeTo(os); + *os << ")"; + } +}; + +// The implementation of a matcher. +template +class MatcherInterface : public MatcherDescriberInterface { + public: + // Returns true iff the matcher matches x; also explains the match + // result to 'listener' if necessary (see the next paragraph), in + // the form of a non-restrictive relative clause ("which ...", + // "whose ...", etc) that describes x. For example, the + // MatchAndExplain() method of the Pointee(...) matcher should + // generate an explanation like "which points to ...". + // + // Implementations of MatchAndExplain() should add an explanation of + // the match result *if and only if* they can provide additional + // information that's not already present (or not obvious) in the + // print-out of x and the matcher's description. Whether the match + // succeeds is not a factor in deciding whether an explanation is + // needed, as sometimes the caller needs to print a failure message + // when the match succeeds (e.g. when the matcher is used inside + // Not()). + // + // For example, a "has at least 10 elements" matcher should explain + // what the actual element count is, regardless of the match result, + // as it is useful information to the reader; on the other hand, an + // "is empty" matcher probably only needs to explain what the actual + // size is when the match fails, as it's redundant to say that the + // size is 0 when the value is already known to be empty. + // + // You should override this method when defining a new matcher. + // + // It's the responsibility of the caller (Google Test) to guarantee + // that 'listener' is not NULL. This helps to simplify a matcher's + // implementation when it doesn't care about the performance, as it + // can talk to 'listener' without checking its validity first. + // However, in order to implement dummy listeners efficiently, + // listener->stream() may be NULL. + virtual bool MatchAndExplain(T x, MatchResultListener* listener) const = 0; + + // Inherits these methods from MatcherDescriberInterface: + // virtual void DescribeTo(::std::ostream* os) const = 0; + // virtual void DescribeNegationTo(::std::ostream* os) const; +}; + +namespace internal { + +// Converts a MatcherInterface to a MatcherInterface. +template +class MatcherInterfaceAdapter : public MatcherInterface { + public: + explicit MatcherInterfaceAdapter(const MatcherInterface* impl) + : impl_(impl) {} + ~MatcherInterfaceAdapter() override { delete impl_; } + + void DescribeTo(::std::ostream* os) const override { impl_->DescribeTo(os); } + + void DescribeNegationTo(::std::ostream* os) const override { + impl_->DescribeNegationTo(os); + } + + bool MatchAndExplain(const T& x, + MatchResultListener* listener) const override { + return impl_->MatchAndExplain(x, listener); + } + + private: + const MatcherInterface* const impl_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(MatcherInterfaceAdapter); +}; + +struct AnyEq { + template + bool operator()(const A& a, const B& b) const { return a == b; } +}; +struct AnyNe { + template + bool operator()(const A& a, const B& b) const { return a != b; } +}; +struct AnyLt { + template + bool operator()(const A& a, const B& b) const { return a < b; } +}; +struct AnyGt { + template + bool operator()(const A& a, const B& b) const { return a > b; } +}; +struct AnyLe { + template + bool operator()(const A& a, const B& b) const { return a <= b; } +}; +struct AnyGe { + template + bool operator()(const A& a, const B& b) const { return a >= b; } +}; + +// A match result listener that ignores the explanation. +class DummyMatchResultListener : public MatchResultListener { + public: + DummyMatchResultListener() : MatchResultListener(nullptr) {} + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(DummyMatchResultListener); +}; + +// A match result listener that forwards the explanation to a given +// ostream. The difference between this and MatchResultListener is +// that the former is concrete. +class StreamMatchResultListener : public MatchResultListener { + public: + explicit StreamMatchResultListener(::std::ostream* os) + : MatchResultListener(os) {} + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(StreamMatchResultListener); +}; + +// An internal class for implementing Matcher, which will derive +// from it. We put functionalities common to all Matcher +// specializations here to avoid code duplication. +template +class MatcherBase { + public: + // Returns true iff the matcher matches x; also explains the match + // result to 'listener'. + bool MatchAndExplain(const T& x, MatchResultListener* listener) const { + return impl_->MatchAndExplain(x, listener); + } + + // Returns true iff this matcher matches x. + bool Matches(const T& x) const { + DummyMatchResultListener dummy; + return MatchAndExplain(x, &dummy); + } + + // Describes this matcher to an ostream. + void DescribeTo(::std::ostream* os) const { impl_->DescribeTo(os); } + + // Describes the negation of this matcher to an ostream. + void DescribeNegationTo(::std::ostream* os) const { + impl_->DescribeNegationTo(os); + } + + // Explains why x matches, or doesn't match, the matcher. + void ExplainMatchResultTo(const T& x, ::std::ostream* os) const { + StreamMatchResultListener listener(os); + MatchAndExplain(x, &listener); + } + + // Returns the describer for this matcher object; retains ownership + // of the describer, which is only guaranteed to be alive when + // this matcher object is alive. + const MatcherDescriberInterface* GetDescriber() const { + return impl_.get(); + } + + protected: + MatcherBase() {} + + // Constructs a matcher from its implementation. + explicit MatcherBase(const MatcherInterface* impl) : impl_(impl) {} + + template + explicit MatcherBase( + const MatcherInterface* impl, + typename internal::EnableIf< + !internal::IsSame::value>::type* = nullptr) + : impl_(new internal::MatcherInterfaceAdapter(impl)) {} + + MatcherBase(const MatcherBase&) = default; + MatcherBase& operator=(const MatcherBase&) = default; + MatcherBase(MatcherBase&&) = default; + MatcherBase& operator=(MatcherBase&&) = default; + + virtual ~MatcherBase() {} + + private: + std::shared_ptr> impl_; +}; + +} // namespace internal + +// A Matcher is a copyable and IMMUTABLE (except by assignment) +// object that can check whether a value of type T matches. The +// implementation of Matcher is just a std::shared_ptr to const +// MatcherInterface. Don't inherit from Matcher! +template +class Matcher : public internal::MatcherBase { + public: + // Constructs a null matcher. Needed for storing Matcher objects in STL + // containers. A default-constructed matcher is not yet initialized. You + // cannot use it until a valid value has been assigned to it. + explicit Matcher() {} // NOLINT + + // Constructs a matcher from its implementation. + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + + template + explicit Matcher(const MatcherInterface* impl, + typename internal::EnableIf< + !internal::IsSame::value>::type* = nullptr) + : internal::MatcherBase(impl) {} + + // Implicit constructor here allows people to write + // EXPECT_CALL(foo, Bar(5)) instead of EXPECT_CALL(foo, Bar(Eq(5))) sometimes + Matcher(T value); // NOLINT +}; + +// The following two specializations allow the user to write str +// instead of Eq(str) and "foo" instead of Eq("foo") when a std::string +// matcher is expected. +template <> +class GTEST_API_ Matcher + : public internal::MatcherBase { + public: + Matcher() {} + + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + + // Allows the user to write str instead of Eq(str) sometimes, where + // str is a std::string object. + Matcher(const std::string& s); // NOLINT + + // Allows the user to write "foo" instead of Eq("foo") sometimes. + Matcher(const char* s); // NOLINT +}; + +template <> +class GTEST_API_ Matcher + : public internal::MatcherBase { + public: + Matcher() {} + + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + + // Allows the user to write str instead of Eq(str) sometimes, where + // str is a string object. + Matcher(const std::string& s); // NOLINT + + // Allows the user to write "foo" instead of Eq("foo") sometimes. + Matcher(const char* s); // NOLINT +}; + +#if GTEST_HAS_ABSL +// The following two specializations allow the user to write str +// instead of Eq(str) and "foo" instead of Eq("foo") when a absl::string_view +// matcher is expected. +template <> +class GTEST_API_ Matcher + : public internal::MatcherBase { + public: + Matcher() {} + + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + + // Allows the user to write str instead of Eq(str) sometimes, where + // str is a std::string object. + Matcher(const std::string& s); // NOLINT + + // Allows the user to write "foo" instead of Eq("foo") sometimes. + Matcher(const char* s); // NOLINT + + // Allows the user to pass absl::string_views directly. + Matcher(absl::string_view s); // NOLINT +}; + +template <> +class GTEST_API_ Matcher + : public internal::MatcherBase { + public: + Matcher() {} + + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + explicit Matcher(const MatcherInterface* impl) + : internal::MatcherBase(impl) {} + + // Allows the user to write str instead of Eq(str) sometimes, where + // str is a std::string object. + Matcher(const std::string& s); // NOLINT + + // Allows the user to write "foo" instead of Eq("foo") sometimes. + Matcher(const char* s); // NOLINT + + // Allows the user to pass absl::string_views directly. + Matcher(absl::string_view s); // NOLINT +}; +#endif // GTEST_HAS_ABSL + +// Prints a matcher in a human-readable format. +template +std::ostream& operator<<(std::ostream& os, const Matcher& matcher) { + matcher.DescribeTo(&os); + return os; +} + +// The PolymorphicMatcher class template makes it easy to implement a +// polymorphic matcher (i.e. a matcher that can match values of more +// than one type, e.g. Eq(n) and NotNull()). +// +// To define a polymorphic matcher, a user should provide an Impl +// class that has a DescribeTo() method and a DescribeNegationTo() +// method, and define a member function (or member function template) +// +// bool MatchAndExplain(const Value& value, +// MatchResultListener* listener) const; +// +// See the definition of NotNull() for a complete example. +template +class PolymorphicMatcher { + public: + explicit PolymorphicMatcher(const Impl& an_impl) : impl_(an_impl) {} + + // Returns a mutable reference to the underlying matcher + // implementation object. + Impl& mutable_impl() { return impl_; } + + // Returns an immutable reference to the underlying matcher + // implementation object. + const Impl& impl() const { return impl_; } + + template + operator Matcher() const { + return Matcher(new MonomorphicImpl(impl_)); + } + + private: + template + class MonomorphicImpl : public MatcherInterface { + public: + explicit MonomorphicImpl(const Impl& impl) : impl_(impl) {} + + virtual void DescribeTo(::std::ostream* os) const { impl_.DescribeTo(os); } + + virtual void DescribeNegationTo(::std::ostream* os) const { + impl_.DescribeNegationTo(os); + } + + virtual bool MatchAndExplain(T x, MatchResultListener* listener) const { + return impl_.MatchAndExplain(x, listener); + } + + private: + const Impl impl_; + }; + + Impl impl_; +}; + +// Creates a matcher from its implementation. +// DEPRECATED: Especially in the generic code, prefer: +// Matcher(new MyMatcherImpl(...)); +// +// MakeMatcher may create a Matcher that accepts its argument by value, which +// leads to unnecessary copies & lack of support for non-copyable types. +template +inline Matcher MakeMatcher(const MatcherInterface* impl) { + return Matcher(impl); +} + +// Creates a polymorphic matcher from its implementation. This is +// easier to use than the PolymorphicMatcher constructor as it +// doesn't require you to explicitly write the template argument, e.g. +// +// MakePolymorphicMatcher(foo); +// vs +// PolymorphicMatcher(foo); +template +inline PolymorphicMatcher MakePolymorphicMatcher(const Impl& impl) { + return PolymorphicMatcher(impl); +} + +namespace internal { +// Implements a matcher that compares a given value with a +// pre-supplied value using one of the ==, <=, <, etc, operators. The +// two values being compared don't have to have the same type. +// +// The matcher defined here is polymorphic (for example, Eq(5) can be +// used to match an int, a short, a double, etc). Therefore we use +// a template type conversion operator in the implementation. +// +// The following template definition assumes that the Rhs parameter is +// a "bare" type (i.e. neither 'const T' nor 'T&'). +template +class ComparisonBase { + public: + explicit ComparisonBase(const Rhs& rhs) : rhs_(rhs) {} + template + operator Matcher() const { + return Matcher(new Impl(rhs_)); + } + + private: + template + static const T& Unwrap(const T& v) { return v; } + template + static const T& Unwrap(std::reference_wrapper v) { return v; } + + template + class Impl : public MatcherInterface { + public: + explicit Impl(const Rhs& rhs) : rhs_(rhs) {} + bool MatchAndExplain(Lhs lhs, + MatchResultListener* /* listener */) const override { + return Op()(lhs, Unwrap(rhs_)); + } + void DescribeTo(::std::ostream* os) const override { + *os << D::Desc() << " "; + UniversalPrint(Unwrap(rhs_), os); + } + void DescribeNegationTo(::std::ostream* os) const override { + *os << D::NegatedDesc() << " "; + UniversalPrint(Unwrap(rhs_), os); + } + + private: + Rhs rhs_; + }; + Rhs rhs_; +}; + +template +class EqMatcher : public ComparisonBase, Rhs, AnyEq> { + public: + explicit EqMatcher(const Rhs& rhs) + : ComparisonBase, Rhs, AnyEq>(rhs) { } + static const char* Desc() { return "is equal to"; } + static const char* NegatedDesc() { return "isn't equal to"; } +}; +template +class NeMatcher : public ComparisonBase, Rhs, AnyNe> { + public: + explicit NeMatcher(const Rhs& rhs) + : ComparisonBase, Rhs, AnyNe>(rhs) { } + static const char* Desc() { return "isn't equal to"; } + static const char* NegatedDesc() { return "is equal to"; } +}; +template +class LtMatcher : public ComparisonBase, Rhs, AnyLt> { + public: + explicit LtMatcher(const Rhs& rhs) + : ComparisonBase, Rhs, AnyLt>(rhs) { } + static const char* Desc() { return "is <"; } + static const char* NegatedDesc() { return "isn't <"; } +}; +template +class GtMatcher : public ComparisonBase, Rhs, AnyGt> { + public: + explicit GtMatcher(const Rhs& rhs) + : ComparisonBase, Rhs, AnyGt>(rhs) { } + static const char* Desc() { return "is >"; } + static const char* NegatedDesc() { return "isn't >"; } +}; +template +class LeMatcher : public ComparisonBase, Rhs, AnyLe> { + public: + explicit LeMatcher(const Rhs& rhs) + : ComparisonBase, Rhs, AnyLe>(rhs) { } + static const char* Desc() { return "is <="; } + static const char* NegatedDesc() { return "isn't <="; } +}; +template +class GeMatcher : public ComparisonBase, Rhs, AnyGe> { + public: + explicit GeMatcher(const Rhs& rhs) + : ComparisonBase, Rhs, AnyGe>(rhs) { } + static const char* Desc() { return "is >="; } + static const char* NegatedDesc() { return "isn't >="; } +}; + +// Implements polymorphic matchers MatchesRegex(regex) and +// ContainsRegex(regex), which can be used as a Matcher as long as +// T can be converted to a string. +class MatchesRegexMatcher { + public: + MatchesRegexMatcher(const RE* regex, bool full_match) + : regex_(regex), full_match_(full_match) {} + +#if GTEST_HAS_ABSL + bool MatchAndExplain(const absl::string_view& s, + MatchResultListener* listener) const { + return MatchAndExplain(std::string(s), listener); + } +#endif // GTEST_HAS_ABSL + + // Accepts pointer types, particularly: + // const char* + // char* + // const wchar_t* + // wchar_t* + template + bool MatchAndExplain(CharType* s, MatchResultListener* listener) const { + return s != nullptr && MatchAndExplain(std::string(s), listener); + } + + // Matches anything that can convert to std::string. + // + // This is a template, not just a plain function with const std::string&, + // because absl::string_view has some interfering non-explicit constructors. + template + bool MatchAndExplain(const MatcheeStringType& s, + MatchResultListener* /* listener */) const { + const std::string& s2(s); + return full_match_ ? RE::FullMatch(s2, *regex_) + : RE::PartialMatch(s2, *regex_); + } + + void DescribeTo(::std::ostream* os) const { + *os << (full_match_ ? "matches" : "contains") << " regular expression "; + UniversalPrinter::Print(regex_->pattern(), os); + } + + void DescribeNegationTo(::std::ostream* os) const { + *os << "doesn't " << (full_match_ ? "match" : "contain") + << " regular expression "; + UniversalPrinter::Print(regex_->pattern(), os); + } + + private: + const std::shared_ptr regex_; + const bool full_match_; +}; +} // namespace internal + +// Matches a string that fully matches regular expression 'regex'. +// The matcher takes ownership of 'regex'. +inline PolymorphicMatcher MatchesRegex( + const internal::RE* regex) { + return MakePolymorphicMatcher(internal::MatchesRegexMatcher(regex, true)); +} +inline PolymorphicMatcher MatchesRegex( + const std::string& regex) { + return MatchesRegex(new internal::RE(regex)); +} + +// Matches a string that contains regular expression 'regex'. +// The matcher takes ownership of 'regex'. +inline PolymorphicMatcher ContainsRegex( + const internal::RE* regex) { + return MakePolymorphicMatcher(internal::MatchesRegexMatcher(regex, false)); +} +inline PolymorphicMatcher ContainsRegex( + const std::string& regex) { + return ContainsRegex(new internal::RE(regex)); +} + +// Creates a polymorphic matcher that matches anything equal to x. +// Note: if the parameter of Eq() were declared as const T&, Eq("foo") +// wouldn't compile. +template +inline internal::EqMatcher Eq(T x) { return internal::EqMatcher(x); } + +// Constructs a Matcher from a 'value' of type T. The constructed +// matcher matches any value that's equal to 'value'. +template +Matcher::Matcher(T value) { *this = Eq(value); } + +// Creates a monomorphic matcher that matches anything with type Lhs +// and equal to rhs. A user may need to use this instead of Eq(...) +// in order to resolve an overloading ambiguity. +// +// TypedEq(x) is just a convenient short-hand for Matcher(Eq(x)) +// or Matcher(x), but more readable than the latter. +// +// We could define similar monomorphic matchers for other comparison +// operations (e.g. TypedLt, TypedGe, and etc), but decided not to do +// it yet as those are used much less than Eq() in practice. A user +// can always write Matcher(Lt(5)) to be explicit about the type, +// for example. +template +inline Matcher TypedEq(const Rhs& rhs) { return Eq(rhs); } + +// Creates a polymorphic matcher that matches anything >= x. +template +inline internal::GeMatcher Ge(Rhs x) { + return internal::GeMatcher(x); +} + +// Creates a polymorphic matcher that matches anything > x. +template +inline internal::GtMatcher Gt(Rhs x) { + return internal::GtMatcher(x); +} + +// Creates a polymorphic matcher that matches anything <= x. +template +inline internal::LeMatcher Le(Rhs x) { + return internal::LeMatcher(x); +} + +// Creates a polymorphic matcher that matches anything < x. +template +inline internal::LtMatcher Lt(Rhs x) { + return internal::LtMatcher(x); +} + +// Creates a polymorphic matcher that matches anything != x. +template +inline internal::NeMatcher Ne(Rhs x) { + return internal::NeMatcher(x); +} +} // namespace testing + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 5046 + +#endif // GTEST_INCLUDE_GTEST_GTEST_MATCHERS_H_ diff --git a/deps/googletest/include/gtest/gtest-message.h b/deps/googletest/include/gtest/gtest-message.h new file mode 100644 index 000000000..4a80e11e6 --- /dev/null +++ b/deps/googletest/include/gtest/gtest-message.h @@ -0,0 +1,218 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file defines the Message class. +// +// IMPORTANT NOTE: Due to limitation of the C++ language, we have to +// leave some internal implementation details in this header file. +// They are clearly marked by comments like this: +// +// // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +// +// Such code is NOT meant to be used by a user directly, and is subject +// to CHANGE WITHOUT NOTICE. Therefore DO NOT DEPEND ON IT in a user +// program! + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_ +#define GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_ + +#include +#include + +#include "gtest/internal/gtest-port.h" + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +// Ensures that there is at least one operator<< in the global namespace. +// See Message& operator<<(...) below for why. +void operator<<(const testing::internal::Secret&, int); + +namespace testing { + +// The Message class works like an ostream repeater. +// +// Typical usage: +// +// 1. You stream a bunch of values to a Message object. +// It will remember the text in a stringstream. +// 2. Then you stream the Message object to an ostream. +// This causes the text in the Message to be streamed +// to the ostream. +// +// For example; +// +// testing::Message foo; +// foo << 1 << " != " << 2; +// std::cout << foo; +// +// will print "1 != 2". +// +// Message is not intended to be inherited from. In particular, its +// destructor is not virtual. +// +// Note that stringstream behaves differently in gcc and in MSVC. You +// can stream a NULL char pointer to it in the former, but not in the +// latter (it causes an access violation if you do). The Message +// class hides this difference by treating a NULL char pointer as +// "(null)". +class GTEST_API_ Message { + private: + // The type of basic IO manipulators (endl, ends, and flush) for + // narrow streams. + typedef std::ostream& (*BasicNarrowIoManip)(std::ostream&); + + public: + // Constructs an empty Message. + Message(); + + // Copy constructor. + Message(const Message& msg) : ss_(new ::std::stringstream) { // NOLINT + *ss_ << msg.GetString(); + } + + // Constructs a Message from a C-string. + explicit Message(const char* str) : ss_(new ::std::stringstream) { + *ss_ << str; + } + + // Streams a non-pointer value to this object. + template + inline Message& operator <<(const T& val) { + // Some libraries overload << for STL containers. These + // overloads are defined in the global namespace instead of ::std. + // + // C++'s symbol lookup rule (i.e. Koenig lookup) says that these + // overloads are visible in either the std namespace or the global + // namespace, but not other namespaces, including the testing + // namespace which Google Test's Message class is in. + // + // To allow STL containers (and other types that has a << operator + // defined in the global namespace) to be used in Google Test + // assertions, testing::Message must access the custom << operator + // from the global namespace. With this using declaration, + // overloads of << defined in the global namespace and those + // visible via Koenig lookup are both exposed in this function. + using ::operator <<; + *ss_ << val; + return *this; + } + + // Streams a pointer value to this object. + // + // This function is an overload of the previous one. When you + // stream a pointer to a Message, this definition will be used as it + // is more specialized. (The C++ Standard, section + // [temp.func.order].) If you stream a non-pointer, then the + // previous definition will be used. + // + // The reason for this overload is that streaming a NULL pointer to + // ostream is undefined behavior. Depending on the compiler, you + // may get "0", "(nil)", "(null)", or an access violation. To + // ensure consistent result across compilers, we always treat NULL + // as "(null)". + template + inline Message& operator <<(T* const& pointer) { // NOLINT + if (pointer == nullptr) { + *ss_ << "(null)"; + } else { + *ss_ << pointer; + } + return *this; + } + + // Since the basic IO manipulators are overloaded for both narrow + // and wide streams, we have to provide this specialized definition + // of operator <<, even though its body is the same as the + // templatized version above. Without this definition, streaming + // endl or other basic IO manipulators to Message will confuse the + // compiler. + Message& operator <<(BasicNarrowIoManip val) { + *ss_ << val; + return *this; + } + + // Instead of 1/0, we want to see true/false for bool values. + Message& operator <<(bool b) { + return *this << (b ? "true" : "false"); + } + + // These two overloads allow streaming a wide C string to a Message + // using the UTF-8 encoding. + Message& operator <<(const wchar_t* wide_c_str); + Message& operator <<(wchar_t* wide_c_str); + +#if GTEST_HAS_STD_WSTRING + // Converts the given wide string to a narrow string using the UTF-8 + // encoding, and streams the result to this Message object. + Message& operator <<(const ::std::wstring& wstr); +#endif // GTEST_HAS_STD_WSTRING + + // Gets the text streamed to this object so far as an std::string. + // Each '\0' character in the buffer is replaced with "\\0". + // + // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. + std::string GetString() const; + + private: + // We'll hold the text streamed to this object here. + const std::unique_ptr< ::std::stringstream> ss_; + + // We declare (but don't implement) this to prevent the compiler + // from implementing the assignment operator. + void operator=(const Message&); +}; + +// Streams a Message to an ostream. +inline std::ostream& operator <<(std::ostream& os, const Message& sb) { + return os << sb.GetString(); +} + +namespace internal { + +// Converts a streamable value to an std::string. A NULL pointer is +// converted to "(null)". When the input value is a ::string, +// ::std::string, ::wstring, or ::std::wstring object, each NUL +// character in it is replaced with "\\0". +template +std::string StreamableToString(const T& streamable) { + return (Message() << streamable).GetString(); +} + +} // namespace internal +} // namespace testing + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +#endif // GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_ diff --git a/deps/googletest/include/gtest/gtest-param-test.h b/deps/googletest/include/gtest/gtest-param-test.h new file mode 100644 index 000000000..d7c9dd8c7 --- /dev/null +++ b/deps/googletest/include/gtest/gtest-param-test.h @@ -0,0 +1,503 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Macros and functions for implementing parameterized tests +// in Google C++ Testing and Mocking Framework (Google Test) +// +// This file is generated by a SCRIPT. DO NOT EDIT BY HAND! +// +// GOOGLETEST_CM0001 DO NOT DELETE +#ifndef GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_ +#define GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_ + + +// Value-parameterized tests allow you to test your code with different +// parameters without writing multiple copies of the same test. +// +// Here is how you use value-parameterized tests: + +#if 0 + +// To write value-parameterized tests, first you should define a fixture +// class. It is usually derived from testing::TestWithParam (see below for +// another inheritance scheme that's sometimes useful in more complicated +// class hierarchies), where the type of your parameter values. +// TestWithParam is itself derived from testing::Test. T can be any +// copyable type. If it's a raw pointer, you are responsible for managing the +// lifespan of the pointed values. + +class FooTest : public ::testing::TestWithParam { + // You can implement all the usual class fixture members here. +}; + +// Then, use the TEST_P macro to define as many parameterized tests +// for this fixture as you want. The _P suffix is for "parameterized" +// or "pattern", whichever you prefer to think. + +TEST_P(FooTest, DoesBlah) { + // Inside a test, access the test parameter with the GetParam() method + // of the TestWithParam class: + EXPECT_TRUE(foo.Blah(GetParam())); + ... +} + +TEST_P(FooTest, HasBlahBlah) { + ... +} + +// Finally, you can use INSTANTIATE_TEST_SUITE_P to instantiate the test +// case with any set of parameters you want. Google Test defines a number +// of functions for generating test parameters. They return what we call +// (surprise!) parameter generators. Here is a summary of them, which +// are all in the testing namespace: +// +// +// Range(begin, end [, step]) - Yields values {begin, begin+step, +// begin+step+step, ...}. The values do not +// include end. step defaults to 1. +// Values(v1, v2, ..., vN) - Yields values {v1, v2, ..., vN}. +// ValuesIn(container) - Yields values from a C-style array, an STL +// ValuesIn(begin,end) container, or an iterator range [begin, end). +// Bool() - Yields sequence {false, true}. +// Combine(g1, g2, ..., gN) - Yields all combinations (the Cartesian product +// for the math savvy) of the values generated +// by the N generators. +// +// For more details, see comments at the definitions of these functions below +// in this file. +// +// The following statement will instantiate tests from the FooTest test suite +// each with parameter values "meeny", "miny", and "moe". + +INSTANTIATE_TEST_SUITE_P(InstantiationName, + FooTest, + Values("meeny", "miny", "moe")); + +// To distinguish different instances of the pattern, (yes, you +// can instantiate it more than once) the first argument to the +// INSTANTIATE_TEST_SUITE_P macro is a prefix that will be added to the +// actual test suite name. Remember to pick unique prefixes for different +// instantiations. The tests from the instantiation above will have +// these names: +// +// * InstantiationName/FooTest.DoesBlah/0 for "meeny" +// * InstantiationName/FooTest.DoesBlah/1 for "miny" +// * InstantiationName/FooTest.DoesBlah/2 for "moe" +// * InstantiationName/FooTest.HasBlahBlah/0 for "meeny" +// * InstantiationName/FooTest.HasBlahBlah/1 for "miny" +// * InstantiationName/FooTest.HasBlahBlah/2 for "moe" +// +// You can use these names in --gtest_filter. +// +// This statement will instantiate all tests from FooTest again, each +// with parameter values "cat" and "dog": + +const char* pets[] = {"cat", "dog"}; +INSTANTIATE_TEST_SUITE_P(AnotherInstantiationName, FooTest, ValuesIn(pets)); + +// The tests from the instantiation above will have these names: +// +// * AnotherInstantiationName/FooTest.DoesBlah/0 for "cat" +// * AnotherInstantiationName/FooTest.DoesBlah/1 for "dog" +// * AnotherInstantiationName/FooTest.HasBlahBlah/0 for "cat" +// * AnotherInstantiationName/FooTest.HasBlahBlah/1 for "dog" +// +// Please note that INSTANTIATE_TEST_SUITE_P will instantiate all tests +// in the given test suite, whether their definitions come before or +// AFTER the INSTANTIATE_TEST_SUITE_P statement. +// +// Please also note that generator expressions (including parameters to the +// generators) are evaluated in InitGoogleTest(), after main() has started. +// This allows the user on one hand, to adjust generator parameters in order +// to dynamically determine a set of tests to run and on the other hand, +// give the user a chance to inspect the generated tests with Google Test +// reflection API before RUN_ALL_TESTS() is executed. +// +// You can see samples/sample7_unittest.cc and samples/sample8_unittest.cc +// for more examples. +// +// In the future, we plan to publish the API for defining new parameter +// generators. But for now this interface remains part of the internal +// implementation and is subject to change. +// +// +// A parameterized test fixture must be derived from testing::Test and from +// testing::WithParamInterface, where T is the type of the parameter +// values. Inheriting from TestWithParam satisfies that requirement because +// TestWithParam inherits from both Test and WithParamInterface. In more +// complicated hierarchies, however, it is occasionally useful to inherit +// separately from Test and WithParamInterface. For example: + +class BaseTest : public ::testing::Test { + // You can inherit all the usual members for a non-parameterized test + // fixture here. +}; + +class DerivedTest : public BaseTest, public ::testing::WithParamInterface { + // The usual test fixture members go here too. +}; + +TEST_F(BaseTest, HasFoo) { + // This is an ordinary non-parameterized test. +} + +TEST_P(DerivedTest, DoesBlah) { + // GetParam works just the same here as if you inherit from TestWithParam. + EXPECT_TRUE(foo.Blah(GetParam())); +} + +#endif // 0 + +#include + +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-param-util.h" +#include "gtest/internal/gtest-port.h" + +namespace testing { + +// Functions producing parameter generators. +// +// Google Test uses these generators to produce parameters for value- +// parameterized tests. When a parameterized test suite is instantiated +// with a particular generator, Google Test creates and runs tests +// for each element in the sequence produced by the generator. +// +// In the following sample, tests from test suite FooTest are instantiated +// each three times with parameter values 3, 5, and 8: +// +// class FooTest : public TestWithParam { ... }; +// +// TEST_P(FooTest, TestThis) { +// } +// TEST_P(FooTest, TestThat) { +// } +// INSTANTIATE_TEST_SUITE_P(TestSequence, FooTest, Values(3, 5, 8)); +// + +// Range() returns generators providing sequences of values in a range. +// +// Synopsis: +// Range(start, end) +// - returns a generator producing a sequence of values {start, start+1, +// start+2, ..., }. +// Range(start, end, step) +// - returns a generator producing a sequence of values {start, start+step, +// start+step+step, ..., }. +// Notes: +// * The generated sequences never include end. For example, Range(1, 5) +// returns a generator producing a sequence {1, 2, 3, 4}. Range(1, 9, 2) +// returns a generator producing {1, 3, 5, 7}. +// * start and end must have the same type. That type may be any integral or +// floating-point type or a user defined type satisfying these conditions: +// * It must be assignable (have operator=() defined). +// * It must have operator+() (operator+(int-compatible type) for +// two-operand version). +// * It must have operator<() defined. +// Elements in the resulting sequences will also have that type. +// * Condition start < end must be satisfied in order for resulting sequences +// to contain any elements. +// +template +internal::ParamGenerator Range(T start, T end, IncrementT step) { + return internal::ParamGenerator( + new internal::RangeGenerator(start, end, step)); +} + +template +internal::ParamGenerator Range(T start, T end) { + return Range(start, end, 1); +} + +// ValuesIn() function allows generation of tests with parameters coming from +// a container. +// +// Synopsis: +// ValuesIn(const T (&array)[N]) +// - returns a generator producing sequences with elements from +// a C-style array. +// ValuesIn(const Container& container) +// - returns a generator producing sequences with elements from +// an STL-style container. +// ValuesIn(Iterator begin, Iterator end) +// - returns a generator producing sequences with elements from +// a range [begin, end) defined by a pair of STL-style iterators. These +// iterators can also be plain C pointers. +// +// Please note that ValuesIn copies the values from the containers +// passed in and keeps them to generate tests in RUN_ALL_TESTS(). +// +// Examples: +// +// This instantiates tests from test suite StringTest +// each with C-string values of "foo", "bar", and "baz": +// +// const char* strings[] = {"foo", "bar", "baz"}; +// INSTANTIATE_TEST_SUITE_P(StringSequence, StringTest, ValuesIn(strings)); +// +// This instantiates tests from test suite StlStringTest +// each with STL strings with values "a" and "b": +// +// ::std::vector< ::std::string> GetParameterStrings() { +// ::std::vector< ::std::string> v; +// v.push_back("a"); +// v.push_back("b"); +// return v; +// } +// +// INSTANTIATE_TEST_SUITE_P(CharSequence, +// StlStringTest, +// ValuesIn(GetParameterStrings())); +// +// +// This will also instantiate tests from CharTest +// each with parameter values 'a' and 'b': +// +// ::std::list GetParameterChars() { +// ::std::list list; +// list.push_back('a'); +// list.push_back('b'); +// return list; +// } +// ::std::list l = GetParameterChars(); +// INSTANTIATE_TEST_SUITE_P(CharSequence2, +// CharTest, +// ValuesIn(l.begin(), l.end())); +// +template +internal::ParamGenerator< + typename ::testing::internal::IteratorTraits::value_type> +ValuesIn(ForwardIterator begin, ForwardIterator end) { + typedef typename ::testing::internal::IteratorTraits + ::value_type ParamType; + return internal::ParamGenerator( + new internal::ValuesInIteratorRangeGenerator(begin, end)); +} + +template +internal::ParamGenerator ValuesIn(const T (&array)[N]) { + return ValuesIn(array, array + N); +} + +template +internal::ParamGenerator ValuesIn( + const Container& container) { + return ValuesIn(container.begin(), container.end()); +} + +// Values() allows generating tests from explicitly specified list of +// parameters. +// +// Synopsis: +// Values(T v1, T v2, ..., T vN) +// - returns a generator producing sequences with elements v1, v2, ..., vN. +// +// For example, this instantiates tests from test suite BarTest each +// with values "one", "two", and "three": +// +// INSTANTIATE_TEST_SUITE_P(NumSequence, +// BarTest, +// Values("one", "two", "three")); +// +// This instantiates tests from test suite BazTest each with values 1, 2, 3.5. +// The exact type of values will depend on the type of parameter in BazTest. +// +// INSTANTIATE_TEST_SUITE_P(FloatingNumbers, BazTest, Values(1, 2, 3.5)); +// +// +template +internal::ValueArray Values(T... v) { + return internal::ValueArray(std::move(v)...); +} + +// Bool() allows generating tests with parameters in a set of (false, true). +// +// Synopsis: +// Bool() +// - returns a generator producing sequences with elements {false, true}. +// +// It is useful when testing code that depends on Boolean flags. Combinations +// of multiple flags can be tested when several Bool()'s are combined using +// Combine() function. +// +// In the following example all tests in the test suite FlagDependentTest +// will be instantiated twice with parameters false and true. +// +// class FlagDependentTest : public testing::TestWithParam { +// virtual void SetUp() { +// external_flag = GetParam(); +// } +// } +// INSTANTIATE_TEST_SUITE_P(BoolSequence, FlagDependentTest, Bool()); +// +inline internal::ParamGenerator Bool() { + return Values(false, true); +} + +// Combine() allows the user to combine two or more sequences to produce +// values of a Cartesian product of those sequences' elements. +// +// Synopsis: +// Combine(gen1, gen2, ..., genN) +// - returns a generator producing sequences with elements coming from +// the Cartesian product of elements from the sequences generated by +// gen1, gen2, ..., genN. The sequence elements will have a type of +// std::tuple where T1, T2, ..., TN are the types +// of elements from sequences produces by gen1, gen2, ..., genN. +// +// Combine can have up to 10 arguments. +// +// Example: +// +// This will instantiate tests in test suite AnimalTest each one with +// the parameter values tuple("cat", BLACK), tuple("cat", WHITE), +// tuple("dog", BLACK), and tuple("dog", WHITE): +// +// enum Color { BLACK, GRAY, WHITE }; +// class AnimalTest +// : public testing::TestWithParam > {...}; +// +// TEST_P(AnimalTest, AnimalLooksNice) {...} +// +// INSTANTIATE_TEST_SUITE_P(AnimalVariations, AnimalTest, +// Combine(Values("cat", "dog"), +// Values(BLACK, WHITE))); +// +// This will instantiate tests in FlagDependentTest with all variations of two +// Boolean flags: +// +// class FlagDependentTest +// : public testing::TestWithParam > { +// virtual void SetUp() { +// // Assigns external_flag_1 and external_flag_2 values from the tuple. +// std::tie(external_flag_1, external_flag_2) = GetParam(); +// } +// }; +// +// TEST_P(FlagDependentTest, TestFeature1) { +// // Test your code using external_flag_1 and external_flag_2 here. +// } +// INSTANTIATE_TEST_SUITE_P(TwoBoolSequence, FlagDependentTest, +// Combine(Bool(), Bool())); +// +template +internal::CartesianProductHolder Combine(const Generator&... g) { + return internal::CartesianProductHolder(g...); +} + +#define TEST_P(test_suite_name, test_name) \ + class GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) \ + : public test_suite_name { \ + public: \ + GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)() {} \ + virtual void TestBody(); \ + \ + private: \ + static int AddToRegistry() { \ + ::testing::UnitTest::GetInstance() \ + ->parameterized_test_registry() \ + .GetTestSuitePatternHolder( \ + #test_suite_name, \ + ::testing::internal::CodeLocation(__FILE__, __LINE__)) \ + ->AddTestPattern( \ + GTEST_STRINGIFY_(test_suite_name), GTEST_STRINGIFY_(test_name), \ + new ::testing::internal::TestMetaFactory()); \ + return 0; \ + } \ + static int gtest_registering_dummy_ GTEST_ATTRIBUTE_UNUSED_; \ + GTEST_DISALLOW_COPY_AND_ASSIGN_(GTEST_TEST_CLASS_NAME_(test_suite_name, \ + test_name)); \ + }; \ + int GTEST_TEST_CLASS_NAME_(test_suite_name, \ + test_name)::gtest_registering_dummy_ = \ + GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)::AddToRegistry(); \ + void GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)::TestBody() + +// The last argument to INSTANTIATE_TEST_SUITE_P allows the user to specify +// generator and an optional function or functor that generates custom test name +// suffixes based on the test parameters. Such a function or functor should +// accept one argument of type testing::TestParamInfo, and +// return std::string. +// +// testing::PrintToStringParamName is a builtin test suffix generator that +// returns the value of testing::PrintToString(GetParam()). +// +// Note: test names must be non-empty, unique, and may only contain ASCII +// alphanumeric characters or underscore. Because PrintToString adds quotes +// to std::string and C strings, it won't work for these types. + +#define GTEST_EXPAND_(arg) arg +#define GTEST_GET_FIRST_(first, ...) first +#define GTEST_GET_SECOND_(first, second, ...) second + +#define INSTANTIATE_TEST_SUITE_P(prefix, test_suite_name, ...) \ + static ::testing::internal::ParamGenerator \ + gtest_##prefix##test_suite_name##_EvalGenerator_() { \ + return GTEST_EXPAND_(GTEST_GET_FIRST_(__VA_ARGS__, DUMMY_PARAM_)); \ + } \ + static ::std::string gtest_##prefix##test_suite_name##_EvalGenerateName_( \ + const ::testing::TestParamInfo& info) { \ + if (::testing::internal::AlwaysFalse()) { \ + ::testing::internal::TestNotEmpty(GTEST_EXPAND_(GTEST_GET_SECOND_( \ + __VA_ARGS__, \ + ::testing::internal::DefaultParamName, \ + DUMMY_PARAM_))); \ + auto t = std::make_tuple(__VA_ARGS__); \ + static_assert(std::tuple_size::value <= 2, \ + "Too Many Args!"); \ + } \ + return ((GTEST_EXPAND_(GTEST_GET_SECOND_( \ + __VA_ARGS__, \ + ::testing::internal::DefaultParamName, \ + DUMMY_PARAM_))))(info); \ + } \ + static int gtest_##prefix##test_suite_name##_dummy_ \ + GTEST_ATTRIBUTE_UNUSED_ = \ + ::testing::UnitTest::GetInstance() \ + ->parameterized_test_registry() \ + .GetTestSuitePatternHolder( \ + #test_suite_name, \ + ::testing::internal::CodeLocation(__FILE__, __LINE__)) \ + ->AddTestSuiteInstantiation( \ + #prefix, >est_##prefix##test_suite_name##_EvalGenerator_, \ + >est_##prefix##test_suite_name##_EvalGenerateName_, \ + __FILE__, __LINE__) + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +#define INSTANTIATE_TEST_CASE_P \ + static_assert(::testing::internal::InstantiateTestCase_P_IsDeprecated(), \ + ""); \ + INSTANTIATE_TEST_SUITE_P +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_ diff --git a/deps/googletest/include/gtest/gtest-printers.h b/deps/googletest/include/gtest/gtest-printers.h new file mode 100644 index 000000000..97cdb2a20 --- /dev/null +++ b/deps/googletest/include/gtest/gtest-printers.h @@ -0,0 +1,927 @@ +// Copyright 2007, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +// Google Test - The Google C++ Testing and Mocking Framework +// +// This file implements a universal value printer that can print a +// value of any type T: +// +// void ::testing::internal::UniversalPrinter::Print(value, ostream_ptr); +// +// A user can teach this function how to print a class type T by +// defining either operator<<() or PrintTo() in the namespace that +// defines T. More specifically, the FIRST defined function in the +// following list will be used (assuming T is defined in namespace +// foo): +// +// 1. foo::PrintTo(const T&, ostream*) +// 2. operator<<(ostream&, const T&) defined in either foo or the +// global namespace. +// +// However if T is an STL-style container then it is printed element-wise +// unless foo::PrintTo(const T&, ostream*) is defined. Note that +// operator<<() is ignored for container types. +// +// If none of the above is defined, it will print the debug string of +// the value if it is a protocol buffer, or print the raw bytes in the +// value otherwise. +// +// To aid debugging: when T is a reference type, the address of the +// value is also printed; when T is a (const) char pointer, both the +// pointer value and the NUL-terminated string it points to are +// printed. +// +// We also provide some convenient wrappers: +// +// // Prints a value to a string. For a (const or not) char +// // pointer, the NUL-terminated string (but not the pointer) is +// // printed. +// std::string ::testing::PrintToString(const T& value); +// +// // Prints a value tersely: for a reference type, the referenced +// // value (but not the address) is printed; for a (const or not) char +// // pointer, the NUL-terminated string (but not the pointer) is +// // printed. +// void ::testing::internal::UniversalTersePrint(const T& value, ostream*); +// +// // Prints value using the type inferred by the compiler. The difference +// // from UniversalTersePrint() is that this function prints both the +// // pointer and the NUL-terminated string for a (const or not) char pointer. +// void ::testing::internal::UniversalPrint(const T& value, ostream*); +// +// // Prints the fields of a tuple tersely to a string vector, one +// // element for each field. Tuple support must be enabled in +// // gtest-port.h. +// std::vector UniversalTersePrintTupleFieldsToStrings( +// const Tuple& value); +// +// Known limitation: +// +// The print primitives print the elements of an STL-style container +// using the compiler-inferred type of *iter where iter is a +// const_iterator of the container. When const_iterator is an input +// iterator but not a forward iterator, this inferred type may not +// match value_type, and the print output may be incorrect. In +// practice, this is rarely a problem as for most containers +// const_iterator is a forward iterator. We'll fix this if there's an +// actual need for it. Note that this fix cannot rely on value_type +// being defined as many user-defined container types don't have +// value_type. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_ +#define GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_ + +#include +#include // NOLINT +#include +#include +#include +#include +#include +#include +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-port.h" + +#if GTEST_HAS_ABSL +#include "absl/strings/string_view.h" +#include "absl/types/optional.h" +#include "absl/types/variant.h" +#endif // GTEST_HAS_ABSL + +namespace testing { + +// Definitions in the 'internal' and 'internal2' name spaces are +// subject to change without notice. DO NOT USE THEM IN USER CODE! +namespace internal2 { + +// Prints the given number of bytes in the given object to the given +// ostream. +GTEST_API_ void PrintBytesInObjectTo(const unsigned char* obj_bytes, + size_t count, + ::std::ostream* os); + +// For selecting which printer to use when a given type has neither << +// nor PrintTo(). +enum TypeKind { + kProtobuf, // a protobuf type + kConvertibleToInteger, // a type implicitly convertible to BiggestInt + // (e.g. a named or unnamed enum type) +#if GTEST_HAS_ABSL + kConvertibleToStringView, // a type implicitly convertible to + // absl::string_view +#endif + kOtherType // anything else +}; + +// TypeWithoutFormatter::PrintValue(value, os) is called +// by the universal printer to print a value of type T when neither +// operator<< nor PrintTo() is defined for T, where kTypeKind is the +// "kind" of T as defined by enum TypeKind. +template +class TypeWithoutFormatter { + public: + // This default version is called when kTypeKind is kOtherType. + static void PrintValue(const T& value, ::std::ostream* os) { + PrintBytesInObjectTo(static_cast( + reinterpret_cast(&value)), + sizeof(value), os); + } +}; + +// We print a protobuf using its ShortDebugString() when the string +// doesn't exceed this many characters; otherwise we print it using +// DebugString() for better readability. +const size_t kProtobufOneLinerMaxLength = 50; + +template +class TypeWithoutFormatter { + public: + static void PrintValue(const T& value, ::std::ostream* os) { + std::string pretty_str = value.ShortDebugString(); + if (pretty_str.length() > kProtobufOneLinerMaxLength) { + pretty_str = "\n" + value.DebugString(); + } + *os << ("<" + pretty_str + ">"); + } +}; + +template +class TypeWithoutFormatter { + public: + // Since T has no << operator or PrintTo() but can be implicitly + // converted to BiggestInt, we print it as a BiggestInt. + // + // Most likely T is an enum type (either named or unnamed), in which + // case printing it as an integer is the desired behavior. In case + // T is not an enum, printing it as an integer is the best we can do + // given that it has no user-defined printer. + static void PrintValue(const T& value, ::std::ostream* os) { + const internal::BiggestInt kBigInt = value; + *os << kBigInt; + } +}; + +#if GTEST_HAS_ABSL +template +class TypeWithoutFormatter { + public: + // Since T has neither operator<< nor PrintTo() but can be implicitly + // converted to absl::string_view, we print it as a absl::string_view. + // + // Note: the implementation is further below, as it depends on + // internal::PrintTo symbol which is defined later in the file. + static void PrintValue(const T& value, ::std::ostream* os); +}; +#endif + +// Prints the given value to the given ostream. If the value is a +// protocol message, its debug string is printed; if it's an enum or +// of a type implicitly convertible to BiggestInt, it's printed as an +// integer; otherwise the bytes in the value are printed. This is +// what UniversalPrinter::Print() does when it knows nothing about +// type T and T has neither << operator nor PrintTo(). +// +// A user can override this behavior for a class type Foo by defining +// a << operator in the namespace where Foo is defined. +// +// We put this operator in namespace 'internal2' instead of 'internal' +// to simplify the implementation, as much code in 'internal' needs to +// use << in STL, which would conflict with our own << were it defined +// in 'internal'. +// +// Note that this operator<< takes a generic std::basic_ostream type instead of the more restricted std::ostream. If +// we define it to take an std::ostream instead, we'll get an +// "ambiguous overloads" compiler error when trying to print a type +// Foo that supports streaming to std::basic_ostream, as the compiler cannot tell whether +// operator<<(std::ostream&, const T&) or +// operator<<(std::basic_stream, const Foo&) is more +// specific. +template +::std::basic_ostream& operator<<( + ::std::basic_ostream& os, const T& x) { + TypeWithoutFormatter::value + ? kProtobuf + : std::is_convertible< + const T&, internal::BiggestInt>::value + ? kConvertibleToInteger + : +#if GTEST_HAS_ABSL + std::is_convertible< + const T&, absl::string_view>::value + ? kConvertibleToStringView + : +#endif + kOtherType)>::PrintValue(x, &os); + return os; +} + +} // namespace internal2 +} // namespace testing + +// This namespace MUST NOT BE NESTED IN ::testing, or the name look-up +// magic needed for implementing UniversalPrinter won't work. +namespace testing_internal { + +// Used to print a value that is not an STL-style container when the +// user doesn't define PrintTo() for it. +template +void DefaultPrintNonContainerTo(const T& value, ::std::ostream* os) { + // With the following statement, during unqualified name lookup, + // testing::internal2::operator<< appears as if it was declared in + // the nearest enclosing namespace that contains both + // ::testing_internal and ::testing::internal2, i.e. the global + // namespace. For more details, refer to the C++ Standard section + // 7.3.4-1 [namespace.udir]. This allows us to fall back onto + // testing::internal2::operator<< in case T doesn't come with a << + // operator. + // + // We cannot write 'using ::testing::internal2::operator<<;', which + // gcc 3.3 fails to compile due to a compiler bug. + using namespace ::testing::internal2; // NOLINT + + // Assuming T is defined in namespace foo, in the next statement, + // the compiler will consider all of: + // + // 1. foo::operator<< (thanks to Koenig look-up), + // 2. ::operator<< (as the current namespace is enclosed in ::), + // 3. testing::internal2::operator<< (thanks to the using statement above). + // + // The operator<< whose type matches T best will be picked. + // + // We deliberately allow #2 to be a candidate, as sometimes it's + // impossible to define #1 (e.g. when foo is ::std, defining + // anything in it is undefined behavior unless you are a compiler + // vendor.). + *os << value; +} + +} // namespace testing_internal + +namespace testing { +namespace internal { + +// FormatForComparison::Format(value) formats a +// value of type ToPrint that is an operand of a comparison assertion +// (e.g. ASSERT_EQ). OtherOperand is the type of the other operand in +// the comparison, and is used to help determine the best way to +// format the value. In particular, when the value is a C string +// (char pointer) and the other operand is an STL string object, we +// want to format the C string as a string, since we know it is +// compared by value with the string object. If the value is a char +// pointer but the other operand is not an STL string object, we don't +// know whether the pointer is supposed to point to a NUL-terminated +// string, and thus want to print it as a pointer to be safe. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. + +// The default case. +template +class FormatForComparison { + public: + static ::std::string Format(const ToPrint& value) { + return ::testing::PrintToString(value); + } +}; + +// Array. +template +class FormatForComparison { + public: + static ::std::string Format(const ToPrint* value) { + return FormatForComparison::Format(value); + } +}; + +// By default, print C string as pointers to be safe, as we don't know +// whether they actually point to a NUL-terminated string. + +#define GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(CharType) \ + template \ + class FormatForComparison { \ + public: \ + static ::std::string Format(CharType* value) { \ + return ::testing::PrintToString(static_cast(value)); \ + } \ + } + +GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(char); +GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const char); +GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(wchar_t); +GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const wchar_t); + +#undef GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_ + +// If a C string is compared with an STL string object, we know it's meant +// to point to a NUL-terminated string, and thus can print it as a string. + +#define GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(CharType, OtherStringType) \ + template <> \ + class FormatForComparison { \ + public: \ + static ::std::string Format(CharType* value) { \ + return ::testing::PrintToString(value); \ + } \ + } + +GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::std::string); +GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::std::string); + +#if GTEST_HAS_STD_WSTRING +GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::std::wstring); +GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::std::wstring); +#endif + +#undef GTEST_IMPL_FORMAT_C_STRING_AS_STRING_ + +// Formats a comparison assertion (e.g. ASSERT_EQ, EXPECT_LT, and etc) +// operand to be used in a failure message. The type (but not value) +// of the other operand may affect the format. This allows us to +// print a char* as a raw pointer when it is compared against another +// char* or void*, and print it as a C string when it is compared +// against an std::string object, for example. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +template +std::string FormatForComparisonFailureMessage( + const T1& value, const T2& /* other_operand */) { + return FormatForComparison::Format(value); +} + +// UniversalPrinter::Print(value, ostream_ptr) prints the given +// value to the given ostream. The caller must ensure that +// 'ostream_ptr' is not NULL, or the behavior is undefined. +// +// We define UniversalPrinter as a class template (as opposed to a +// function template), as we need to partially specialize it for +// reference types, which cannot be done with function templates. +template +class UniversalPrinter; + +template +void UniversalPrint(const T& value, ::std::ostream* os); + +enum DefaultPrinterType { + kPrintContainer, + kPrintPointer, + kPrintFunctionPointer, + kPrintOther, +}; +template struct WrapPrinterType {}; + +// Used to print an STL-style container when the user doesn't define +// a PrintTo() for it. +template +void DefaultPrintTo(WrapPrinterType /* dummy */, + const C& container, ::std::ostream* os) { + const size_t kMaxCount = 32; // The maximum number of elements to print. + *os << '{'; + size_t count = 0; + for (typename C::const_iterator it = container.begin(); + it != container.end(); ++it, ++count) { + if (count > 0) { + *os << ','; + if (count == kMaxCount) { // Enough has been printed. + *os << " ..."; + break; + } + } + *os << ' '; + // We cannot call PrintTo(*it, os) here as PrintTo() doesn't + // handle *it being a native array. + internal::UniversalPrint(*it, os); + } + + if (count > 0) { + *os << ' '; + } + *os << '}'; +} + +// Used to print a pointer that is neither a char pointer nor a member +// pointer, when the user doesn't define PrintTo() for it. (A member +// variable pointer or member function pointer doesn't really point to +// a location in the address space. Their representation is +// implementation-defined. Therefore they will be printed as raw +// bytes.) +template +void DefaultPrintTo(WrapPrinterType /* dummy */, + T* p, ::std::ostream* os) { + if (p == nullptr) { + *os << "NULL"; + } else { + // T is not a function type. We just call << to print p, + // relying on ADL to pick up user-defined << for their pointer + // types, if any. + *os << p; + } +} +template +void DefaultPrintTo(WrapPrinterType /* dummy */, + T* p, ::std::ostream* os) { + if (p == nullptr) { + *os << "NULL"; + } else { + // T is a function type, so '*os << p' doesn't do what we want + // (it just prints p as bool). We want to print p as a const + // void*. + *os << reinterpret_cast(p); + } +} + +// Used to print a non-container, non-pointer value when the user +// doesn't define PrintTo() for it. +template +void DefaultPrintTo(WrapPrinterType /* dummy */, + const T& value, ::std::ostream* os) { + ::testing_internal::DefaultPrintNonContainerTo(value, os); +} + +// Prints the given value using the << operator if it has one; +// otherwise prints the bytes in it. This is what +// UniversalPrinter::Print() does when PrintTo() is not specialized +// or overloaded for type T. +// +// A user can override this behavior for a class type Foo by defining +// an overload of PrintTo() in the namespace where Foo is defined. We +// give the user this option as sometimes defining a << operator for +// Foo is not desirable (e.g. the coding style may prevent doing it, +// or there is already a << operator but it doesn't do what the user +// wants). +template +void PrintTo(const T& value, ::std::ostream* os) { + // DefaultPrintTo() is overloaded. The type of its first argument + // determines which version will be picked. + // + // Note that we check for container types here, prior to we check + // for protocol message types in our operator<<. The rationale is: + // + // For protocol messages, we want to give people a chance to + // override Google Mock's format by defining a PrintTo() or + // operator<<. For STL containers, other formats can be + // incompatible with Google Mock's format for the container + // elements; therefore we check for container types here to ensure + // that our format is used. + // + // Note that MSVC and clang-cl do allow an implicit conversion from + // pointer-to-function to pointer-to-object, but clang-cl warns on it. + // So don't use ImplicitlyConvertible if it can be helped since it will + // cause this warning, and use a separate overload of DefaultPrintTo for + // function pointers so that the `*os << p` in the object pointer overload + // doesn't cause that warning either. + DefaultPrintTo( + WrapPrinterType < + (sizeof(IsContainerTest(0)) == sizeof(IsContainer)) && + !IsRecursiveContainer::value + ? kPrintContainer + : !std::is_pointer::value + ? kPrintOther + : std::is_function::type>::value + ? kPrintFunctionPointer + : kPrintPointer > (), + value, os); +} + +// The following list of PrintTo() overloads tells +// UniversalPrinter::Print() how to print standard types (built-in +// types, strings, plain arrays, and pointers). + +// Overloads for various char types. +GTEST_API_ void PrintTo(unsigned char c, ::std::ostream* os); +GTEST_API_ void PrintTo(signed char c, ::std::ostream* os); +inline void PrintTo(char c, ::std::ostream* os) { + // When printing a plain char, we always treat it as unsigned. This + // way, the output won't be affected by whether the compiler thinks + // char is signed or not. + PrintTo(static_cast(c), os); +} + +// Overloads for other simple built-in types. +inline void PrintTo(bool x, ::std::ostream* os) { + *os << (x ? "true" : "false"); +} + +// Overload for wchar_t type. +// Prints a wchar_t as a symbol if it is printable or as its internal +// code otherwise and also as its decimal code (except for L'\0'). +// The L'\0' char is printed as "L'\\0'". The decimal code is printed +// as signed integer when wchar_t is implemented by the compiler +// as a signed type and is printed as an unsigned integer when wchar_t +// is implemented as an unsigned type. +GTEST_API_ void PrintTo(wchar_t wc, ::std::ostream* os); + +// Overloads for C strings. +GTEST_API_ void PrintTo(const char* s, ::std::ostream* os); +inline void PrintTo(char* s, ::std::ostream* os) { + PrintTo(ImplicitCast_(s), os); +} + +// signed/unsigned char is often used for representing binary data, so +// we print pointers to it as void* to be safe. +inline void PrintTo(const signed char* s, ::std::ostream* os) { + PrintTo(ImplicitCast_(s), os); +} +inline void PrintTo(signed char* s, ::std::ostream* os) { + PrintTo(ImplicitCast_(s), os); +} +inline void PrintTo(const unsigned char* s, ::std::ostream* os) { + PrintTo(ImplicitCast_(s), os); +} +inline void PrintTo(unsigned char* s, ::std::ostream* os) { + PrintTo(ImplicitCast_(s), os); +} + +// MSVC can be configured to define wchar_t as a typedef of unsigned +// short. It defines _NATIVE_WCHAR_T_DEFINED when wchar_t is a native +// type. When wchar_t is a typedef, defining an overload for const +// wchar_t* would cause unsigned short* be printed as a wide string, +// possibly causing invalid memory accesses. +#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED) +// Overloads for wide C strings +GTEST_API_ void PrintTo(const wchar_t* s, ::std::ostream* os); +inline void PrintTo(wchar_t* s, ::std::ostream* os) { + PrintTo(ImplicitCast_(s), os); +} +#endif + +// Overload for C arrays. Multi-dimensional arrays are printed +// properly. + +// Prints the given number of elements in an array, without printing +// the curly braces. +template +void PrintRawArrayTo(const T a[], size_t count, ::std::ostream* os) { + UniversalPrint(a[0], os); + for (size_t i = 1; i != count; i++) { + *os << ", "; + UniversalPrint(a[i], os); + } +} + +// Overloads for ::std::string. +GTEST_API_ void PrintStringTo(const ::std::string&s, ::std::ostream* os); +inline void PrintTo(const ::std::string& s, ::std::ostream* os) { + PrintStringTo(s, os); +} + +// Overloads for ::std::wstring. +#if GTEST_HAS_STD_WSTRING +GTEST_API_ void PrintWideStringTo(const ::std::wstring&s, ::std::ostream* os); +inline void PrintTo(const ::std::wstring& s, ::std::ostream* os) { + PrintWideStringTo(s, os); +} +#endif // GTEST_HAS_STD_WSTRING + +#if GTEST_HAS_ABSL +// Overload for absl::string_view. +inline void PrintTo(absl::string_view sp, ::std::ostream* os) { + PrintTo(::std::string(sp), os); +} +#endif // GTEST_HAS_ABSL + +inline void PrintTo(std::nullptr_t, ::std::ostream* os) { *os << "(nullptr)"; } + +template +void PrintTo(std::reference_wrapper ref, ::std::ostream* os) { + UniversalPrinter::Print(ref.get(), os); +} + +// Helper function for printing a tuple. T must be instantiated with +// a tuple type. +template +void PrintTupleTo(const T&, std::integral_constant, + ::std::ostream*) {} + +template +void PrintTupleTo(const T& t, std::integral_constant, + ::std::ostream* os) { + PrintTupleTo(t, std::integral_constant(), os); + GTEST_INTENTIONAL_CONST_COND_PUSH_() + if (I > 1) { + GTEST_INTENTIONAL_CONST_COND_POP_() + *os << ", "; + } + UniversalPrinter::type>::Print( + std::get(t), os); +} + +template +void PrintTo(const ::std::tuple& t, ::std::ostream* os) { + *os << "("; + PrintTupleTo(t, std::integral_constant(), os); + *os << ")"; +} + +// Overload for std::pair. +template +void PrintTo(const ::std::pair& value, ::std::ostream* os) { + *os << '('; + // We cannot use UniversalPrint(value.first, os) here, as T1 may be + // a reference type. The same for printing value.second. + UniversalPrinter::Print(value.first, os); + *os << ", "; + UniversalPrinter::Print(value.second, os); + *os << ')'; +} + +// Implements printing a non-reference type T by letting the compiler +// pick the right overload of PrintTo() for T. +template +class UniversalPrinter { + public: + // MSVC warns about adding const to a function type, so we want to + // disable the warning. + GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180) + + // Note: we deliberately don't call this PrintTo(), as that name + // conflicts with ::testing::internal::PrintTo in the body of the + // function. + static void Print(const T& value, ::std::ostream* os) { + // By default, ::testing::internal::PrintTo() is used for printing + // the value. + // + // Thanks to Koenig look-up, if T is a class and has its own + // PrintTo() function defined in its namespace, that function will + // be visible here. Since it is more specific than the generic ones + // in ::testing::internal, it will be picked by the compiler in the + // following statement - exactly what we want. + PrintTo(value, os); + } + + GTEST_DISABLE_MSC_WARNINGS_POP_() +}; + +#if GTEST_HAS_ABSL + +// Printer for absl::optional + +template +class UniversalPrinter<::absl::optional> { + public: + static void Print(const ::absl::optional& value, ::std::ostream* os) { + *os << '('; + if (!value) { + *os << "nullopt"; + } else { + UniversalPrint(*value, os); + } + *os << ')'; + } +}; + +// Printer for absl::variant + +template +class UniversalPrinter<::absl::variant> { + public: + static void Print(const ::absl::variant& value, ::std::ostream* os) { + *os << '('; + absl::visit(Visitor{os}, value); + *os << ')'; + } + + private: + struct Visitor { + template + void operator()(const U& u) const { + *os << "'" << GetTypeName() << "' with value "; + UniversalPrint(u, os); + } + ::std::ostream* os; + }; +}; + +#endif // GTEST_HAS_ABSL + +// UniversalPrintArray(begin, len, os) prints an array of 'len' +// elements, starting at address 'begin'. +template +void UniversalPrintArray(const T* begin, size_t len, ::std::ostream* os) { + if (len == 0) { + *os << "{}"; + } else { + *os << "{ "; + const size_t kThreshold = 18; + const size_t kChunkSize = 8; + // If the array has more than kThreshold elements, we'll have to + // omit some details by printing only the first and the last + // kChunkSize elements. + if (len <= kThreshold) { + PrintRawArrayTo(begin, len, os); + } else { + PrintRawArrayTo(begin, kChunkSize, os); + *os << ", ..., "; + PrintRawArrayTo(begin + len - kChunkSize, kChunkSize, os); + } + *os << " }"; + } +} +// This overload prints a (const) char array compactly. +GTEST_API_ void UniversalPrintArray( + const char* begin, size_t len, ::std::ostream* os); + +// This overload prints a (const) wchar_t array compactly. +GTEST_API_ void UniversalPrintArray( + const wchar_t* begin, size_t len, ::std::ostream* os); + +// Implements printing an array type T[N]. +template +class UniversalPrinter { + public: + // Prints the given array, omitting some elements when there are too + // many. + static void Print(const T (&a)[N], ::std::ostream* os) { + UniversalPrintArray(a, N, os); + } +}; + +// Implements printing a reference type T&. +template +class UniversalPrinter { + public: + // MSVC warns about adding const to a function type, so we want to + // disable the warning. + GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180) + + static void Print(const T& value, ::std::ostream* os) { + // Prints the address of the value. We use reinterpret_cast here + // as static_cast doesn't compile when T is a function type. + *os << "@" << reinterpret_cast(&value) << " "; + + // Then prints the value itself. + UniversalPrint(value, os); + } + + GTEST_DISABLE_MSC_WARNINGS_POP_() +}; + +// Prints a value tersely: for a reference type, the referenced value +// (but not the address) is printed; for a (const) char pointer, the +// NUL-terminated string (but not the pointer) is printed. + +template +class UniversalTersePrinter { + public: + static void Print(const T& value, ::std::ostream* os) { + UniversalPrint(value, os); + } +}; +template +class UniversalTersePrinter { + public: + static void Print(const T& value, ::std::ostream* os) { + UniversalPrint(value, os); + } +}; +template +class UniversalTersePrinter { + public: + static void Print(const T (&value)[N], ::std::ostream* os) { + UniversalPrinter::Print(value, os); + } +}; +template <> +class UniversalTersePrinter { + public: + static void Print(const char* str, ::std::ostream* os) { + if (str == nullptr) { + *os << "NULL"; + } else { + UniversalPrint(std::string(str), os); + } + } +}; +template <> +class UniversalTersePrinter { + public: + static void Print(char* str, ::std::ostream* os) { + UniversalTersePrinter::Print(str, os); + } +}; + +#if GTEST_HAS_STD_WSTRING +template <> +class UniversalTersePrinter { + public: + static void Print(const wchar_t* str, ::std::ostream* os) { + if (str == nullptr) { + *os << "NULL"; + } else { + UniversalPrint(::std::wstring(str), os); + } + } +}; +#endif + +template <> +class UniversalTersePrinter { + public: + static void Print(wchar_t* str, ::std::ostream* os) { + UniversalTersePrinter::Print(str, os); + } +}; + +template +void UniversalTersePrint(const T& value, ::std::ostream* os) { + UniversalTersePrinter::Print(value, os); +} + +// Prints a value using the type inferred by the compiler. The +// difference between this and UniversalTersePrint() is that for a +// (const) char pointer, this prints both the pointer and the +// NUL-terminated string. +template +void UniversalPrint(const T& value, ::std::ostream* os) { + // A workarond for the bug in VC++ 7.1 that prevents us from instantiating + // UniversalPrinter with T directly. + typedef T T1; + UniversalPrinter::Print(value, os); +} + +typedef ::std::vector< ::std::string> Strings; + + // Tersely prints the first N fields of a tuple to a string vector, + // one element for each field. +template +void TersePrintPrefixToStrings(const Tuple&, std::integral_constant, + Strings*) {} +template +void TersePrintPrefixToStrings(const Tuple& t, + std::integral_constant, + Strings* strings) { + TersePrintPrefixToStrings(t, std::integral_constant(), + strings); + ::std::stringstream ss; + UniversalTersePrint(std::get(t), &ss); + strings->push_back(ss.str()); +} + +// Prints the fields of a tuple tersely to a string vector, one +// element for each field. See the comment before +// UniversalTersePrint() for how we define "tersely". +template +Strings UniversalTersePrintTupleFieldsToStrings(const Tuple& value) { + Strings result; + TersePrintPrefixToStrings( + value, std::integral_constant::value>(), + &result); + return result; +} + +} // namespace internal + +#if GTEST_HAS_ABSL +namespace internal2 { +template +void TypeWithoutFormatter::PrintValue( + const T& value, ::std::ostream* os) { + internal::PrintTo(absl::string_view(value), os); +} +} // namespace internal2 +#endif + +template +::std::string PrintToString(const T& value) { + ::std::stringstream ss; + internal::UniversalTersePrinter::Print(value, &ss); + return ss.str(); +} + +} // namespace testing + +// Include any custom printer added by the local installation. +// We must include this header at the end to make sure it can use the +// declarations from this file. +#include "gtest/internal/custom/gtest-printers.h" + +#endif // GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_ diff --git a/deps/googletest/include/gtest/gtest-spi.h b/deps/googletest/include/gtest/gtest-spi.h new file mode 100644 index 000000000..aa38870e8 --- /dev/null +++ b/deps/googletest/include/gtest/gtest-spi.h @@ -0,0 +1,238 @@ +// Copyright 2007, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// Utilities for testing Google Test itself and code that uses Google Test +// (e.g. frameworks built on top of Google Test). + +// GOOGLETEST_CM0004 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_SPI_H_ +#define GTEST_INCLUDE_GTEST_GTEST_SPI_H_ + +#include "gtest/gtest.h" + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +namespace testing { + +// This helper class can be used to mock out Google Test failure reporting +// so that we can test Google Test or code that builds on Google Test. +// +// An object of this class appends a TestPartResult object to the +// TestPartResultArray object given in the constructor whenever a Google Test +// failure is reported. It can either intercept only failures that are +// generated in the same thread that created this object or it can intercept +// all generated failures. The scope of this mock object can be controlled with +// the second argument to the two arguments constructor. +class GTEST_API_ ScopedFakeTestPartResultReporter + : public TestPartResultReporterInterface { + public: + // The two possible mocking modes of this object. + enum InterceptMode { + INTERCEPT_ONLY_CURRENT_THREAD, // Intercepts only thread local failures. + INTERCEPT_ALL_THREADS // Intercepts all failures. + }; + + // The c'tor sets this object as the test part result reporter used + // by Google Test. The 'result' parameter specifies where to report the + // results. This reporter will only catch failures generated in the current + // thread. DEPRECATED + explicit ScopedFakeTestPartResultReporter(TestPartResultArray* result); + + // Same as above, but you can choose the interception scope of this object. + ScopedFakeTestPartResultReporter(InterceptMode intercept_mode, + TestPartResultArray* result); + + // The d'tor restores the previous test part result reporter. + ~ScopedFakeTestPartResultReporter() override; + + // Appends the TestPartResult object to the TestPartResultArray + // received in the constructor. + // + // This method is from the TestPartResultReporterInterface + // interface. + void ReportTestPartResult(const TestPartResult& result) override; + + private: + void Init(); + + const InterceptMode intercept_mode_; + TestPartResultReporterInterface* old_reporter_; + TestPartResultArray* const result_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedFakeTestPartResultReporter); +}; + +namespace internal { + +// A helper class for implementing EXPECT_FATAL_FAILURE() and +// EXPECT_NONFATAL_FAILURE(). Its destructor verifies that the given +// TestPartResultArray contains exactly one failure that has the given +// type and contains the given substring. If that's not the case, a +// non-fatal failure will be generated. +class GTEST_API_ SingleFailureChecker { + public: + // The constructor remembers the arguments. + SingleFailureChecker(const TestPartResultArray* results, + TestPartResult::Type type, const std::string& substr); + ~SingleFailureChecker(); + private: + const TestPartResultArray* const results_; + const TestPartResult::Type type_; + const std::string substr_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(SingleFailureChecker); +}; + +} // namespace internal + +} // namespace testing + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +// A set of macros for testing Google Test assertions or code that's expected +// to generate Google Test fatal failures. It verifies that the given +// statement will cause exactly one fatal Google Test failure with 'substr' +// being part of the failure message. +// +// There are two different versions of this macro. EXPECT_FATAL_FAILURE only +// affects and considers failures generated in the current thread and +// EXPECT_FATAL_FAILURE_ON_ALL_THREADS does the same but for all threads. +// +// The verification of the assertion is done correctly even when the statement +// throws an exception or aborts the current function. +// +// Known restrictions: +// - 'statement' cannot reference local non-static variables or +// non-static members of the current object. +// - 'statement' cannot return a value. +// - You cannot stream a failure message to this macro. +// +// Note that even though the implementations of the following two +// macros are much alike, we cannot refactor them to use a common +// helper macro, due to some peculiarity in how the preprocessor +// works. The AcceptsMacroThatExpandsToUnprotectedComma test in +// gtest_unittest.cc will fail to compile if we do that. +#define EXPECT_FATAL_FAILURE(statement, substr) \ + do { \ + class GTestExpectFatalFailureHelper {\ + public:\ + static void Execute() { statement; }\ + };\ + ::testing::TestPartResultArray gtest_failures;\ + ::testing::internal::SingleFailureChecker gtest_checker(\ + >est_failures, ::testing::TestPartResult::kFatalFailure, (substr));\ + {\ + ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ + ::testing::ScopedFakeTestPartResultReporter:: \ + INTERCEPT_ONLY_CURRENT_THREAD, >est_failures);\ + GTestExpectFatalFailureHelper::Execute();\ + }\ + } while (::testing::internal::AlwaysFalse()) + +#define EXPECT_FATAL_FAILURE_ON_ALL_THREADS(statement, substr) \ + do { \ + class GTestExpectFatalFailureHelper {\ + public:\ + static void Execute() { statement; }\ + };\ + ::testing::TestPartResultArray gtest_failures;\ + ::testing::internal::SingleFailureChecker gtest_checker(\ + >est_failures, ::testing::TestPartResult::kFatalFailure, (substr));\ + {\ + ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ + ::testing::ScopedFakeTestPartResultReporter:: \ + INTERCEPT_ALL_THREADS, >est_failures);\ + GTestExpectFatalFailureHelper::Execute();\ + }\ + } while (::testing::internal::AlwaysFalse()) + +// A macro for testing Google Test assertions or code that's expected to +// generate Google Test non-fatal failures. It asserts that the given +// statement will cause exactly one non-fatal Google Test failure with 'substr' +// being part of the failure message. +// +// There are two different versions of this macro. EXPECT_NONFATAL_FAILURE only +// affects and considers failures generated in the current thread and +// EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS does the same but for all threads. +// +// 'statement' is allowed to reference local variables and members of +// the current object. +// +// The verification of the assertion is done correctly even when the statement +// throws an exception or aborts the current function. +// +// Known restrictions: +// - You cannot stream a failure message to this macro. +// +// Note that even though the implementations of the following two +// macros are much alike, we cannot refactor them to use a common +// helper macro, due to some peculiarity in how the preprocessor +// works. If we do that, the code won't compile when the user gives +// EXPECT_NONFATAL_FAILURE() a statement that contains a macro that +// expands to code containing an unprotected comma. The +// AcceptsMacroThatExpandsToUnprotectedComma test in gtest_unittest.cc +// catches that. +// +// For the same reason, we have to write +// if (::testing::internal::AlwaysTrue()) { statement; } +// instead of +// GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) +// to avoid an MSVC warning on unreachable code. +#define EXPECT_NONFATAL_FAILURE(statement, substr) \ + do {\ + ::testing::TestPartResultArray gtest_failures;\ + ::testing::internal::SingleFailureChecker gtest_checker(\ + >est_failures, ::testing::TestPartResult::kNonFatalFailure, \ + (substr));\ + {\ + ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ + ::testing::ScopedFakeTestPartResultReporter:: \ + INTERCEPT_ONLY_CURRENT_THREAD, >est_failures);\ + if (::testing::internal::AlwaysTrue()) { statement; }\ + }\ + } while (::testing::internal::AlwaysFalse()) + +#define EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(statement, substr) \ + do {\ + ::testing::TestPartResultArray gtest_failures;\ + ::testing::internal::SingleFailureChecker gtest_checker(\ + >est_failures, ::testing::TestPartResult::kNonFatalFailure, \ + (substr));\ + {\ + ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\ + ::testing::ScopedFakeTestPartResultReporter::INTERCEPT_ALL_THREADS, \ + >est_failures);\ + if (::testing::internal::AlwaysTrue()) { statement; }\ + }\ + } while (::testing::internal::AlwaysFalse()) + +#endif // GTEST_INCLUDE_GTEST_GTEST_SPI_H_ diff --git a/deps/googletest/include/gtest/gtest-test-part.h b/deps/googletest/include/gtest/gtest-test-part.h new file mode 100644 index 000000000..1e1cb097a --- /dev/null +++ b/deps/googletest/include/gtest/gtest-test-part.h @@ -0,0 +1,184 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_ +#define GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_ + +#include +#include +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-string.h" + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +namespace testing { + +// A copyable object representing the result of a test part (i.e. an +// assertion or an explicit FAIL(), ADD_FAILURE(), or SUCCESS()). +// +// Don't inherit from TestPartResult as its destructor is not virtual. +class GTEST_API_ TestPartResult { + public: + // The possible outcomes of a test part (i.e. an assertion or an + // explicit SUCCEED(), FAIL(), or ADD_FAILURE()). + enum Type { + kSuccess, // Succeeded. + kNonFatalFailure, // Failed but the test can continue. + kFatalFailure, // Failed and the test should be terminated. + kSkip // Skipped. + }; + + // C'tor. TestPartResult does NOT have a default constructor. + // Always use this constructor (with parameters) to create a + // TestPartResult object. + TestPartResult(Type a_type, const char* a_file_name, int a_line_number, + const char* a_message) + : type_(a_type), + file_name_(a_file_name == nullptr ? "" : a_file_name), + line_number_(a_line_number), + summary_(ExtractSummary(a_message)), + message_(a_message) {} + + // Gets the outcome of the test part. + Type type() const { return type_; } + + // Gets the name of the source file where the test part took place, or + // NULL if it's unknown. + const char* file_name() const { + return file_name_.empty() ? nullptr : file_name_.c_str(); + } + + // Gets the line in the source file where the test part took place, + // or -1 if it's unknown. + int line_number() const { return line_number_; } + + // Gets the summary of the failure message. + const char* summary() const { return summary_.c_str(); } + + // Gets the message associated with the test part. + const char* message() const { return message_.c_str(); } + + // Returns true iff the test part was skipped. + bool skipped() const { return type_ == kSkip; } + + // Returns true iff the test part passed. + bool passed() const { return type_ == kSuccess; } + + // Returns true iff the test part non-fatally failed. + bool nonfatally_failed() const { return type_ == kNonFatalFailure; } + + // Returns true iff the test part fatally failed. + bool fatally_failed() const { return type_ == kFatalFailure; } + + // Returns true iff the test part failed. + bool failed() const { return fatally_failed() || nonfatally_failed(); } + + private: + Type type_; + + // Gets the summary of the failure message by omitting the stack + // trace in it. + static std::string ExtractSummary(const char* message); + + // The name of the source file where the test part took place, or + // "" if the source file is unknown. + std::string file_name_; + // The line in the source file where the test part took place, or -1 + // if the line number is unknown. + int line_number_; + std::string summary_; // The test failure summary. + std::string message_; // The test failure message. +}; + +// Prints a TestPartResult object. +std::ostream& operator<<(std::ostream& os, const TestPartResult& result); + +// An array of TestPartResult objects. +// +// Don't inherit from TestPartResultArray as its destructor is not +// virtual. +class GTEST_API_ TestPartResultArray { + public: + TestPartResultArray() {} + + // Appends the given TestPartResult to the array. + void Append(const TestPartResult& result); + + // Returns the TestPartResult at the given index (0-based). + const TestPartResult& GetTestPartResult(int index) const; + + // Returns the number of TestPartResult objects in the array. + int size() const; + + private: + std::vector array_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestPartResultArray); +}; + +// This interface knows how to report a test part result. +class GTEST_API_ TestPartResultReporterInterface { + public: + virtual ~TestPartResultReporterInterface() {} + + virtual void ReportTestPartResult(const TestPartResult& result) = 0; +}; + +namespace internal { + +// This helper class is used by {ASSERT|EXPECT}_NO_FATAL_FAILURE to check if a +// statement generates new fatal failures. To do so it registers itself as the +// current test part result reporter. Besides checking if fatal failures were +// reported, it only delegates the reporting to the former result reporter. +// The original result reporter is restored in the destructor. +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +class GTEST_API_ HasNewFatalFailureHelper + : public TestPartResultReporterInterface { + public: + HasNewFatalFailureHelper(); + ~HasNewFatalFailureHelper() override; + void ReportTestPartResult(const TestPartResult& result) override; + bool has_new_fatal_failure() const { return has_new_fatal_failure_; } + private: + bool has_new_fatal_failure_; + TestPartResultReporterInterface* original_reporter_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(HasNewFatalFailureHelper); +}; + +} // namespace internal + +} // namespace testing + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +#endif // GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_ diff --git a/deps/googletest/include/gtest/gtest-typed-test.h b/deps/googletest/include/gtest/gtest-typed-test.h new file mode 100644 index 000000000..b3319f682 --- /dev/null +++ b/deps/googletest/include/gtest/gtest-typed-test.h @@ -0,0 +1,336 @@ +// Copyright 2008 Google Inc. +// All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_ +#define GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_ + +// This header implements typed tests and type-parameterized tests. + +// Typed (aka type-driven) tests repeat the same test for types in a +// list. You must know which types you want to test with when writing +// typed tests. Here's how you do it: + +#if 0 + +// First, define a fixture class template. It should be parameterized +// by a type. Remember to derive it from testing::Test. +template +class FooTest : public testing::Test { + public: + ... + typedef std::list List; + static T shared_; + T value_; +}; + +// Next, associate a list of types with the test suite, which will be +// repeated for each type in the list. The typedef is necessary for +// the macro to parse correctly. +typedef testing::Types MyTypes; +TYPED_TEST_SUITE(FooTest, MyTypes); + +// If the type list contains only one type, you can write that type +// directly without Types<...>: +// TYPED_TEST_SUITE(FooTest, int); + +// Then, use TYPED_TEST() instead of TEST_F() to define as many typed +// tests for this test suite as you want. +TYPED_TEST(FooTest, DoesBlah) { + // Inside a test, refer to TypeParam to get the type parameter. + // Since we are inside a derived class template, C++ requires use to + // visit the members of FooTest via 'this'. + TypeParam n = this->value_; + + // To visit static members of the fixture, add the TestFixture:: + // prefix. + n += TestFixture::shared_; + + // To refer to typedefs in the fixture, add the "typename + // TestFixture::" prefix. + typename TestFixture::List values; + values.push_back(n); + ... +} + +TYPED_TEST(FooTest, HasPropertyA) { ... } + +// TYPED_TEST_SUITE takes an optional third argument which allows to specify a +// class that generates custom test name suffixes based on the type. This should +// be a class which has a static template function GetName(int index) returning +// a string for each type. The provided integer index equals the index of the +// type in the provided type list. In many cases the index can be ignored. +// +// For example: +// class MyTypeNames { +// public: +// template +// static std::string GetName(int) { +// if (std::is_same()) return "char"; +// if (std::is_same()) return "int"; +// if (std::is_same()) return "unsignedInt"; +// } +// }; +// TYPED_TEST_SUITE(FooTest, MyTypes, MyTypeNames); + +#endif // 0 + +// Type-parameterized tests are abstract test patterns parameterized +// by a type. Compared with typed tests, type-parameterized tests +// allow you to define the test pattern without knowing what the type +// parameters are. The defined pattern can be instantiated with +// different types any number of times, in any number of translation +// units. +// +// If you are designing an interface or concept, you can define a +// suite of type-parameterized tests to verify properties that any +// valid implementation of the interface/concept should have. Then, +// each implementation can easily instantiate the test suite to verify +// that it conforms to the requirements, without having to write +// similar tests repeatedly. Here's an example: + +#if 0 + +// First, define a fixture class template. It should be parameterized +// by a type. Remember to derive it from testing::Test. +template +class FooTest : public testing::Test { + ... +}; + +// Next, declare that you will define a type-parameterized test suite +// (the _P suffix is for "parameterized" or "pattern", whichever you +// prefer): +TYPED_TEST_SUITE_P(FooTest); + +// Then, use TYPED_TEST_P() to define as many type-parameterized tests +// for this type-parameterized test suite as you want. +TYPED_TEST_P(FooTest, DoesBlah) { + // Inside a test, refer to TypeParam to get the type parameter. + TypeParam n = 0; + ... +} + +TYPED_TEST_P(FooTest, HasPropertyA) { ... } + +// Now the tricky part: you need to register all test patterns before +// you can instantiate them. The first argument of the macro is the +// test suite name; the rest are the names of the tests in this test +// case. +REGISTER_TYPED_TEST_SUITE_P(FooTest, + DoesBlah, HasPropertyA); + +// Finally, you are free to instantiate the pattern with the types you +// want. If you put the above code in a header file, you can #include +// it in multiple C++ source files and instantiate it multiple times. +// +// To distinguish different instances of the pattern, the first +// argument to the INSTANTIATE_* macro is a prefix that will be added +// to the actual test suite name. Remember to pick unique prefixes for +// different instances. +typedef testing::Types MyTypes; +INSTANTIATE_TYPED_TEST_SUITE_P(My, FooTest, MyTypes); + +// If the type list contains only one type, you can write that type +// directly without Types<...>: +// INSTANTIATE_TYPED_TEST_SUITE_P(My, FooTest, int); +// +// Similar to the optional argument of TYPED_TEST_SUITE above, +// INSTANTIATE_TEST_SUITE_P takes an optional fourth argument which allows to +// generate custom names. +// INSTANTIATE_TYPED_TEST_SUITE_P(My, FooTest, MyTypes, MyTypeNames); + +#endif // 0 + +#include "gtest/internal/gtest-port.h" +#include "gtest/internal/gtest-type-util.h" + +// Implements typed tests. + +#if GTEST_HAS_TYPED_TEST + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// Expands to the name of the typedef for the type parameters of the +// given test suite. +#define GTEST_TYPE_PARAMS_(TestSuiteName) gtest_type_params_##TestSuiteName##_ + +// Expands to the name of the typedef for the NameGenerator, responsible for +// creating the suffixes of the name. +#define GTEST_NAME_GENERATOR_(TestSuiteName) \ + gtest_type_params_##TestSuiteName##_NameGenerator + +// The 'Types' template argument below must have spaces around it +// since some compilers may choke on '>>' when passing a template +// instance (e.g. Types) +#define TYPED_TEST_SUITE(CaseName, Types, ...) \ + typedef ::testing::internal::TypeList::type GTEST_TYPE_PARAMS_( \ + CaseName); \ + typedef ::testing::internal::NameGeneratorSelector<__VA_ARGS__>::type \ + GTEST_NAME_GENERATOR_(CaseName) + +# define TYPED_TEST(CaseName, TestName) \ + template \ + class GTEST_TEST_CLASS_NAME_(CaseName, TestName) \ + : public CaseName { \ + private: \ + typedef CaseName TestFixture; \ + typedef gtest_TypeParam_ TypeParam; \ + virtual void TestBody(); \ + }; \ + static bool gtest_##CaseName##_##TestName##_registered_ \ + GTEST_ATTRIBUTE_UNUSED_ = \ + ::testing::internal::TypeParameterizedTest< \ + CaseName, \ + ::testing::internal::TemplateSel, \ + GTEST_TYPE_PARAMS_( \ + CaseName)>::Register("", \ + ::testing::internal::CodeLocation( \ + __FILE__, __LINE__), \ + #CaseName, #TestName, 0, \ + ::testing::internal::GenerateNames< \ + GTEST_NAME_GENERATOR_(CaseName), \ + GTEST_TYPE_PARAMS_(CaseName)>()); \ + template \ + void GTEST_TEST_CLASS_NAME_(CaseName, \ + TestName)::TestBody() + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +#define TYPED_TEST_CASE \ + static_assert(::testing::internal::TypedTestCaseIsDeprecated(), ""); \ + TYPED_TEST_SUITE +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +#endif // GTEST_HAS_TYPED_TEST + +// Implements type-parameterized tests. + +#if GTEST_HAS_TYPED_TEST_P + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// Expands to the namespace name that the type-parameterized tests for +// the given type-parameterized test suite are defined in. The exact +// name of the namespace is subject to change without notice. +#define GTEST_SUITE_NAMESPACE_(TestSuiteName) gtest_suite_##TestSuiteName##_ + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// Expands to the name of the variable used to remember the names of +// the defined tests in the given test suite. +#define GTEST_TYPED_TEST_SUITE_P_STATE_(TestSuiteName) \ + gtest_typed_test_suite_p_state_##TestSuiteName##_ + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE DIRECTLY. +// +// Expands to the name of the variable used to remember the names of +// the registered tests in the given test suite. +#define GTEST_REGISTERED_TEST_NAMES_(TestSuiteName) \ + gtest_registered_test_names_##TestSuiteName##_ + +// The variables defined in the type-parameterized test macros are +// static as typically these macros are used in a .h file that can be +// #included in multiple translation units linked together. +#define TYPED_TEST_SUITE_P(SuiteName) \ + static ::testing::internal::TypedTestSuitePState \ + GTEST_TYPED_TEST_SUITE_P_STATE_(SuiteName) + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +#define TYPED_TEST_CASE_P \ + static_assert(::testing::internal::TypedTestCase_P_IsDeprecated(), ""); \ + TYPED_TEST_SUITE_P +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +#define TYPED_TEST_P(SuiteName, TestName) \ + namespace GTEST_SUITE_NAMESPACE_(SuiteName) { \ + template \ + class TestName : public SuiteName { \ + private: \ + typedef SuiteName TestFixture; \ + typedef gtest_TypeParam_ TypeParam; \ + virtual void TestBody(); \ + }; \ + static bool gtest_##TestName##_defined_ GTEST_ATTRIBUTE_UNUSED_ = \ + GTEST_TYPED_TEST_SUITE_P_STATE_(SuiteName).AddTestName( \ + __FILE__, __LINE__, #SuiteName, #TestName); \ + } \ + template \ + void GTEST_SUITE_NAMESPACE_( \ + SuiteName)::TestName::TestBody() + +#define REGISTER_TYPED_TEST_SUITE_P(SuiteName, ...) \ + namespace GTEST_SUITE_NAMESPACE_(SuiteName) { \ + typedef ::testing::internal::Templates<__VA_ARGS__>::type gtest_AllTests_; \ + } \ + static const char* const GTEST_REGISTERED_TEST_NAMES_( \ + SuiteName) GTEST_ATTRIBUTE_UNUSED_ = \ + GTEST_TYPED_TEST_SUITE_P_STATE_(SuiteName).VerifyRegisteredTestNames( \ + __FILE__, __LINE__, #__VA_ARGS__) + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +#define REGISTER_TYPED_TEST_CASE_P \ + static_assert(::testing::internal::RegisterTypedTestCase_P_IsDeprecated(), \ + ""); \ + REGISTER_TYPED_TEST_SUITE_P +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +// The 'Types' template argument below must have spaces around it +// since some compilers may choke on '>>' when passing a template +// instance (e.g. Types) +#define INSTANTIATE_TYPED_TEST_SUITE_P(Prefix, SuiteName, Types, ...) \ + static bool gtest_##Prefix##_##SuiteName GTEST_ATTRIBUTE_UNUSED_ = \ + ::testing::internal::TypeParameterizedTestSuite< \ + SuiteName, GTEST_SUITE_NAMESPACE_(SuiteName)::gtest_AllTests_, \ + ::testing::internal::TypeList::type>:: \ + Register(#Prefix, \ + ::testing::internal::CodeLocation(__FILE__, __LINE__), \ + >EST_TYPED_TEST_SUITE_P_STATE_(SuiteName), #SuiteName, \ + GTEST_REGISTERED_TEST_NAMES_(SuiteName), \ + ::testing::internal::GenerateNames< \ + ::testing::internal::NameGeneratorSelector< \ + __VA_ARGS__>::type, \ + ::testing::internal::TypeList::type>()) + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +#define INSTANTIATE_TYPED_TEST_CASE_P \ + static_assert( \ + ::testing::internal::InstantiateTypedTestCase_P_IsDeprecated(), ""); \ + INSTANTIATE_TYPED_TEST_SUITE_P +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +#endif // GTEST_HAS_TYPED_TEST_P + +#endif // GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_ diff --git a/deps/googletest/include/gtest/gtest.h b/deps/googletest/include/gtest/gtest.h new file mode 100644 index 000000000..76f4098bf --- /dev/null +++ b/deps/googletest/include/gtest/gtest.h @@ -0,0 +1,2453 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file defines the public API for Google Test. It should be +// included by any test program that uses Google Test. +// +// IMPORTANT NOTE: Due to limitation of the C++ language, we have to +// leave some internal implementation details in this header file. +// They are clearly marked by comments like this: +// +// // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +// +// Such code is NOT meant to be used by a user directly, and is subject +// to CHANGE WITHOUT NOTICE. Therefore DO NOT DEPEND ON IT in a user +// program! +// +// Acknowledgment: Google Test borrowed the idea of automatic test +// registration from Barthelemy Dagenais' (barthelemy@prologique.com) +// easyUnit framework. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_H_ +#define GTEST_INCLUDE_GTEST_GTEST_H_ + +#include +#include +#include +#include +#include +#include + +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-string.h" +#include "gtest/gtest-death-test.h" +#include "gtest/gtest-matchers.h" +#include "gtest/gtest-message.h" +#include "gtest/gtest-param-test.h" +#include "gtest/gtest-printers.h" +#include "gtest/gtest_prod.h" +#include "gtest/gtest-test-part.h" +#include "gtest/gtest-typed-test.h" + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +namespace testing { + +// Silence C4100 (unreferenced formal parameter) and 4805 +// unsafe mix of type 'const int' and type 'const bool' +#ifdef _MSC_VER +# pragma warning(push) +# pragma warning(disable:4805) +# pragma warning(disable:4100) +#endif + + +// Declares the flags. + +// This flag temporary enables the disabled tests. +GTEST_DECLARE_bool_(also_run_disabled_tests); + +// This flag brings the debugger on an assertion failure. +GTEST_DECLARE_bool_(break_on_failure); + +// This flag controls whether Google Test catches all test-thrown exceptions +// and logs them as failures. +GTEST_DECLARE_bool_(catch_exceptions); + +// This flag enables using colors in terminal output. Available values are +// "yes" to enable colors, "no" (disable colors), or "auto" (the default) +// to let Google Test decide. +GTEST_DECLARE_string_(color); + +// This flag sets up the filter to select by name using a glob pattern +// the tests to run. If the filter is not given all tests are executed. +GTEST_DECLARE_string_(filter); + +// This flag controls whether Google Test installs a signal handler that dumps +// debugging information when fatal signals are raised. +GTEST_DECLARE_bool_(install_failure_signal_handler); + +// This flag causes the Google Test to list tests. None of the tests listed +// are actually run if the flag is provided. +GTEST_DECLARE_bool_(list_tests); + +// This flag controls whether Google Test emits a detailed XML report to a file +// in addition to its normal textual output. +GTEST_DECLARE_string_(output); + +// This flags control whether Google Test prints the elapsed time for each +// test. +GTEST_DECLARE_bool_(print_time); + +// This flags control whether Google Test prints UTF8 characters as text. +GTEST_DECLARE_bool_(print_utf8); + +// This flag specifies the random number seed. +GTEST_DECLARE_int32_(random_seed); + +// This flag sets how many times the tests are repeated. The default value +// is 1. If the value is -1 the tests are repeating forever. +GTEST_DECLARE_int32_(repeat); + +// This flag controls whether Google Test includes Google Test internal +// stack frames in failure stack traces. +GTEST_DECLARE_bool_(show_internal_stack_frames); + +// When this flag is specified, tests' order is randomized on every iteration. +GTEST_DECLARE_bool_(shuffle); + +// This flag specifies the maximum number of stack frames to be +// printed in a failure message. +GTEST_DECLARE_int32_(stack_trace_depth); + +// When this flag is specified, a failed assertion will throw an +// exception if exceptions are enabled, or exit the program with a +// non-zero code otherwise. For use with an external test framework. +GTEST_DECLARE_bool_(throw_on_failure); + +// When this flag is set with a "host:port" string, on supported +// platforms test results are streamed to the specified port on +// the specified host machine. +GTEST_DECLARE_string_(stream_result_to); + +#if GTEST_USE_OWN_FLAGFILE_FLAG_ +GTEST_DECLARE_string_(flagfile); +#endif // GTEST_USE_OWN_FLAGFILE_FLAG_ + +// The upper limit for valid stack trace depths. +const int kMaxStackTraceDepth = 100; + +namespace internal { + +class AssertHelper; +class DefaultGlobalTestPartResultReporter; +class ExecDeathTest; +class NoExecDeathTest; +class FinalSuccessChecker; +class GTestFlagSaver; +class StreamingListenerTest; +class TestResultAccessor; +class TestEventListenersAccessor; +class TestEventRepeater; +class UnitTestRecordPropertyTestHelper; +class WindowsDeathTest; +class FuchsiaDeathTest; +class UnitTestImpl* GetUnitTestImpl(); +void ReportFailureInUnknownLocation(TestPartResult::Type result_type, + const std::string& message); + +} // namespace internal + +// The friend relationship of some of these classes is cyclic. +// If we don't forward declare them the compiler might confuse the classes +// in friendship clauses with same named classes on the scope. +class Test; +class TestSuite; + +// Old API is still available but deprecated +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +using TestCase = TestSuite; +#endif +class TestInfo; +class UnitTest; + +// A class for indicating whether an assertion was successful. When +// the assertion wasn't successful, the AssertionResult object +// remembers a non-empty message that describes how it failed. +// +// To create an instance of this class, use one of the factory functions +// (AssertionSuccess() and AssertionFailure()). +// +// This class is useful for two purposes: +// 1. Defining predicate functions to be used with Boolean test assertions +// EXPECT_TRUE/EXPECT_FALSE and their ASSERT_ counterparts +// 2. Defining predicate-format functions to be +// used with predicate assertions (ASSERT_PRED_FORMAT*, etc). +// +// For example, if you define IsEven predicate: +// +// testing::AssertionResult IsEven(int n) { +// if ((n % 2) == 0) +// return testing::AssertionSuccess(); +// else +// return testing::AssertionFailure() << n << " is odd"; +// } +// +// Then the failed expectation EXPECT_TRUE(IsEven(Fib(5))) +// will print the message +// +// Value of: IsEven(Fib(5)) +// Actual: false (5 is odd) +// Expected: true +// +// instead of a more opaque +// +// Value of: IsEven(Fib(5)) +// Actual: false +// Expected: true +// +// in case IsEven is a simple Boolean predicate. +// +// If you expect your predicate to be reused and want to support informative +// messages in EXPECT_FALSE and ASSERT_FALSE (negative assertions show up +// about half as often as positive ones in our tests), supply messages for +// both success and failure cases: +// +// testing::AssertionResult IsEven(int n) { +// if ((n % 2) == 0) +// return testing::AssertionSuccess() << n << " is even"; +// else +// return testing::AssertionFailure() << n << " is odd"; +// } +// +// Then a statement EXPECT_FALSE(IsEven(Fib(6))) will print +// +// Value of: IsEven(Fib(6)) +// Actual: true (8 is even) +// Expected: false +// +// NB: Predicates that support negative Boolean assertions have reduced +// performance in positive ones so be careful not to use them in tests +// that have lots (tens of thousands) of positive Boolean assertions. +// +// To use this class with EXPECT_PRED_FORMAT assertions such as: +// +// // Verifies that Foo() returns an even number. +// EXPECT_PRED_FORMAT1(IsEven, Foo()); +// +// you need to define: +// +// testing::AssertionResult IsEven(const char* expr, int n) { +// if ((n % 2) == 0) +// return testing::AssertionSuccess(); +// else +// return testing::AssertionFailure() +// << "Expected: " << expr << " is even\n Actual: it's " << n; +// } +// +// If Foo() returns 5, you will see the following message: +// +// Expected: Foo() is even +// Actual: it's 5 +// +class GTEST_API_ AssertionResult { + public: + // Copy constructor. + // Used in EXPECT_TRUE/FALSE(assertion_result). + AssertionResult(const AssertionResult& other); + +#if defined(_MSC_VER) && _MSC_VER < 1910 + GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 /* forcing value to bool */) +#endif + + // Used in the EXPECT_TRUE/FALSE(bool_expression). + // + // T must be contextually convertible to bool. + // + // The second parameter prevents this overload from being considered if + // the argument is implicitly convertible to AssertionResult. In that case + // we want AssertionResult's copy constructor to be used. + template + explicit AssertionResult( + const T& success, + typename internal::EnableIf< + !std::is_convertible::value>::type* + /*enabler*/ + = nullptr) + : success_(success) {} + +#if defined(_MSC_VER) && _MSC_VER < 1910 + GTEST_DISABLE_MSC_WARNINGS_POP_() +#endif + + // Assignment operator. + AssertionResult& operator=(AssertionResult other) { + swap(other); + return *this; + } + + // Returns true iff the assertion succeeded. + operator bool() const { return success_; } // NOLINT + + // Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE. + AssertionResult operator!() const; + + // Returns the text streamed into this AssertionResult. Test assertions + // use it when they fail (i.e., the predicate's outcome doesn't match the + // assertion's expectation). When nothing has been streamed into the + // object, returns an empty string. + const char* message() const { + return message_.get() != nullptr ? message_->c_str() : ""; + } + // Deprecated; please use message() instead. + const char* failure_message() const { return message(); } + + // Streams a custom failure message into this object. + template AssertionResult& operator<<(const T& value) { + AppendMessage(Message() << value); + return *this; + } + + // Allows streaming basic output manipulators such as endl or flush into + // this object. + AssertionResult& operator<<( + ::std::ostream& (*basic_manipulator)(::std::ostream& stream)) { + AppendMessage(Message() << basic_manipulator); + return *this; + } + + private: + // Appends the contents of message to message_. + void AppendMessage(const Message& a_message) { + if (message_.get() == nullptr) message_.reset(new ::std::string); + message_->append(a_message.GetString().c_str()); + } + + // Swap the contents of this AssertionResult with other. + void swap(AssertionResult& other); + + // Stores result of the assertion predicate. + bool success_; + // Stores the message describing the condition in case the expectation + // construct is not satisfied with the predicate's outcome. + // Referenced via a pointer to avoid taking too much stack frame space + // with test assertions. + std::unique_ptr< ::std::string> message_; +}; + +// Makes a successful assertion result. +GTEST_API_ AssertionResult AssertionSuccess(); + +// Makes a failed assertion result. +GTEST_API_ AssertionResult AssertionFailure(); + +// Makes a failed assertion result with the given failure message. +// Deprecated; use AssertionFailure() << msg. +GTEST_API_ AssertionResult AssertionFailure(const Message& msg); + +} // namespace testing + +// Includes the auto-generated header that implements a family of generic +// predicate assertion macros. This include comes late because it relies on +// APIs declared above. +#include "gtest/gtest_pred_impl.h" + +namespace testing { + +// The abstract class that all tests inherit from. +// +// In Google Test, a unit test program contains one or many TestSuites, and +// each TestSuite contains one or many Tests. +// +// When you define a test using the TEST macro, you don't need to +// explicitly derive from Test - the TEST macro automatically does +// this for you. +// +// The only time you derive from Test is when defining a test fixture +// to be used in a TEST_F. For example: +// +// class FooTest : public testing::Test { +// protected: +// void SetUp() override { ... } +// void TearDown() override { ... } +// ... +// }; +// +// TEST_F(FooTest, Bar) { ... } +// TEST_F(FooTest, Baz) { ... } +// +// Test is not copyable. +class GTEST_API_ Test { + public: + friend class TestInfo; + + // The d'tor is virtual as we intend to inherit from Test. + virtual ~Test(); + + // Sets up the stuff shared by all tests in this test case. + // + // Google Test will call Foo::SetUpTestSuite() before running the first + // test in test case Foo. Hence a sub-class can define its own + // SetUpTestSuite() method to shadow the one defined in the super + // class. + static void SetUpTestSuite() {} + + // Tears down the stuff shared by all tests in this test case. + // + // Google Test will call Foo::TearDownTestSuite() after running the last + // test in test case Foo. Hence a sub-class can define its own + // TearDownTestSuite() method to shadow the one defined in the super + // class. + static void TearDownTestSuite() {} + + // Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + static void TearDownTestCase() {} + static void SetUpTestCase() {} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Returns true iff the current test has a fatal failure. + static bool HasFatalFailure(); + + // Returns true iff the current test has a non-fatal failure. + static bool HasNonfatalFailure(); + + // Returns true iff the current test was skipped. + static bool IsSkipped(); + + // Returns true iff the current test has a (either fatal or + // non-fatal) failure. + static bool HasFailure() { return HasFatalFailure() || HasNonfatalFailure(); } + + // Logs a property for the current test, test suite, or for the entire + // invocation of the test program when used outside of the context of a + // test suite. Only the last value for a given key is remembered. These + // are public static so they can be called from utility functions that are + // not members of the test fixture. Calls to RecordProperty made during + // lifespan of the test (from the moment its constructor starts to the + // moment its destructor finishes) will be output in XML as attributes of + // the element. Properties recorded from fixture's + // SetUpTestSuite or TearDownTestSuite are logged as attributes of the + // corresponding element. Calls to RecordProperty made in the + // global context (before or after invocation of RUN_ALL_TESTS and from + // SetUp/TearDown method of Environment objects registered with Google + // Test) will be output as attributes of the element. + static void RecordProperty(const std::string& key, const std::string& value); + static void RecordProperty(const std::string& key, int value); + + protected: + // Creates a Test object. + Test(); + + // Sets up the test fixture. + virtual void SetUp(); + + // Tears down the test fixture. + virtual void TearDown(); + + private: + // Returns true iff the current test has the same fixture class as + // the first test in the current test suite. + static bool HasSameFixtureClass(); + + // Runs the test after the test fixture has been set up. + // + // A sub-class must implement this to define the test logic. + // + // DO NOT OVERRIDE THIS FUNCTION DIRECTLY IN A USER PROGRAM. + // Instead, use the TEST or TEST_F macro. + virtual void TestBody() = 0; + + // Sets up, executes, and tears down the test. + void Run(); + + // Deletes self. We deliberately pick an unusual name for this + // internal method to avoid clashing with names used in user TESTs. + void DeleteSelf_() { delete this; } + + const std::unique_ptr gtest_flag_saver_; + + // Often a user misspells SetUp() as Setup() and spends a long time + // wondering why it is never called by Google Test. The declaration of + // the following method is solely for catching such an error at + // compile time: + // + // - The return type is deliberately chosen to be not void, so it + // will be a conflict if void Setup() is declared in the user's + // test fixture. + // + // - This method is private, so it will be another compiler error + // if the method is called from the user's test fixture. + // + // DO NOT OVERRIDE THIS FUNCTION. + // + // If you see an error about overriding the following function or + // about it being private, you have mis-spelled SetUp() as Setup(). + struct Setup_should_be_spelled_SetUp {}; + virtual Setup_should_be_spelled_SetUp* Setup() { return nullptr; } + + // We disallow copying Tests. + GTEST_DISALLOW_COPY_AND_ASSIGN_(Test); +}; + +typedef internal::TimeInMillis TimeInMillis; + +// A copyable object representing a user specified test property which can be +// output as a key/value string pair. +// +// Don't inherit from TestProperty as its destructor is not virtual. +class TestProperty { + public: + // C'tor. TestProperty does NOT have a default constructor. + // Always use this constructor (with parameters) to create a + // TestProperty object. + TestProperty(const std::string& a_key, const std::string& a_value) : + key_(a_key), value_(a_value) { + } + + // Gets the user supplied key. + const char* key() const { + return key_.c_str(); + } + + // Gets the user supplied value. + const char* value() const { + return value_.c_str(); + } + + // Sets a new value, overriding the one supplied in the constructor. + void SetValue(const std::string& new_value) { + value_ = new_value; + } + + private: + // The key supplied by the user. + std::string key_; + // The value supplied by the user. + std::string value_; +}; + +// The result of a single Test. This includes a list of +// TestPartResults, a list of TestProperties, a count of how many +// death tests there are in the Test, and how much time it took to run +// the Test. +// +// TestResult is not copyable. +class GTEST_API_ TestResult { + public: + // Creates an empty TestResult. + TestResult(); + + // D'tor. Do not inherit from TestResult. + ~TestResult(); + + // Gets the number of all test parts. This is the sum of the number + // of successful test parts and the number of failed test parts. + int total_part_count() const; + + // Returns the number of the test properties. + int test_property_count() const; + + // Returns true iff the test passed (i.e. no test part failed). + bool Passed() const { return !Skipped() && !Failed(); } + + // Returns true iff the test was skipped. + bool Skipped() const; + + // Returns true iff the test failed. + bool Failed() const; + + // Returns true iff the test fatally failed. + bool HasFatalFailure() const; + + // Returns true iff the test has a non-fatal failure. + bool HasNonfatalFailure() const; + + // Returns the elapsed time, in milliseconds. + TimeInMillis elapsed_time() const { return elapsed_time_; } + + // Returns the i-th test part result among all the results. i can range from 0 + // to total_part_count() - 1. If i is not in that range, aborts the program. + const TestPartResult& GetTestPartResult(int i) const; + + // Returns the i-th test property. i can range from 0 to + // test_property_count() - 1. If i is not in that range, aborts the + // program. + const TestProperty& GetTestProperty(int i) const; + + private: + friend class TestInfo; + friend class TestSuite; + friend class UnitTest; + friend class internal::DefaultGlobalTestPartResultReporter; + friend class internal::ExecDeathTest; + friend class internal::TestResultAccessor; + friend class internal::UnitTestImpl; + friend class internal::WindowsDeathTest; + friend class internal::FuchsiaDeathTest; + + // Gets the vector of TestPartResults. + const std::vector& test_part_results() const { + return test_part_results_; + } + + // Gets the vector of TestProperties. + const std::vector& test_properties() const { + return test_properties_; + } + + // Sets the elapsed time. + void set_elapsed_time(TimeInMillis elapsed) { elapsed_time_ = elapsed; } + + // Adds a test property to the list. The property is validated and may add + // a non-fatal failure if invalid (e.g., if it conflicts with reserved + // key names). If a property is already recorded for the same key, the + // value will be updated, rather than storing multiple values for the same + // key. xml_element specifies the element for which the property is being + // recorded and is used for validation. + void RecordProperty(const std::string& xml_element, + const TestProperty& test_property); + + // Adds a failure if the key is a reserved attribute of Google Test + // testsuite tags. Returns true if the property is valid. + // FIXME: Validate attribute names are legal and human readable. + static bool ValidateTestProperty(const std::string& xml_element, + const TestProperty& test_property); + + // Adds a test part result to the list. + void AddTestPartResult(const TestPartResult& test_part_result); + + // Returns the death test count. + int death_test_count() const { return death_test_count_; } + + // Increments the death test count, returning the new count. + int increment_death_test_count() { return ++death_test_count_; } + + // Clears the test part results. + void ClearTestPartResults(); + + // Clears the object. + void Clear(); + + // Protects mutable state of the property vector and of owned + // properties, whose values may be updated. + internal::Mutex test_properites_mutex_; + + // The vector of TestPartResults + std::vector test_part_results_; + // The vector of TestProperties + std::vector test_properties_; + // Running count of death tests. + int death_test_count_; + // The elapsed time, in milliseconds. + TimeInMillis elapsed_time_; + + // We disallow copying TestResult. + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestResult); +}; // class TestResult + +// A TestInfo object stores the following information about a test: +// +// Test suite name +// Test name +// Whether the test should be run +// A function pointer that creates the test object when invoked +// Test result +// +// The constructor of TestInfo registers itself with the UnitTest +// singleton such that the RUN_ALL_TESTS() macro knows which tests to +// run. +class GTEST_API_ TestInfo { + public: + // Destructs a TestInfo object. This function is not virtual, so + // don't inherit from TestInfo. + ~TestInfo(); + + // Returns the test suite name. + const char* test_suite_name() const { return test_suite_name_.c_str(); } + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + const char* test_case_name() const { return test_suite_name(); } +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Returns the test name. + const char* name() const { return name_.c_str(); } + + // Returns the name of the parameter type, or NULL if this is not a typed + // or a type-parameterized test. + const char* type_param() const { + if (type_param_.get() != nullptr) return type_param_->c_str(); + return nullptr; + } + + // Returns the text representation of the value parameter, or NULL if this + // is not a value-parameterized test. + const char* value_param() const { + if (value_param_.get() != nullptr) return value_param_->c_str(); + return nullptr; + } + + // Returns the file name where this test is defined. + const char* file() const { return location_.file.c_str(); } + + // Returns the line where this test is defined. + int line() const { return location_.line; } + + // Return true if this test should not be run because it's in another shard. + bool is_in_another_shard() const { return is_in_another_shard_; } + + // Returns true if this test should run, that is if the test is not + // disabled (or it is disabled but the also_run_disabled_tests flag has + // been specified) and its full name matches the user-specified filter. + // + // Google Test allows the user to filter the tests by their full names. + // The full name of a test Bar in test suite Foo is defined as + // "Foo.Bar". Only the tests that match the filter will run. + // + // A filter is a colon-separated list of glob (not regex) patterns, + // optionally followed by a '-' and a colon-separated list of + // negative patterns (tests to exclude). A test is run if it + // matches one of the positive patterns and does not match any of + // the negative patterns. + // + // For example, *A*:Foo.* is a filter that matches any string that + // contains the character 'A' or starts with "Foo.". + bool should_run() const { return should_run_; } + + // Returns true iff this test will appear in the XML report. + bool is_reportable() const { + // The XML report includes tests matching the filter, excluding those + // run in other shards. + return matches_filter_ && !is_in_another_shard_; + } + + // Returns the result of the test. + const TestResult* result() const { return &result_; } + + private: +#if GTEST_HAS_DEATH_TEST + friend class internal::DefaultDeathTestFactory; +#endif // GTEST_HAS_DEATH_TEST + friend class Test; + friend class TestSuite; + friend class internal::UnitTestImpl; + friend class internal::StreamingListenerTest; + friend TestInfo* internal::MakeAndRegisterTestInfo( + const char* test_suite_name, const char* name, const char* type_param, + const char* value_param, internal::CodeLocation code_location, + internal::TypeId fixture_class_id, internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc, + internal::TestFactoryBase* factory); + + // Constructs a TestInfo object. The newly constructed instance assumes + // ownership of the factory object. + TestInfo(const std::string& test_suite_name, const std::string& name, + const char* a_type_param, // NULL if not a type-parameterized test + const char* a_value_param, // NULL if not a value-parameterized test + internal::CodeLocation a_code_location, + internal::TypeId fixture_class_id, + internal::TestFactoryBase* factory); + + // Increments the number of death tests encountered in this test so + // far. + int increment_death_test_count() { + return result_.increment_death_test_count(); + } + + // Creates the test object, runs it, records its result, and then + // deletes it. + void Run(); + + static void ClearTestResult(TestInfo* test_info) { + test_info->result_.Clear(); + } + + // These fields are immutable properties of the test. + const std::string test_suite_name_; // test suite name + const std::string name_; // Test name + // Name of the parameter type, or NULL if this is not a typed or a + // type-parameterized test. + const std::unique_ptr type_param_; + // Text representation of the value parameter, or NULL if this is not a + // value-parameterized test. + const std::unique_ptr value_param_; + internal::CodeLocation location_; + const internal::TypeId fixture_class_id_; // ID of the test fixture class + bool should_run_; // True iff this test should run + bool is_disabled_; // True iff this test is disabled + bool matches_filter_; // True if this test matches the + // user-specified filter. + bool is_in_another_shard_; // Will be run in another shard. + internal::TestFactoryBase* const factory_; // The factory that creates + // the test object + + // This field is mutable and needs to be reset before running the + // test for the second time. + TestResult result_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestInfo); +}; + +// A test suite, which consists of a vector of TestInfos. +// +// TestSuite is not copyable. +class GTEST_API_ TestSuite { + public: + // Creates a TestSuite with the given name. + // + // TestSuite does NOT have a default constructor. Always use this + // constructor to create a TestSuite object. + // + // Arguments: + // + // name: name of the test suite + // a_type_param: the name of the test's type parameter, or NULL if + // this is not a type-parameterized test. + // set_up_tc: pointer to the function that sets up the test suite + // tear_down_tc: pointer to the function that tears down the test suite + TestSuite(const char* name, const char* a_type_param, + internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc); + + // Destructor of TestSuite. + virtual ~TestSuite(); + + // Gets the name of the TestSuite. + const char* name() const { return name_.c_str(); } + + // Returns the name of the parameter type, or NULL if this is not a + // type-parameterized test suite. + const char* type_param() const { + if (type_param_.get() != nullptr) return type_param_->c_str(); + return nullptr; + } + + // Returns true if any test in this test suite should run. + bool should_run() const { return should_run_; } + + // Gets the number of successful tests in this test suite. + int successful_test_count() const; + + // Gets the number of skipped tests in this test suite. + int skipped_test_count() const; + + // Gets the number of failed tests in this test suite. + int failed_test_count() const; + + // Gets the number of disabled tests that will be reported in the XML report. + int reportable_disabled_test_count() const; + + // Gets the number of disabled tests in this test suite. + int disabled_test_count() const; + + // Gets the number of tests to be printed in the XML report. + int reportable_test_count() const; + + // Get the number of tests in this test suite that should run. + int test_to_run_count() const; + + // Gets the number of all tests in this test suite. + int total_test_count() const; + + // Returns true iff the test suite passed. + bool Passed() const { return !Failed(); } + + // Returns true iff the test suite failed. + bool Failed() const { return failed_test_count() > 0; } + + // Returns the elapsed time, in milliseconds. + TimeInMillis elapsed_time() const { return elapsed_time_; } + + // Returns the i-th test among all the tests. i can range from 0 to + // total_test_count() - 1. If i is not in that range, returns NULL. + const TestInfo* GetTestInfo(int i) const; + + // Returns the TestResult that holds test properties recorded during + // execution of SetUpTestSuite and TearDownTestSuite. + const TestResult& ad_hoc_test_result() const { return ad_hoc_test_result_; } + + private: + friend class Test; + friend class internal::UnitTestImpl; + + // Gets the (mutable) vector of TestInfos in this TestSuite. + std::vector& test_info_list() { return test_info_list_; } + + // Gets the (immutable) vector of TestInfos in this TestSuite. + const std::vector& test_info_list() const { + return test_info_list_; + } + + // Returns the i-th test among all the tests. i can range from 0 to + // total_test_count() - 1. If i is not in that range, returns NULL. + TestInfo* GetMutableTestInfo(int i); + + // Sets the should_run member. + void set_should_run(bool should) { should_run_ = should; } + + // Adds a TestInfo to this test suite. Will delete the TestInfo upon + // destruction of the TestSuite object. + void AddTestInfo(TestInfo * test_info); + + // Clears the results of all tests in this test suite. + void ClearResult(); + + // Clears the results of all tests in the given test suite. + static void ClearTestSuiteResult(TestSuite* test_suite) { + test_suite->ClearResult(); + } + + // Runs every test in this TestSuite. + void Run(); + + // Runs SetUpTestSuite() for this TestSuite. This wrapper is needed + // for catching exceptions thrown from SetUpTestSuite(). + void RunSetUpTestSuite() { + if (set_up_tc_ != nullptr) { + (*set_up_tc_)(); + } + } + + // Runs TearDownTestSuite() for this TestSuite. This wrapper is + // needed for catching exceptions thrown from TearDownTestSuite(). + void RunTearDownTestSuite() { + if (tear_down_tc_ != nullptr) { + (*tear_down_tc_)(); + } + } + + // Returns true iff test passed. + static bool TestPassed(const TestInfo* test_info) { + return test_info->should_run() && test_info->result()->Passed(); + } + + // Returns true iff test skipped. + static bool TestSkipped(const TestInfo* test_info) { + return test_info->should_run() && test_info->result()->Skipped(); + } + + // Returns true iff test failed. + static bool TestFailed(const TestInfo* test_info) { + return test_info->should_run() && test_info->result()->Failed(); + } + + // Returns true iff the test is disabled and will be reported in the XML + // report. + static bool TestReportableDisabled(const TestInfo* test_info) { + return test_info->is_reportable() && test_info->is_disabled_; + } + + // Returns true iff test is disabled. + static bool TestDisabled(const TestInfo* test_info) { + return test_info->is_disabled_; + } + + // Returns true iff this test will appear in the XML report. + static bool TestReportable(const TestInfo* test_info) { + return test_info->is_reportable(); + } + + // Returns true if the given test should run. + static bool ShouldRunTest(const TestInfo* test_info) { + return test_info->should_run(); + } + + // Shuffles the tests in this test suite. + void ShuffleTests(internal::Random* random); + + // Restores the test order to before the first shuffle. + void UnshuffleTests(); + + // Name of the test suite. + std::string name_; + // Name of the parameter type, or NULL if this is not a typed or a + // type-parameterized test. + const std::unique_ptr type_param_; + // The vector of TestInfos in their original order. It owns the + // elements in the vector. + std::vector test_info_list_; + // Provides a level of indirection for the test list to allow easy + // shuffling and restoring the test order. The i-th element in this + // vector is the index of the i-th test in the shuffled test list. + std::vector test_indices_; + // Pointer to the function that sets up the test suite. + internal::SetUpTestSuiteFunc set_up_tc_; + // Pointer to the function that tears down the test suite. + internal::TearDownTestSuiteFunc tear_down_tc_; + // True iff any test in this test suite should run. + bool should_run_; + // Elapsed time, in milliseconds. + TimeInMillis elapsed_time_; + // Holds test properties recorded during execution of SetUpTestSuite and + // TearDownTestSuite. + TestResult ad_hoc_test_result_; + + // We disallow copying TestSuites. + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestSuite); +}; + +// An Environment object is capable of setting up and tearing down an +// environment. You should subclass this to define your own +// environment(s). +// +// An Environment object does the set-up and tear-down in virtual +// methods SetUp() and TearDown() instead of the constructor and the +// destructor, as: +// +// 1. You cannot safely throw from a destructor. This is a problem +// as in some cases Google Test is used where exceptions are enabled, and +// we may want to implement ASSERT_* using exceptions where they are +// available. +// 2. You cannot use ASSERT_* directly in a constructor or +// destructor. +class Environment { + public: + // The d'tor is virtual as we need to subclass Environment. + virtual ~Environment() {} + + // Override this to define how to set up the environment. + virtual void SetUp() {} + + // Override this to define how to tear down the environment. + virtual void TearDown() {} + private: + // If you see an error about overriding the following function or + // about it being private, you have mis-spelled SetUp() as Setup(). + struct Setup_should_be_spelled_SetUp {}; + virtual Setup_should_be_spelled_SetUp* Setup() { return nullptr; } +}; + +#if GTEST_HAS_EXCEPTIONS + +// Exception which can be thrown from TestEventListener::OnTestPartResult. +class GTEST_API_ AssertionException + : public internal::GoogleTestFailureException { + public: + explicit AssertionException(const TestPartResult& result) + : GoogleTestFailureException(result) {} +}; + +#endif // GTEST_HAS_EXCEPTIONS + +// The interface for tracing execution of tests. The methods are organized in +// the order the corresponding events are fired. +class TestEventListener { + public: + virtual ~TestEventListener() {} + + // Fired before any test activity starts. + virtual void OnTestProgramStart(const UnitTest& unit_test) = 0; + + // Fired before each iteration of tests starts. There may be more than + // one iteration if GTEST_FLAG(repeat) is set. iteration is the iteration + // index, starting from 0. + virtual void OnTestIterationStart(const UnitTest& unit_test, + int iteration) = 0; + + // Fired before environment set-up for each iteration of tests starts. + virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test) = 0; + + // Fired after environment set-up for each iteration of tests ends. + virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test) = 0; + + // Fired before the test suite starts. + virtual void OnTestSuiteStart(const TestSuite& /*test_suite*/) {} + + // Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + virtual void OnTestCaseStart(const TestCase& /*test_case*/) {} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Fired before the test starts. + virtual void OnTestStart(const TestInfo& test_info) = 0; + + // Fired after a failed assertion or a SUCCEED() invocation. + // If you want to throw an exception from this function to skip to the next + // TEST, it must be AssertionException defined above, or inherited from it. + virtual void OnTestPartResult(const TestPartResult& test_part_result) = 0; + + // Fired after the test ends. + virtual void OnTestEnd(const TestInfo& test_info) = 0; + + // Fired after the test suite ends. + virtual void OnTestSuiteEnd(const TestSuite& /*test_suite*/) {} + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + virtual void OnTestCaseEnd(const TestCase& /*test_case*/) {} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Fired before environment tear-down for each iteration of tests starts. + virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test) = 0; + + // Fired after environment tear-down for each iteration of tests ends. + virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test) = 0; + + // Fired after each iteration of tests finishes. + virtual void OnTestIterationEnd(const UnitTest& unit_test, + int iteration) = 0; + + // Fired after all test activities have ended. + virtual void OnTestProgramEnd(const UnitTest& unit_test) = 0; +}; + +// The convenience class for users who need to override just one or two +// methods and are not concerned that a possible change to a signature of +// the methods they override will not be caught during the build. For +// comments about each method please see the definition of TestEventListener +// above. +class EmptyTestEventListener : public TestEventListener { + public: + void OnTestProgramStart(const UnitTest& /*unit_test*/) override {} + void OnTestIterationStart(const UnitTest& /*unit_test*/, + int /*iteration*/) override {} + void OnEnvironmentsSetUpStart(const UnitTest& /*unit_test*/) override {} + void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) override {} + void OnTestSuiteStart(const TestSuite& /*test_suite*/) override {} +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + void OnTestCaseStart(const TestCase& /*test_case*/) override {} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + void OnTestStart(const TestInfo& /*test_info*/) override {} + void OnTestPartResult(const TestPartResult& /*test_part_result*/) override {} + void OnTestEnd(const TestInfo& /*test_info*/) override {} + void OnTestSuiteEnd(const TestSuite& /*test_suite*/) override {} +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + void OnTestCaseEnd(const TestCase& /*test_case*/) override {} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + void OnEnvironmentsTearDownStart(const UnitTest& /*unit_test*/) override {} + void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) override {} + void OnTestIterationEnd(const UnitTest& /*unit_test*/, + int /*iteration*/) override {} + void OnTestProgramEnd(const UnitTest& /*unit_test*/) override {} +}; + +// TestEventListeners lets users add listeners to track events in Google Test. +class GTEST_API_ TestEventListeners { + public: + TestEventListeners(); + ~TestEventListeners(); + + // Appends an event listener to the end of the list. Google Test assumes + // the ownership of the listener (i.e. it will delete the listener when + // the test program finishes). + void Append(TestEventListener* listener); + + // Removes the given event listener from the list and returns it. It then + // becomes the caller's responsibility to delete the listener. Returns + // NULL if the listener is not found in the list. + TestEventListener* Release(TestEventListener* listener); + + // Returns the standard listener responsible for the default console + // output. Can be removed from the listeners list to shut down default + // console output. Note that removing this object from the listener list + // with Release transfers its ownership to the caller and makes this + // function return NULL the next time. + TestEventListener* default_result_printer() const { + return default_result_printer_; + } + + // Returns the standard listener responsible for the default XML output + // controlled by the --gtest_output=xml flag. Can be removed from the + // listeners list by users who want to shut down the default XML output + // controlled by this flag and substitute it with custom one. Note that + // removing this object from the listener list with Release transfers its + // ownership to the caller and makes this function return NULL the next + // time. + TestEventListener* default_xml_generator() const { + return default_xml_generator_; + } + + private: + friend class TestSuite; + friend class TestInfo; + friend class internal::DefaultGlobalTestPartResultReporter; + friend class internal::NoExecDeathTest; + friend class internal::TestEventListenersAccessor; + friend class internal::UnitTestImpl; + + // Returns repeater that broadcasts the TestEventListener events to all + // subscribers. + TestEventListener* repeater(); + + // Sets the default_result_printer attribute to the provided listener. + // The listener is also added to the listener list and previous + // default_result_printer is removed from it and deleted. The listener can + // also be NULL in which case it will not be added to the list. Does + // nothing if the previous and the current listener objects are the same. + void SetDefaultResultPrinter(TestEventListener* listener); + + // Sets the default_xml_generator attribute to the provided listener. The + // listener is also added to the listener list and previous + // default_xml_generator is removed from it and deleted. The listener can + // also be NULL in which case it will not be added to the list. Does + // nothing if the previous and the current listener objects are the same. + void SetDefaultXmlGenerator(TestEventListener* listener); + + // Controls whether events will be forwarded by the repeater to the + // listeners in the list. + bool EventForwardingEnabled() const; + void SuppressEventForwarding(); + + // The actual list of listeners. + internal::TestEventRepeater* repeater_; + // Listener responsible for the standard result output. + TestEventListener* default_result_printer_; + // Listener responsible for the creation of the XML output file. + TestEventListener* default_xml_generator_; + + // We disallow copying TestEventListeners. + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventListeners); +}; + +// A UnitTest consists of a vector of TestSuites. +// +// This is a singleton class. The only instance of UnitTest is +// created when UnitTest::GetInstance() is first called. This +// instance is never deleted. +// +// UnitTest is not copyable. +// +// This class is thread-safe as long as the methods are called +// according to their specification. +class GTEST_API_ UnitTest { + public: + // Gets the singleton UnitTest object. The first time this method + // is called, a UnitTest object is constructed and returned. + // Consecutive calls will return the same object. + static UnitTest* GetInstance(); + + // Runs all tests in this UnitTest object and prints the result. + // Returns 0 if successful, or 1 otherwise. + // + // This method can only be called from the main thread. + // + // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. + int Run() GTEST_MUST_USE_RESULT_; + + // Returns the working directory when the first TEST() or TEST_F() + // was executed. The UnitTest object owns the string. + const char* original_working_dir() const; + + // Returns the TestSuite object for the test that's currently running, + // or NULL if no test is running. + const TestSuite* current_test_suite() const GTEST_LOCK_EXCLUDED_(mutex_); + +// Legacy API is still available but deprecated +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + const TestCase* current_test_case() const GTEST_LOCK_EXCLUDED_(mutex_); +#endif + + // Returns the TestInfo object for the test that's currently running, + // or NULL if no test is running. + const TestInfo* current_test_info() const + GTEST_LOCK_EXCLUDED_(mutex_); + + // Returns the random seed used at the start of the current test run. + int random_seed() const; + + // Returns the ParameterizedTestSuiteRegistry object used to keep track of + // value-parameterized tests and instantiate and register them. + // + // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. + internal::ParameterizedTestSuiteRegistry& parameterized_test_registry() + GTEST_LOCK_EXCLUDED_(mutex_); + + // Gets the number of successful test suites. + int successful_test_suite_count() const; + + // Gets the number of failed test suites. + int failed_test_suite_count() const; + + // Gets the number of all test suites. + int total_test_suite_count() const; + + // Gets the number of all test suites that contain at least one test + // that should run. + int test_suite_to_run_count() const; + + // Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + int successful_test_case_count() const; + int failed_test_case_count() const; + int total_test_case_count() const; + int test_case_to_run_count() const; +#endif // EMOVE_LEGACY_TEST_CASEAPI + + // Gets the number of successful tests. + int successful_test_count() const; + + // Gets the number of skipped tests. + int skipped_test_count() const; + + // Gets the number of failed tests. + int failed_test_count() const; + + // Gets the number of disabled tests that will be reported in the XML report. + int reportable_disabled_test_count() const; + + // Gets the number of disabled tests. + int disabled_test_count() const; + + // Gets the number of tests to be printed in the XML report. + int reportable_test_count() const; + + // Gets the number of all tests. + int total_test_count() const; + + // Gets the number of tests that should run. + int test_to_run_count() const; + + // Gets the time of the test program start, in ms from the start of the + // UNIX epoch. + TimeInMillis start_timestamp() const; + + // Gets the elapsed time, in milliseconds. + TimeInMillis elapsed_time() const; + + // Returns true iff the unit test passed (i.e. all test suites passed). + bool Passed() const; + + // Returns true iff the unit test failed (i.e. some test suite failed + // or something outside of all tests failed). + bool Failed() const; + + // Gets the i-th test suite among all the test suites. i can range from 0 to + // total_test_suite_count() - 1. If i is not in that range, returns NULL. + const TestSuite* GetTestSuite(int i) const; + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + const TestCase* GetTestCase(int i) const; +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Returns the TestResult containing information on test failures and + // properties logged outside of individual test suites. + const TestResult& ad_hoc_test_result() const; + + // Returns the list of event listeners that can be used to track events + // inside Google Test. + TestEventListeners& listeners(); + + private: + // Registers and returns a global test environment. When a test + // program is run, all global test environments will be set-up in + // the order they were registered. After all tests in the program + // have finished, all global test environments will be torn-down in + // the *reverse* order they were registered. + // + // The UnitTest object takes ownership of the given environment. + // + // This method can only be called from the main thread. + Environment* AddEnvironment(Environment* env); + + // Adds a TestPartResult to the current TestResult object. All + // Google Test assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) + // eventually call this to report their results. The user code + // should use the assertion macros instead of calling this directly. + void AddTestPartResult(TestPartResult::Type result_type, + const char* file_name, + int line_number, + const std::string& message, + const std::string& os_stack_trace) + GTEST_LOCK_EXCLUDED_(mutex_); + + // Adds a TestProperty to the current TestResult object when invoked from + // inside a test, to current TestSuite's ad_hoc_test_result_ when invoked + // from SetUpTestSuite or TearDownTestSuite, or to the global property set + // when invoked elsewhere. If the result already contains a property with + // the same key, the value will be updated. + void RecordProperty(const std::string& key, const std::string& value); + + // Gets the i-th test suite among all the test suites. i can range from 0 to + // total_test_suite_count() - 1. If i is not in that range, returns NULL. + TestSuite* GetMutableTestSuite(int i); + + // Accessors for the implementation object. + internal::UnitTestImpl* impl() { return impl_; } + const internal::UnitTestImpl* impl() const { return impl_; } + + // These classes and functions are friends as they need to access private + // members of UnitTest. + friend class ScopedTrace; + friend class Test; + friend class internal::AssertHelper; + friend class internal::StreamingListenerTest; + friend class internal::UnitTestRecordPropertyTestHelper; + friend Environment* AddGlobalTestEnvironment(Environment* env); + friend internal::UnitTestImpl* internal::GetUnitTestImpl(); + friend void internal::ReportFailureInUnknownLocation( + TestPartResult::Type result_type, + const std::string& message); + + // Creates an empty UnitTest. + UnitTest(); + + // D'tor + virtual ~UnitTest(); + + // Pushes a trace defined by SCOPED_TRACE() on to the per-thread + // Google Test trace stack. + void PushGTestTrace(const internal::TraceInfo& trace) + GTEST_LOCK_EXCLUDED_(mutex_); + + // Pops a trace from the per-thread Google Test trace stack. + void PopGTestTrace() + GTEST_LOCK_EXCLUDED_(mutex_); + + // Protects mutable state in *impl_. This is mutable as some const + // methods need to lock it too. + mutable internal::Mutex mutex_; + + // Opaque implementation object. This field is never changed once + // the object is constructed. We don't mark it as const here, as + // doing so will cause a warning in the constructor of UnitTest. + // Mutable state in *impl_ is protected by mutex_. + internal::UnitTestImpl* impl_; + + // We disallow copying UnitTest. + GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTest); +}; + +// A convenient wrapper for adding an environment for the test +// program. +// +// You should call this before RUN_ALL_TESTS() is called, probably in +// main(). If you use gtest_main, you need to call this before main() +// starts for it to take effect. For example, you can define a global +// variable like this: +// +// testing::Environment* const foo_env = +// testing::AddGlobalTestEnvironment(new FooEnvironment); +// +// However, we strongly recommend you to write your own main() and +// call AddGlobalTestEnvironment() there, as relying on initialization +// of global variables makes the code harder to read and may cause +// problems when you register multiple environments from different +// translation units and the environments have dependencies among them +// (remember that the compiler doesn't guarantee the order in which +// global variables from different translation units are initialized). +inline Environment* AddGlobalTestEnvironment(Environment* env) { + return UnitTest::GetInstance()->AddEnvironment(env); +} + +// Initializes Google Test. This must be called before calling +// RUN_ALL_TESTS(). In particular, it parses a command line for the +// flags that Google Test recognizes. Whenever a Google Test flag is +// seen, it is removed from argv, and *argc is decremented. +// +// No value is returned. Instead, the Google Test flag variables are +// updated. +// +// Calling the function for the second time has no user-visible effect. +GTEST_API_ void InitGoogleTest(int* argc, char** argv); + +// This overloaded version can be used in Windows programs compiled in +// UNICODE mode. +GTEST_API_ void InitGoogleTest(int* argc, wchar_t** argv); + +// This overloaded version can be used on Arduino/embedded platforms where +// there is no argc/argv. +GTEST_API_ void InitGoogleTest(); + +namespace internal { + +// Separate the error generating code from the code path to reduce the stack +// frame size of CmpHelperEQ. This helps reduce the overhead of some sanitizers +// when calling EXPECT_* in a tight loop. +template +AssertionResult CmpHelperEQFailure(const char* lhs_expression, + const char* rhs_expression, + const T1& lhs, const T2& rhs) { + return EqFailure(lhs_expression, + rhs_expression, + FormatForComparisonFailureMessage(lhs, rhs), + FormatForComparisonFailureMessage(rhs, lhs), + false); +} + +// This block of code defines operator==/!= +// to block lexical scope lookup. +// It prevents using invalid operator==/!= defined at namespace scope. +struct faketype {}; +inline bool operator==(faketype, faketype) { return true; } +inline bool operator!=(faketype, faketype) { return false; } + +// The helper function for {ASSERT|EXPECT}_EQ. +template +AssertionResult CmpHelperEQ(const char* lhs_expression, + const char* rhs_expression, + const T1& lhs, + const T2& rhs) { + if (lhs == rhs) { + return AssertionSuccess(); + } + + return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs); +} + +// With this overloaded version, we allow anonymous enums to be used +// in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous enums +// can be implicitly cast to BiggestInt. +GTEST_API_ AssertionResult CmpHelperEQ(const char* lhs_expression, + const char* rhs_expression, + BiggestInt lhs, + BiggestInt rhs); + +class EqHelper { + public: + // This templatized version is for the general case. + template < + typename T1, typename T2, + // Disable this overload for cases where one argument is a pointer + // and the other is the null pointer constant. + typename std::enable_if::value || + !std::is_pointer::value>::type* = nullptr> + static AssertionResult Compare(const char* lhs_expression, + const char* rhs_expression, const T1& lhs, + const T2& rhs) { + return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs); + } + + // With this overloaded version, we allow anonymous enums to be used + // in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous + // enums can be implicitly cast to BiggestInt. + // + // Even though its body looks the same as the above version, we + // cannot merge the two, as it will make anonymous enums unhappy. + static AssertionResult Compare(const char* lhs_expression, + const char* rhs_expression, + BiggestInt lhs, + BiggestInt rhs) { + return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs); + } + + template + static AssertionResult Compare( + const char* lhs_expression, const char* rhs_expression, + // Handle cases where '0' is used as a null pointer literal. + std::nullptr_t /* lhs */, T* rhs) { + // We already know that 'lhs' is a null pointer. + return CmpHelperEQ(lhs_expression, rhs_expression, static_cast(nullptr), + rhs); + } +}; + +// Separate the error generating code from the code path to reduce the stack +// frame size of CmpHelperOP. This helps reduce the overhead of some sanitizers +// when calling EXPECT_OP in a tight loop. +template +AssertionResult CmpHelperOpFailure(const char* expr1, const char* expr2, + const T1& val1, const T2& val2, + const char* op) { + return AssertionFailure() + << "Expected: (" << expr1 << ") " << op << " (" << expr2 + << "), actual: " << FormatForComparisonFailureMessage(val1, val2) + << " vs " << FormatForComparisonFailureMessage(val2, val1); +} + +// A macro for implementing the helper functions needed to implement +// ASSERT_?? and EXPECT_??. It is here just to avoid copy-and-paste +// of similar code. +// +// For each templatized helper function, we also define an overloaded +// version for BiggestInt in order to reduce code bloat and allow +// anonymous enums to be used with {ASSERT|EXPECT}_?? when compiled +// with gcc 4. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. + +#define GTEST_IMPL_CMP_HELPER_(op_name, op)\ +template \ +AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \ + const T1& val1, const T2& val2) {\ + if (val1 op val2) {\ + return AssertionSuccess();\ + } else {\ + return CmpHelperOpFailure(expr1, expr2, val1, val2, #op);\ + }\ +}\ +GTEST_API_ AssertionResult CmpHelper##op_name(\ + const char* expr1, const char* expr2, BiggestInt val1, BiggestInt val2) + +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. + +// Implements the helper function for {ASSERT|EXPECT}_NE +GTEST_IMPL_CMP_HELPER_(NE, !=); +// Implements the helper function for {ASSERT|EXPECT}_LE +GTEST_IMPL_CMP_HELPER_(LE, <=); +// Implements the helper function for {ASSERT|EXPECT}_LT +GTEST_IMPL_CMP_HELPER_(LT, <); +// Implements the helper function for {ASSERT|EXPECT}_GE +GTEST_IMPL_CMP_HELPER_(GE, >=); +// Implements the helper function for {ASSERT|EXPECT}_GT +GTEST_IMPL_CMP_HELPER_(GT, >); + +#undef GTEST_IMPL_CMP_HELPER_ + +// The helper function for {ASSERT|EXPECT}_STREQ. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult CmpHelperSTREQ(const char* s1_expression, + const char* s2_expression, + const char* s1, + const char* s2); + +// The helper function for {ASSERT|EXPECT}_STRCASEEQ. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult CmpHelperSTRCASEEQ(const char* s1_expression, + const char* s2_expression, + const char* s1, + const char* s2); + +// The helper function for {ASSERT|EXPECT}_STRNE. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult CmpHelperSTRNE(const char* s1_expression, + const char* s2_expression, + const char* s1, + const char* s2); + +// The helper function for {ASSERT|EXPECT}_STRCASENE. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult CmpHelperSTRCASENE(const char* s1_expression, + const char* s2_expression, + const char* s1, + const char* s2); + + +// Helper function for *_STREQ on wide strings. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult CmpHelperSTREQ(const char* s1_expression, + const char* s2_expression, + const wchar_t* s1, + const wchar_t* s2); + +// Helper function for *_STRNE on wide strings. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult CmpHelperSTRNE(const char* s1_expression, + const char* s2_expression, + const wchar_t* s1, + const wchar_t* s2); + +} // namespace internal + +// IsSubstring() and IsNotSubstring() are intended to be used as the +// first argument to {EXPECT,ASSERT}_PRED_FORMAT2(), not by +// themselves. They check whether needle is a substring of haystack +// (NULL is considered a substring of itself only), and return an +// appropriate error message when they fail. +// +// The {needle,haystack}_expr arguments are the stringified +// expressions that generated the two real arguments. +GTEST_API_ AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const char* needle, const char* haystack); +GTEST_API_ AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const wchar_t* needle, const wchar_t* haystack); +GTEST_API_ AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const char* needle, const char* haystack); +GTEST_API_ AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const wchar_t* needle, const wchar_t* haystack); +GTEST_API_ AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::string& needle, const ::std::string& haystack); +GTEST_API_ AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::string& needle, const ::std::string& haystack); + +#if GTEST_HAS_STD_WSTRING +GTEST_API_ AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::wstring& needle, const ::std::wstring& haystack); +GTEST_API_ AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::wstring& needle, const ::std::wstring& haystack); +#endif // GTEST_HAS_STD_WSTRING + +namespace internal { + +// Helper template function for comparing floating-points. +// +// Template parameter: +// +// RawType: the raw floating-point type (either float or double) +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +template +AssertionResult CmpHelperFloatingPointEQ(const char* lhs_expression, + const char* rhs_expression, + RawType lhs_value, + RawType rhs_value) { + const FloatingPoint lhs(lhs_value), rhs(rhs_value); + + if (lhs.AlmostEquals(rhs)) { + return AssertionSuccess(); + } + + ::std::stringstream lhs_ss; + lhs_ss << std::setprecision(std::numeric_limits::digits10 + 2) + << lhs_value; + + ::std::stringstream rhs_ss; + rhs_ss << std::setprecision(std::numeric_limits::digits10 + 2) + << rhs_value; + + return EqFailure(lhs_expression, + rhs_expression, + StringStreamToString(&lhs_ss), + StringStreamToString(&rhs_ss), + false); +} + +// Helper function for implementing ASSERT_NEAR. +// +// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM. +GTEST_API_ AssertionResult DoubleNearPredFormat(const char* expr1, + const char* expr2, + const char* abs_error_expr, + double val1, + double val2, + double abs_error); + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// A class that enables one to stream messages to assertion macros +class GTEST_API_ AssertHelper { + public: + // Constructor. + AssertHelper(TestPartResult::Type type, + const char* file, + int line, + const char* message); + ~AssertHelper(); + + // Message assignment is a semantic trick to enable assertion + // streaming; see the GTEST_MESSAGE_ macro below. + void operator=(const Message& message) const; + + private: + // We put our data in a struct so that the size of the AssertHelper class can + // be as small as possible. This is important because gcc is incapable of + // re-using stack space even for temporary variables, so every EXPECT_EQ + // reserves stack space for another AssertHelper. + struct AssertHelperData { + AssertHelperData(TestPartResult::Type t, + const char* srcfile, + int line_num, + const char* msg) + : type(t), file(srcfile), line(line_num), message(msg) { } + + TestPartResult::Type const type; + const char* const file; + int const line; + std::string const message; + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelperData); + }; + + AssertHelperData* const data_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelper); +}; + +enum GTestColor { COLOR_DEFAULT, COLOR_RED, COLOR_GREEN, COLOR_YELLOW }; + +GTEST_API_ GTEST_ATTRIBUTE_PRINTF_(2, 3) void ColoredPrintf(GTestColor color, + const char* fmt, + ...); + +} // namespace internal + +// The pure interface class that all value-parameterized tests inherit from. +// A value-parameterized class must inherit from both ::testing::Test and +// ::testing::WithParamInterface. In most cases that just means inheriting +// from ::testing::TestWithParam, but more complicated test hierarchies +// may need to inherit from Test and WithParamInterface at different levels. +// +// This interface has support for accessing the test parameter value via +// the GetParam() method. +// +// Use it with one of the parameter generator defining functions, like Range(), +// Values(), ValuesIn(), Bool(), and Combine(). +// +// class FooTest : public ::testing::TestWithParam { +// protected: +// FooTest() { +// // Can use GetParam() here. +// } +// ~FooTest() override { +// // Can use GetParam() here. +// } +// void SetUp() override { +// // Can use GetParam() here. +// } +// void TearDown override { +// // Can use GetParam() here. +// } +// }; +// TEST_P(FooTest, DoesBar) { +// // Can use GetParam() method here. +// Foo foo; +// ASSERT_TRUE(foo.DoesBar(GetParam())); +// } +// INSTANTIATE_TEST_SUITE_P(OneToTenRange, FooTest, ::testing::Range(1, 10)); + +template +class WithParamInterface { + public: + typedef T ParamType; + virtual ~WithParamInterface() {} + + // The current parameter value. Is also available in the test fixture's + // constructor. + static const ParamType& GetParam() { + GTEST_CHECK_(parameter_ != nullptr) + << "GetParam() can only be called inside a value-parameterized test " + << "-- did you intend to write TEST_P instead of TEST_F?"; + return *parameter_; + } + + private: + // Sets parameter value. The caller is responsible for making sure the value + // remains alive and unchanged throughout the current test. + static void SetParam(const ParamType* parameter) { + parameter_ = parameter; + } + + // Static value used for accessing parameter during a test lifetime. + static const ParamType* parameter_; + + // TestClass must be a subclass of WithParamInterface and Test. + template friend class internal::ParameterizedTestFactory; +}; + +template +const T* WithParamInterface::parameter_ = nullptr; + +// Most value-parameterized classes can ignore the existence of +// WithParamInterface, and can just inherit from ::testing::TestWithParam. + +template +class TestWithParam : public Test, public WithParamInterface { +}; + +// Macros for indicating success/failure in test code. + +// Skips test in runtime. +// Skipping test aborts current function. +// Skipped tests are neither successful nor failed. +#define GTEST_SKIP() GTEST_SKIP_("Skipped") + +// ADD_FAILURE unconditionally adds a failure to the current test. +// SUCCEED generates a success - it doesn't automatically make the +// current test successful, as a test is only successful when it has +// no failure. +// +// EXPECT_* verifies that a certain condition is satisfied. If not, +// it behaves like ADD_FAILURE. In particular: +// +// EXPECT_TRUE verifies that a Boolean condition is true. +// EXPECT_FALSE verifies that a Boolean condition is false. +// +// FAIL and ASSERT_* are similar to ADD_FAILURE and EXPECT_*, except +// that they will also abort the current function on failure. People +// usually want the fail-fast behavior of FAIL and ASSERT_*, but those +// writing data-driven tests often find themselves using ADD_FAILURE +// and EXPECT_* more. + +// Generates a nonfatal failure with a generic message. +#define ADD_FAILURE() GTEST_NONFATAL_FAILURE_("Failed") + +// Generates a nonfatal failure at the given source file location with +// a generic message. +#define ADD_FAILURE_AT(file, line) \ + GTEST_MESSAGE_AT_(file, line, "Failed", \ + ::testing::TestPartResult::kNonFatalFailure) + +// Generates a fatal failure with a generic message. +#define GTEST_FAIL() GTEST_FATAL_FAILURE_("Failed") + +// Define this macro to 1 to omit the definition of FAIL(), which is a +// generic name and clashes with some other libraries. +#if !GTEST_DONT_DEFINE_FAIL +# define FAIL() GTEST_FAIL() +#endif + +// Generates a success with a generic message. +#define GTEST_SUCCEED() GTEST_SUCCESS_("Succeeded") + +// Define this macro to 1 to omit the definition of SUCCEED(), which +// is a generic name and clashes with some other libraries. +#if !GTEST_DONT_DEFINE_SUCCEED +# define SUCCEED() GTEST_SUCCEED() +#endif + +// Macros for testing exceptions. +// +// * {ASSERT|EXPECT}_THROW(statement, expected_exception): +// Tests that the statement throws the expected exception. +// * {ASSERT|EXPECT}_NO_THROW(statement): +// Tests that the statement doesn't throw any exception. +// * {ASSERT|EXPECT}_ANY_THROW(statement): +// Tests that the statement throws an exception. + +#define EXPECT_THROW(statement, expected_exception) \ + GTEST_TEST_THROW_(statement, expected_exception, GTEST_NONFATAL_FAILURE_) +#define EXPECT_NO_THROW(statement) \ + GTEST_TEST_NO_THROW_(statement, GTEST_NONFATAL_FAILURE_) +#define EXPECT_ANY_THROW(statement) \ + GTEST_TEST_ANY_THROW_(statement, GTEST_NONFATAL_FAILURE_) +#define ASSERT_THROW(statement, expected_exception) \ + GTEST_TEST_THROW_(statement, expected_exception, GTEST_FATAL_FAILURE_) +#define ASSERT_NO_THROW(statement) \ + GTEST_TEST_NO_THROW_(statement, GTEST_FATAL_FAILURE_) +#define ASSERT_ANY_THROW(statement) \ + GTEST_TEST_ANY_THROW_(statement, GTEST_FATAL_FAILURE_) + +// Boolean assertions. Condition can be either a Boolean expression or an +// AssertionResult. For more information on how to use AssertionResult with +// these macros see comments on that class. +#define EXPECT_TRUE(condition) \ + GTEST_TEST_BOOLEAN_(condition, #condition, false, true, \ + GTEST_NONFATAL_FAILURE_) +#define EXPECT_FALSE(condition) \ + GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \ + GTEST_NONFATAL_FAILURE_) +#define ASSERT_TRUE(condition) \ + GTEST_TEST_BOOLEAN_(condition, #condition, false, true, \ + GTEST_FATAL_FAILURE_) +#define ASSERT_FALSE(condition) \ + GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \ + GTEST_FATAL_FAILURE_) + +// Macros for testing equalities and inequalities. +// +// * {ASSERT|EXPECT}_EQ(v1, v2): Tests that v1 == v2 +// * {ASSERT|EXPECT}_NE(v1, v2): Tests that v1 != v2 +// * {ASSERT|EXPECT}_LT(v1, v2): Tests that v1 < v2 +// * {ASSERT|EXPECT}_LE(v1, v2): Tests that v1 <= v2 +// * {ASSERT|EXPECT}_GT(v1, v2): Tests that v1 > v2 +// * {ASSERT|EXPECT}_GE(v1, v2): Tests that v1 >= v2 +// +// When they are not, Google Test prints both the tested expressions and +// their actual values. The values must be compatible built-in types, +// or you will get a compiler error. By "compatible" we mean that the +// values can be compared by the respective operator. +// +// Note: +// +// 1. It is possible to make a user-defined type work with +// {ASSERT|EXPECT}_??(), but that requires overloading the +// comparison operators and is thus discouraged by the Google C++ +// Usage Guide. Therefore, you are advised to use the +// {ASSERT|EXPECT}_TRUE() macro to assert that two objects are +// equal. +// +// 2. The {ASSERT|EXPECT}_??() macros do pointer comparisons on +// pointers (in particular, C strings). Therefore, if you use it +// with two C strings, you are testing how their locations in memory +// are related, not how their content is related. To compare two C +// strings by content, use {ASSERT|EXPECT}_STR*(). +// +// 3. {ASSERT|EXPECT}_EQ(v1, v2) is preferred to +// {ASSERT|EXPECT}_TRUE(v1 == v2), as the former tells you +// what the actual value is when it fails, and similarly for the +// other comparisons. +// +// 4. Do not depend on the order in which {ASSERT|EXPECT}_??() +// evaluate their arguments, which is undefined. +// +// 5. These macros evaluate their arguments exactly once. +// +// Examples: +// +// EXPECT_NE(Foo(), 5); +// EXPECT_EQ(a_pointer, NULL); +// ASSERT_LT(i, array_size); +// ASSERT_GT(records.size(), 0) << "There is no record left."; + +#define EXPECT_EQ(val1, val2) \ + EXPECT_PRED_FORMAT2(::testing::internal::EqHelper::Compare, val1, val2) +#define EXPECT_NE(val1, val2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2) +#define EXPECT_LE(val1, val2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2) +#define EXPECT_LT(val1, val2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2) +#define EXPECT_GE(val1, val2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2) +#define EXPECT_GT(val1, val2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2) + +#define GTEST_ASSERT_EQ(val1, val2) \ + ASSERT_PRED_FORMAT2(::testing::internal::EqHelper::Compare, val1, val2) +#define GTEST_ASSERT_NE(val1, val2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2) +#define GTEST_ASSERT_LE(val1, val2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2) +#define GTEST_ASSERT_LT(val1, val2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2) +#define GTEST_ASSERT_GE(val1, val2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2) +#define GTEST_ASSERT_GT(val1, val2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2) + +// Define macro GTEST_DONT_DEFINE_ASSERT_XY to 1 to omit the definition of +// ASSERT_XY(), which clashes with some users' own code. + +#if !GTEST_DONT_DEFINE_ASSERT_EQ +# define ASSERT_EQ(val1, val2) GTEST_ASSERT_EQ(val1, val2) +#endif + +#if !GTEST_DONT_DEFINE_ASSERT_NE +# define ASSERT_NE(val1, val2) GTEST_ASSERT_NE(val1, val2) +#endif + +#if !GTEST_DONT_DEFINE_ASSERT_LE +# define ASSERT_LE(val1, val2) GTEST_ASSERT_LE(val1, val2) +#endif + +#if !GTEST_DONT_DEFINE_ASSERT_LT +# define ASSERT_LT(val1, val2) GTEST_ASSERT_LT(val1, val2) +#endif + +#if !GTEST_DONT_DEFINE_ASSERT_GE +# define ASSERT_GE(val1, val2) GTEST_ASSERT_GE(val1, val2) +#endif + +#if !GTEST_DONT_DEFINE_ASSERT_GT +# define ASSERT_GT(val1, val2) GTEST_ASSERT_GT(val1, val2) +#endif + +// C-string Comparisons. All tests treat NULL and any non-NULL string +// as different. Two NULLs are equal. +// +// * {ASSERT|EXPECT}_STREQ(s1, s2): Tests that s1 == s2 +// * {ASSERT|EXPECT}_STRNE(s1, s2): Tests that s1 != s2 +// * {ASSERT|EXPECT}_STRCASEEQ(s1, s2): Tests that s1 == s2, ignoring case +// * {ASSERT|EXPECT}_STRCASENE(s1, s2): Tests that s1 != s2, ignoring case +// +// For wide or narrow string objects, you can use the +// {ASSERT|EXPECT}_??() macros. +// +// Don't depend on the order in which the arguments are evaluated, +// which is undefined. +// +// These macros evaluate their arguments exactly once. + +#define EXPECT_STREQ(s1, s2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, s1, s2) +#define EXPECT_STRNE(s1, s2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2) +#define EXPECT_STRCASEEQ(s1, s2) \ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, s1, s2) +#define EXPECT_STRCASENE(s1, s2)\ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2) + +#define ASSERT_STREQ(s1, s2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, s1, s2) +#define ASSERT_STRNE(s1, s2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2) +#define ASSERT_STRCASEEQ(s1, s2) \ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, s1, s2) +#define ASSERT_STRCASENE(s1, s2)\ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2) + +// Macros for comparing floating-point numbers. +// +// * {ASSERT|EXPECT}_FLOAT_EQ(val1, val2): +// Tests that two float values are almost equal. +// * {ASSERT|EXPECT}_DOUBLE_EQ(val1, val2): +// Tests that two double values are almost equal. +// * {ASSERT|EXPECT}_NEAR(v1, v2, abs_error): +// Tests that v1 and v2 are within the given distance to each other. +// +// Google Test uses ULP-based comparison to automatically pick a default +// error bound that is appropriate for the operands. See the +// FloatingPoint template class in gtest-internal.h if you are +// interested in the implementation details. + +#define EXPECT_FLOAT_EQ(val1, val2)\ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ + val1, val2) + +#define EXPECT_DOUBLE_EQ(val1, val2)\ + EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ + val1, val2) + +#define ASSERT_FLOAT_EQ(val1, val2)\ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ + val1, val2) + +#define ASSERT_DOUBLE_EQ(val1, val2)\ + ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ, \ + val1, val2) + +#define EXPECT_NEAR(val1, val2, abs_error)\ + EXPECT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \ + val1, val2, abs_error) + +#define ASSERT_NEAR(val1, val2, abs_error)\ + ASSERT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \ + val1, val2, abs_error) + +// These predicate format functions work on floating-point values, and +// can be used in {ASSERT|EXPECT}_PRED_FORMAT2*(), e.g. +// +// EXPECT_PRED_FORMAT2(testing::DoubleLE, Foo(), 5.0); + +// Asserts that val1 is less than, or almost equal to, val2. Fails +// otherwise. In particular, it fails if either val1 or val2 is NaN. +GTEST_API_ AssertionResult FloatLE(const char* expr1, const char* expr2, + float val1, float val2); +GTEST_API_ AssertionResult DoubleLE(const char* expr1, const char* expr2, + double val1, double val2); + + +#if GTEST_OS_WINDOWS + +// Macros that test for HRESULT failure and success, these are only useful +// on Windows, and rely on Windows SDK macros and APIs to compile. +// +// * {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}(expr) +// +// When expr unexpectedly fails or succeeds, Google Test prints the +// expected result and the actual result with both a human-readable +// string representation of the error, if available, as well as the +// hex result code. +# define EXPECT_HRESULT_SUCCEEDED(expr) \ + EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr)) + +# define ASSERT_HRESULT_SUCCEEDED(expr) \ + ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr)) + +# define EXPECT_HRESULT_FAILED(expr) \ + EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr)) + +# define ASSERT_HRESULT_FAILED(expr) \ + ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr)) + +#endif // GTEST_OS_WINDOWS + +// Macros that execute statement and check that it doesn't generate new fatal +// failures in the current thread. +// +// * {ASSERT|EXPECT}_NO_FATAL_FAILURE(statement); +// +// Examples: +// +// EXPECT_NO_FATAL_FAILURE(Process()); +// ASSERT_NO_FATAL_FAILURE(Process()) << "Process() failed"; +// +#define ASSERT_NO_FATAL_FAILURE(statement) \ + GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_FATAL_FAILURE_) +#define EXPECT_NO_FATAL_FAILURE(statement) \ + GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_NONFATAL_FAILURE_) + +// Causes a trace (including the given source file path and line number, +// and the given message) to be included in every test failure message generated +// by code in the scope of the lifetime of an instance of this class. The effect +// is undone with the destruction of the instance. +// +// The message argument can be anything streamable to std::ostream. +// +// Example: +// testing::ScopedTrace trace("file.cc", 123, "message"); +// +class GTEST_API_ ScopedTrace { + public: + // The c'tor pushes the given source file location and message onto + // a trace stack maintained by Google Test. + + // Template version. Uses Message() to convert the values into strings. + // Slow, but flexible. + template + ScopedTrace(const char* file, int line, const T& message) { + PushTrace(file, line, (Message() << message).GetString()); + } + + // Optimize for some known types. + ScopedTrace(const char* file, int line, const char* message) { + PushTrace(file, line, message ? message : "(null)"); + } + + ScopedTrace(const char* file, int line, const std::string& message) { + PushTrace(file, line, message); + } + + // The d'tor pops the info pushed by the c'tor. + // + // Note that the d'tor is not virtual in order to be efficient. + // Don't inherit from ScopedTrace! + ~ScopedTrace(); + + private: + void PushTrace(const char* file, int line, std::string message); + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedTrace); +} GTEST_ATTRIBUTE_UNUSED_; // A ScopedTrace object does its job in its + // c'tor and d'tor. Therefore it doesn't + // need to be used otherwise. + +// Causes a trace (including the source file path, the current line +// number, and the given message) to be included in every test failure +// message generated by code in the current scope. The effect is +// undone when the control leaves the current scope. +// +// The message argument can be anything streamable to std::ostream. +// +// In the implementation, we include the current line number as part +// of the dummy variable name, thus allowing multiple SCOPED_TRACE()s +// to appear in the same block - as long as they are on different +// lines. +// +// Assuming that each thread maintains its own stack of traces. +// Therefore, a SCOPED_TRACE() would (correctly) only affect the +// assertions in its own thread. +#define SCOPED_TRACE(message) \ + ::testing::ScopedTrace GTEST_CONCAT_TOKEN_(gtest_trace_, __LINE__)(\ + __FILE__, __LINE__, (message)) + + +// Compile-time assertion for type equality. +// StaticAssertTypeEq() compiles iff type1 and type2 are +// the same type. The value it returns is not interesting. +// +// Instead of making StaticAssertTypeEq a class template, we make it a +// function template that invokes a helper class template. This +// prevents a user from misusing StaticAssertTypeEq by +// defining objects of that type. +// +// CAVEAT: +// +// When used inside a method of a class template, +// StaticAssertTypeEq() is effective ONLY IF the method is +// instantiated. For example, given: +// +// template class Foo { +// public: +// void Bar() { testing::StaticAssertTypeEq(); } +// }; +// +// the code: +// +// void Test1() { Foo foo; } +// +// will NOT generate a compiler error, as Foo::Bar() is never +// actually instantiated. Instead, you need: +// +// void Test2() { Foo foo; foo.Bar(); } +// +// to cause a compiler error. +template +bool StaticAssertTypeEq() { + (void)internal::StaticAssertTypeEqHelper(); + return true; +} + +// Defines a test. +// +// The first parameter is the name of the test suite, and the second +// parameter is the name of the test within the test suite. +// +// The convention is to end the test suite name with "Test". For +// example, a test suite for the Foo class can be named FooTest. +// +// Test code should appear between braces after an invocation of +// this macro. Example: +// +// TEST(FooTest, InitializesCorrectly) { +// Foo foo; +// EXPECT_TRUE(foo.StatusIsOK()); +// } + +// Note that we call GetTestTypeId() instead of GetTypeId< +// ::testing::Test>() here to get the type ID of testing::Test. This +// is to work around a suspected linker bug when using Google Test as +// a framework on Mac OS X. The bug causes GetTypeId< +// ::testing::Test>() to return different values depending on whether +// the call is from the Google Test framework itself or from user test +// code. GetTestTypeId() is guaranteed to always return the same +// value, as it always calls GetTypeId<>() from the Google Test +// framework. +#define GTEST_TEST(test_suite_name, test_name) \ + GTEST_TEST_(test_suite_name, test_name, ::testing::Test, \ + ::testing::internal::GetTestTypeId()) + +// Define this macro to 1 to omit the definition of TEST(), which +// is a generic name and clashes with some other libraries. +#if !GTEST_DONT_DEFINE_TEST +#define TEST(test_suite_name, test_name) GTEST_TEST(test_suite_name, test_name) +#endif + +// Defines a test that uses a test fixture. +// +// The first parameter is the name of the test fixture class, which +// also doubles as the test suite name. The second parameter is the +// name of the test within the test suite. +// +// A test fixture class must be declared earlier. The user should put +// the test code between braces after using this macro. Example: +// +// class FooTest : public testing::Test { +// protected: +// void SetUp() override { b_.AddElement(3); } +// +// Foo a_; +// Foo b_; +// }; +// +// TEST_F(FooTest, InitializesCorrectly) { +// EXPECT_TRUE(a_.StatusIsOK()); +// } +// +// TEST_F(FooTest, ReturnsElementCountCorrectly) { +// EXPECT_EQ(a_.size(), 0); +// EXPECT_EQ(b_.size(), 1); +// } +// +// GOOGLETEST_CM0011 DO NOT DELETE +#define TEST_F(test_fixture, test_name)\ + GTEST_TEST_(test_fixture, test_name, test_fixture, \ + ::testing::internal::GetTypeId()) + +// Returns a path to temporary directory. +// Tries to determine an appropriate directory for the platform. +GTEST_API_ std::string TempDir(); + +#ifdef _MSC_VER +# pragma warning(pop) +#endif + +// Dynamically registers a test with the framework. +// +// This is an advanced API only to be used when the `TEST` macros are +// insufficient. The macros should be preferred when possible, as they avoid +// most of the complexity of calling this function. +// +// The `factory` argument is a factory callable (move-constructible) object or +// function pointer that creates a new instance of the Test object. It +// handles ownership to the caller. The signature of the callable is +// `Fixture*()`, where `Fixture` is the test fixture class for the test. All +// tests registered with the same `test_suite_name` must return the same +// fixture type. This is checked at runtime. +// +// The framework will infer the fixture class from the factory and will call +// the `SetUpTestSuite` and `TearDownTestSuite` for it. +// +// Must be called before `RUN_ALL_TESTS()` is invoked, otherwise behavior is +// undefined. +// +// Use case example: +// +// class MyFixture : public ::testing::Test { +// public: +// // All of these optional, just like in regular macro usage. +// static void SetUpTestSuite() { ... } +// static void TearDownTestSuite() { ... } +// void SetUp() override { ... } +// void TearDown() override { ... } +// }; +// +// class MyTest : public MyFixture { +// public: +// explicit MyTest(int data) : data_(data) {} +// void TestBody() override { ... } +// +// private: +// int data_; +// }; +// +// void RegisterMyTests(const std::vector& values) { +// for (int v : values) { +// ::testing::RegisterTest( +// "MyFixture", ("Test" + std::to_string(v)).c_str(), nullptr, +// std::to_string(v).c_str(), +// __FILE__, __LINE__, +// // Important to use the fixture type as the return type here. +// [=]() -> MyFixture* { return new MyTest(v); }); +// } +// } +// ... +// int main(int argc, char** argv) { +// std::vector values_to_test = LoadValuesFromConfig(); +// RegisterMyTests(values_to_test); +// ... +// return RUN_ALL_TESTS(); +// } +// +template +TestInfo* RegisterTest(const char* test_suite_name, const char* test_name, + const char* type_param, const char* value_param, + const char* file, int line, Factory factory) { + using TestT = typename std::remove_pointer::type; + + class FactoryImpl : public internal::TestFactoryBase { + public: + explicit FactoryImpl(Factory f) : factory_(std::move(f)) {} + Test* CreateTest() override { return factory_(); } + + private: + Factory factory_; + }; + + return internal::MakeAndRegisterTestInfo( + test_suite_name, test_name, type_param, value_param, + internal::CodeLocation(file, line), internal::GetTypeId(), + internal::SuiteApiResolver::GetSetUpCaseOrSuite(file, line), + internal::SuiteApiResolver::GetTearDownCaseOrSuite(file, line), + new FactoryImpl{std::move(factory)}); +} + +} // namespace testing + +// Use this function in main() to run all tests. It returns 0 if all +// tests are successful, or 1 otherwise. +// +// RUN_ALL_TESTS() should be invoked after the command line has been +// parsed by InitGoogleTest(). +// +// This function was formerly a macro; thus, it is in the global +// namespace and has an all-caps name. +int RUN_ALL_TESTS() GTEST_MUST_USE_RESULT_; + +inline int RUN_ALL_TESTS() { + return ::testing::UnitTest::GetInstance()->Run(); +} + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +#endif // GTEST_INCLUDE_GTEST_GTEST_H_ diff --git a/deps/googletest/include/gtest/gtest_pred_impl.h b/deps/googletest/include/gtest/gtest_pred_impl.h new file mode 100644 index 000000000..d514255c7 --- /dev/null +++ b/deps/googletest/include/gtest/gtest_pred_impl.h @@ -0,0 +1,359 @@ +// Copyright 2006, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// This file is AUTOMATICALLY GENERATED on 01/02/2019 by command +// 'gen_gtest_pred_impl.py 5'. DO NOT EDIT BY HAND! +// +// Implements a family of generic predicate assertion macros. +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_ +#define GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_ + +#include "gtest/gtest.h" + +namespace testing { + +// This header implements a family of generic predicate assertion +// macros: +// +// ASSERT_PRED_FORMAT1(pred_format, v1) +// ASSERT_PRED_FORMAT2(pred_format, v1, v2) +// ... +// +// where pred_format is a function or functor that takes n (in the +// case of ASSERT_PRED_FORMATn) values and their source expression +// text, and returns a testing::AssertionResult. See the definition +// of ASSERT_EQ in gtest.h for an example. +// +// If you don't care about formatting, you can use the more +// restrictive version: +// +// ASSERT_PRED1(pred, v1) +// ASSERT_PRED2(pred, v1, v2) +// ... +// +// where pred is an n-ary function or functor that returns bool, +// and the values v1, v2, ..., must support the << operator for +// streaming to std::ostream. +// +// We also define the EXPECT_* variations. +// +// For now we only support predicates whose arity is at most 5. +// Please email googletestframework@googlegroups.com if you need +// support for higher arities. + +// GTEST_ASSERT_ is the basic statement to which all of the assertions +// in this file reduce. Don't use this in your code. + +#define GTEST_ASSERT_(expression, on_failure) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (const ::testing::AssertionResult gtest_ar = (expression)) \ + ; \ + else \ + on_failure(gtest_ar.failure_message()) + + +// Helper function for implementing {EXPECT|ASSERT}_PRED1. Don't use +// this in your code. +template +AssertionResult AssertPred1Helper(const char* pred_text, + const char* e1, + Pred pred, + const T1& v1) { + if (pred(v1)) return AssertionSuccess(); + + return AssertionFailure() + << pred_text << "(" << e1 << ") evaluates to false, where" + << "\n" + << e1 << " evaluates to " << ::testing::PrintToString(v1); +} + +// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT1. +// Don't use this in your code. +#define GTEST_PRED_FORMAT1_(pred_format, v1, on_failure)\ + GTEST_ASSERT_(pred_format(#v1, v1), \ + on_failure) + +// Internal macro for implementing {EXPECT|ASSERT}_PRED1. Don't use +// this in your code. +#define GTEST_PRED1_(pred, v1, on_failure)\ + GTEST_ASSERT_(::testing::AssertPred1Helper(#pred, \ + #v1, \ + pred, \ + v1), on_failure) + +// Unary predicate assertion macros. +#define EXPECT_PRED_FORMAT1(pred_format, v1) \ + GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_NONFATAL_FAILURE_) +#define EXPECT_PRED1(pred, v1) \ + GTEST_PRED1_(pred, v1, GTEST_NONFATAL_FAILURE_) +#define ASSERT_PRED_FORMAT1(pred_format, v1) \ + GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_FATAL_FAILURE_) +#define ASSERT_PRED1(pred, v1) \ + GTEST_PRED1_(pred, v1, GTEST_FATAL_FAILURE_) + + + +// Helper function for implementing {EXPECT|ASSERT}_PRED2. Don't use +// this in your code. +template +AssertionResult AssertPred2Helper(const char* pred_text, + const char* e1, + const char* e2, + Pred pred, + const T1& v1, + const T2& v2) { + if (pred(v1, v2)) return AssertionSuccess(); + + return AssertionFailure() + << pred_text << "(" << e1 << ", " << e2 + << ") evaluates to false, where" + << "\n" + << e1 << " evaluates to " << ::testing::PrintToString(v1) << "\n" + << e2 << " evaluates to " << ::testing::PrintToString(v2); +} + +// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT2. +// Don't use this in your code. +#define GTEST_PRED_FORMAT2_(pred_format, v1, v2, on_failure)\ + GTEST_ASSERT_(pred_format(#v1, #v2, v1, v2), \ + on_failure) + +// Internal macro for implementing {EXPECT|ASSERT}_PRED2. Don't use +// this in your code. +#define GTEST_PRED2_(pred, v1, v2, on_failure)\ + GTEST_ASSERT_(::testing::AssertPred2Helper(#pred, \ + #v1, \ + #v2, \ + pred, \ + v1, \ + v2), on_failure) + +// Binary predicate assertion macros. +#define EXPECT_PRED_FORMAT2(pred_format, v1, v2) \ + GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_NONFATAL_FAILURE_) +#define EXPECT_PRED2(pred, v1, v2) \ + GTEST_PRED2_(pred, v1, v2, GTEST_NONFATAL_FAILURE_) +#define ASSERT_PRED_FORMAT2(pred_format, v1, v2) \ + GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_FATAL_FAILURE_) +#define ASSERT_PRED2(pred, v1, v2) \ + GTEST_PRED2_(pred, v1, v2, GTEST_FATAL_FAILURE_) + + + +// Helper function for implementing {EXPECT|ASSERT}_PRED3. Don't use +// this in your code. +template +AssertionResult AssertPred3Helper(const char* pred_text, + const char* e1, + const char* e2, + const char* e3, + Pred pred, + const T1& v1, + const T2& v2, + const T3& v3) { + if (pred(v1, v2, v3)) return AssertionSuccess(); + + return AssertionFailure() + << pred_text << "(" << e1 << ", " << e2 << ", " << e3 + << ") evaluates to false, where" + << "\n" + << e1 << " evaluates to " << ::testing::PrintToString(v1) << "\n" + << e2 << " evaluates to " << ::testing::PrintToString(v2) << "\n" + << e3 << " evaluates to " << ::testing::PrintToString(v3); +} + +// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT3. +// Don't use this in your code. +#define GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, on_failure)\ + GTEST_ASSERT_(pred_format(#v1, #v2, #v3, v1, v2, v3), \ + on_failure) + +// Internal macro for implementing {EXPECT|ASSERT}_PRED3. Don't use +// this in your code. +#define GTEST_PRED3_(pred, v1, v2, v3, on_failure)\ + GTEST_ASSERT_(::testing::AssertPred3Helper(#pred, \ + #v1, \ + #v2, \ + #v3, \ + pred, \ + v1, \ + v2, \ + v3), on_failure) + +// Ternary predicate assertion macros. +#define EXPECT_PRED_FORMAT3(pred_format, v1, v2, v3) \ + GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_NONFATAL_FAILURE_) +#define EXPECT_PRED3(pred, v1, v2, v3) \ + GTEST_PRED3_(pred, v1, v2, v3, GTEST_NONFATAL_FAILURE_) +#define ASSERT_PRED_FORMAT3(pred_format, v1, v2, v3) \ + GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_FATAL_FAILURE_) +#define ASSERT_PRED3(pred, v1, v2, v3) \ + GTEST_PRED3_(pred, v1, v2, v3, GTEST_FATAL_FAILURE_) + + + +// Helper function for implementing {EXPECT|ASSERT}_PRED4. Don't use +// this in your code. +template +AssertionResult AssertPred4Helper(const char* pred_text, + const char* e1, + const char* e2, + const char* e3, + const char* e4, + Pred pred, + const T1& v1, + const T2& v2, + const T3& v3, + const T4& v4) { + if (pred(v1, v2, v3, v4)) return AssertionSuccess(); + + return AssertionFailure() + << pred_text << "(" << e1 << ", " << e2 << ", " << e3 << ", " << e4 + << ") evaluates to false, where" + << "\n" + << e1 << " evaluates to " << ::testing::PrintToString(v1) << "\n" + << e2 << " evaluates to " << ::testing::PrintToString(v2) << "\n" + << e3 << " evaluates to " << ::testing::PrintToString(v3) << "\n" + << e4 << " evaluates to " << ::testing::PrintToString(v4); +} + +// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT4. +// Don't use this in your code. +#define GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, on_failure)\ + GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, v1, v2, v3, v4), \ + on_failure) + +// Internal macro for implementing {EXPECT|ASSERT}_PRED4. Don't use +// this in your code. +#define GTEST_PRED4_(pred, v1, v2, v3, v4, on_failure)\ + GTEST_ASSERT_(::testing::AssertPred4Helper(#pred, \ + #v1, \ + #v2, \ + #v3, \ + #v4, \ + pred, \ + v1, \ + v2, \ + v3, \ + v4), on_failure) + +// 4-ary predicate assertion macros. +#define EXPECT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \ + GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_) +#define EXPECT_PRED4(pred, v1, v2, v3, v4) \ + GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_) +#define ASSERT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \ + GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_FATAL_FAILURE_) +#define ASSERT_PRED4(pred, v1, v2, v3, v4) \ + GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_FATAL_FAILURE_) + + + +// Helper function for implementing {EXPECT|ASSERT}_PRED5. Don't use +// this in your code. +template +AssertionResult AssertPred5Helper(const char* pred_text, + const char* e1, + const char* e2, + const char* e3, + const char* e4, + const char* e5, + Pred pred, + const T1& v1, + const T2& v2, + const T3& v3, + const T4& v4, + const T5& v5) { + if (pred(v1, v2, v3, v4, v5)) return AssertionSuccess(); + + return AssertionFailure() + << pred_text << "(" << e1 << ", " << e2 << ", " << e3 << ", " << e4 + << ", " << e5 << ") evaluates to false, where" + << "\n" + << e1 << " evaluates to " << ::testing::PrintToString(v1) << "\n" + << e2 << " evaluates to " << ::testing::PrintToString(v2) << "\n" + << e3 << " evaluates to " << ::testing::PrintToString(v3) << "\n" + << e4 << " evaluates to " << ::testing::PrintToString(v4) << "\n" + << e5 << " evaluates to " << ::testing::PrintToString(v5); +} + +// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT5. +// Don't use this in your code. +#define GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, on_failure)\ + GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, #v5, v1, v2, v3, v4, v5), \ + on_failure) + +// Internal macro for implementing {EXPECT|ASSERT}_PRED5. Don't use +// this in your code. +#define GTEST_PRED5_(pred, v1, v2, v3, v4, v5, on_failure)\ + GTEST_ASSERT_(::testing::AssertPred5Helper(#pred, \ + #v1, \ + #v2, \ + #v3, \ + #v4, \ + #v5, \ + pred, \ + v1, \ + v2, \ + v3, \ + v4, \ + v5), on_failure) + +// 5-ary predicate assertion macros. +#define EXPECT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \ + GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_) +#define EXPECT_PRED5(pred, v1, v2, v3, v4, v5) \ + GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_) +#define ASSERT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \ + GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_) +#define ASSERT_PRED5(pred, v1, v2, v3, v4, v5) \ + GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_) + + + +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_ diff --git a/deps/googletest/include/gtest/gtest_prod.h b/deps/googletest/include/gtest/gtest_prod.h new file mode 100644 index 000000000..e651671eb --- /dev/null +++ b/deps/googletest/include/gtest/gtest_prod.h @@ -0,0 +1,61 @@ +// Copyright 2006, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// Google C++ Testing and Mocking Framework definitions useful in production code. +// GOOGLETEST_CM0003 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_GTEST_PROD_H_ +#define GTEST_INCLUDE_GTEST_GTEST_PROD_H_ + +// When you need to test the private or protected members of a class, +// use the FRIEND_TEST macro to declare your tests as friends of the +// class. For example: +// +// class MyClass { +// private: +// void PrivateMethod(); +// FRIEND_TEST(MyClassTest, PrivateMethodWorks); +// }; +// +// class MyClassTest : public testing::Test { +// // ... +// }; +// +// TEST_F(MyClassTest, PrivateMethodWorks) { +// // Can call MyClass::PrivateMethod() here. +// } +// +// Note: The test class must be in the same namespace as the class being tested. +// For example, putting MyClassTest in an anonymous namespace will not work. + +#define FRIEND_TEST(test_case_name, test_name)\ +friend class test_case_name##_##test_name##_Test + +#endif // GTEST_INCLUDE_GTEST_GTEST_PROD_H_ diff --git a/deps/googletest/include/gtest/internal/custom/README.md b/deps/googletest/include/gtest/internal/custom/README.md new file mode 100644 index 000000000..ff391fb4e --- /dev/null +++ b/deps/googletest/include/gtest/internal/custom/README.md @@ -0,0 +1,56 @@ +# Customization Points + +The custom directory is an injection point for custom user configurations. + +## Header `gtest.h` + +### The following macros can be defined: + +* `GTEST_OS_STACK_TRACE_GETTER_` - The name of an implementation of + `OsStackTraceGetterInterface`. +* `GTEST_CUSTOM_TEMPDIR_FUNCTION_` - An override for `testing::TempDir()`. See + `testing::TempDir` for semantics and signature. + +## Header `gtest-port.h` + +The following macros can be defined: + +### Flag related macros: + +* `GTEST_FLAG(flag_name)` +* `GTEST_USE_OWN_FLAGFILE_FLAG_` - Define to 0 when the system provides its + own flagfile flag parsing. +* `GTEST_DECLARE_bool_(name)` +* `GTEST_DECLARE_int32_(name)` +* `GTEST_DECLARE_string_(name)` +* `GTEST_DEFINE_bool_(name, default_val, doc)` +* `GTEST_DEFINE_int32_(name, default_val, doc)` +* `GTEST_DEFINE_string_(name, default_val, doc)` + +### Logging: + +* `GTEST_LOG_(severity)` +* `GTEST_CHECK_(condition)` +* Functions `LogToStderr()` and `FlushInfoLog()` have to be provided too. + +### Threading: + +* `GTEST_HAS_NOTIFICATION_` - Enabled if Notification is already provided. +* `GTEST_HAS_MUTEX_AND_THREAD_LOCAL_` - Enabled if `Mutex` and `ThreadLocal` + are already provided. Must also provide `GTEST_DECLARE_STATIC_MUTEX_(mutex)` + and `GTEST_DEFINE_STATIC_MUTEX_(mutex)` +* `GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks)` +* `GTEST_LOCK_EXCLUDED_(locks)` + +### Underlying library support features + +* `GTEST_HAS_CXXABI_H_` + +### Exporting API symbols: + +* `GTEST_API_` - Specifier for exported symbols. + +## Header `gtest-printers.h` + +* See documentation at `gtest/gtest-printers.h` for details on how to define a + custom printer. diff --git a/deps/googletest/include/gtest/internal/custom/gtest-port.h b/deps/googletest/include/gtest/internal/custom/gtest-port.h new file mode 100644 index 000000000..cd85d956d --- /dev/null +++ b/deps/googletest/include/gtest/internal/custom/gtest-port.h @@ -0,0 +1,37 @@ +// Copyright 2015, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Injection point for custom user configurations. See README for details +// +// ** Custom implementation starts here ** + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_ + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_ diff --git a/deps/googletest/include/gtest/internal/custom/gtest-printers.h b/deps/googletest/include/gtest/internal/custom/gtest-printers.h new file mode 100644 index 000000000..eb4467abc --- /dev/null +++ b/deps/googletest/include/gtest/internal/custom/gtest-printers.h @@ -0,0 +1,42 @@ +// Copyright 2015, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// This file provides an injection point for custom printers in a local +// installation of gTest. +// It will be included from gtest-printers.h and the overrides in this file +// will be visible to everyone. +// +// Injection point for custom user configurations. See README for details +// +// ** Custom implementation starts here ** + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_ + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_ diff --git a/deps/googletest/include/gtest/internal/custom/gtest.h b/deps/googletest/include/gtest/internal/custom/gtest.h new file mode 100644 index 000000000..4c8e07be2 --- /dev/null +++ b/deps/googletest/include/gtest/internal/custom/gtest.h @@ -0,0 +1,37 @@ +// Copyright 2015, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Injection point for custom user configurations. See README for details +// +// ** Custom implementation starts here ** + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_H_ + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-death-test-internal.h b/deps/googletest/include/gtest/internal/gtest-death-test-internal.h new file mode 100644 index 000000000..68bd35306 --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-death-test-internal.h @@ -0,0 +1,304 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file defines internal utilities needed for implementing +// death tests. They are subject to change without notice. +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_ + +#include "gtest/gtest-matchers.h" +#include "gtest/internal/gtest-internal.h" + +#include +#include + +namespace testing { +namespace internal { + +GTEST_DECLARE_string_(internal_run_death_test); + +// Names of the flags (needed for parsing Google Test flags). +const char kDeathTestStyleFlag[] = "death_test_style"; +const char kDeathTestUseFork[] = "death_test_use_fork"; +const char kInternalRunDeathTestFlag[] = "internal_run_death_test"; + +#if GTEST_HAS_DEATH_TEST + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +// DeathTest is a class that hides much of the complexity of the +// GTEST_DEATH_TEST_ macro. It is abstract; its static Create method +// returns a concrete class that depends on the prevailing death test +// style, as defined by the --gtest_death_test_style and/or +// --gtest_internal_run_death_test flags. + +// In describing the results of death tests, these terms are used with +// the corresponding definitions: +// +// exit status: The integer exit information in the format specified +// by wait(2) +// exit code: The integer code passed to exit(3), _exit(2), or +// returned from main() +class GTEST_API_ DeathTest { + public: + // Create returns false if there was an error determining the + // appropriate action to take for the current death test; for example, + // if the gtest_death_test_style flag is set to an invalid value. + // The LastMessage method will return a more detailed message in that + // case. Otherwise, the DeathTest pointer pointed to by the "test" + // argument is set. If the death test should be skipped, the pointer + // is set to NULL; otherwise, it is set to the address of a new concrete + // DeathTest object that controls the execution of the current test. + static bool Create(const char* statement, Matcher matcher, + const char* file, int line, DeathTest** test); + DeathTest(); + virtual ~DeathTest() { } + + // A helper class that aborts a death test when it's deleted. + class ReturnSentinel { + public: + explicit ReturnSentinel(DeathTest* test) : test_(test) { } + ~ReturnSentinel() { test_->Abort(TEST_ENCOUNTERED_RETURN_STATEMENT); } + private: + DeathTest* const test_; + GTEST_DISALLOW_COPY_AND_ASSIGN_(ReturnSentinel); + } GTEST_ATTRIBUTE_UNUSED_; + + // An enumeration of possible roles that may be taken when a death + // test is encountered. EXECUTE means that the death test logic should + // be executed immediately. OVERSEE means that the program should prepare + // the appropriate environment for a child process to execute the death + // test, then wait for it to complete. + enum TestRole { OVERSEE_TEST, EXECUTE_TEST }; + + // An enumeration of the three reasons that a test might be aborted. + enum AbortReason { + TEST_ENCOUNTERED_RETURN_STATEMENT, + TEST_THREW_EXCEPTION, + TEST_DID_NOT_DIE + }; + + // Assumes one of the above roles. + virtual TestRole AssumeRole() = 0; + + // Waits for the death test to finish and returns its status. + virtual int Wait() = 0; + + // Returns true if the death test passed; that is, the test process + // exited during the test, its exit status matches a user-supplied + // predicate, and its stderr output matches a user-supplied regular + // expression. + // The user-supplied predicate may be a macro expression rather + // than a function pointer or functor, or else Wait and Passed could + // be combined. + virtual bool Passed(bool exit_status_ok) = 0; + + // Signals that the death test did not die as expected. + virtual void Abort(AbortReason reason) = 0; + + // Returns a human-readable outcome message regarding the outcome of + // the last death test. + static const char* LastMessage(); + + static void set_last_death_test_message(const std::string& message); + + private: + // A string containing a description of the outcome of the last death test. + static std::string last_death_test_message_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(DeathTest); +}; + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +// Factory interface for death tests. May be mocked out for testing. +class DeathTestFactory { + public: + virtual ~DeathTestFactory() { } + virtual bool Create(const char* statement, + Matcher matcher, const char* file, + int line, DeathTest** test) = 0; +}; + +// A concrete DeathTestFactory implementation for normal use. +class DefaultDeathTestFactory : public DeathTestFactory { + public: + bool Create(const char* statement, Matcher matcher, + const char* file, int line, DeathTest** test) override; +}; + +// Returns true if exit_status describes a process that was terminated +// by a signal, or exited normally with a nonzero exit code. +GTEST_API_ bool ExitedUnsuccessfully(int exit_status); + +// A string passed to EXPECT_DEATH (etc.) is caught by one of these overloads +// and interpreted as a regex (rather than an Eq matcher) for legacy +// compatibility. +inline Matcher MakeDeathTestMatcher( + ::testing::internal::RE regex) { + return ContainsRegex(regex.pattern()); +} +inline Matcher MakeDeathTestMatcher(const char* regex) { + return ContainsRegex(regex); +} +inline Matcher MakeDeathTestMatcher( + const ::std::string& regex) { + return ContainsRegex(regex); +} + +// If a Matcher is passed to EXPECT_DEATH (etc.), it's +// used directly. +inline Matcher MakeDeathTestMatcher( + Matcher matcher) { + return matcher; +} + +// Traps C++ exceptions escaping statement and reports them as test +// failures. Note that trapping SEH exceptions is not implemented here. +# if GTEST_HAS_EXCEPTIONS +# define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \ + try { \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + } catch (const ::std::exception& gtest_exception) { \ + fprintf(\ + stderr, \ + "\n%s: Caught std::exception-derived exception escaping the " \ + "death test statement. Exception message: %s\n", \ + ::testing::internal::FormatFileLocation(__FILE__, __LINE__).c_str(), \ + gtest_exception.what()); \ + fflush(stderr); \ + death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \ + } catch (...) { \ + death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \ + } + +# else +# define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) + +# endif + +// This macro is for implementing ASSERT_DEATH*, EXPECT_DEATH*, +// ASSERT_EXIT*, and EXPECT_EXIT*. +#define GTEST_DEATH_TEST_(statement, predicate, regex_or_matcher, fail) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::AlwaysTrue()) { \ + ::testing::internal::DeathTest* gtest_dt; \ + if (!::testing::internal::DeathTest::Create( \ + #statement, \ + ::testing::internal::MakeDeathTestMatcher(regex_or_matcher), \ + __FILE__, __LINE__, >est_dt)) { \ + goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \ + } \ + if (gtest_dt != nullptr) { \ + std::unique_ptr< ::testing::internal::DeathTest> gtest_dt_ptr(gtest_dt); \ + switch (gtest_dt->AssumeRole()) { \ + case ::testing::internal::DeathTest::OVERSEE_TEST: \ + if (!gtest_dt->Passed(predicate(gtest_dt->Wait()))) { \ + goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \ + } \ + break; \ + case ::testing::internal::DeathTest::EXECUTE_TEST: { \ + ::testing::internal::DeathTest::ReturnSentinel gtest_sentinel( \ + gtest_dt); \ + GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, gtest_dt); \ + gtest_dt->Abort(::testing::internal::DeathTest::TEST_DID_NOT_DIE); \ + break; \ + } \ + default: \ + break; \ + } \ + } \ + } else \ + GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__) \ + : fail(::testing::internal::DeathTest::LastMessage()) +// The symbol "fail" here expands to something into which a message +// can be streamed. + +// This macro is for implementing ASSERT/EXPECT_DEBUG_DEATH when compiled in +// NDEBUG mode. In this case we need the statements to be executed and the macro +// must accept a streamed message even though the message is never printed. +// The regex object is not evaluated, but it is used to prevent "unused" +// warnings and to avoid an expression that doesn't compile in debug mode. +#define GTEST_EXECUTE_STATEMENT_(statement, regex_or_matcher) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::AlwaysTrue()) { \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + } else if (!::testing::internal::AlwaysTrue()) { \ + ::testing::internal::MakeDeathTestMatcher(regex_or_matcher); \ + } else \ + ::testing::Message() + +// A class representing the parsed contents of the +// --gtest_internal_run_death_test flag, as it existed when +// RUN_ALL_TESTS was called. +class InternalRunDeathTestFlag { + public: + InternalRunDeathTestFlag(const std::string& a_file, + int a_line, + int an_index, + int a_write_fd) + : file_(a_file), line_(a_line), index_(an_index), + write_fd_(a_write_fd) {} + + ~InternalRunDeathTestFlag() { + if (write_fd_ >= 0) + posix::Close(write_fd_); + } + + const std::string& file() const { return file_; } + int line() const { return line_; } + int index() const { return index_; } + int write_fd() const { return write_fd_; } + + private: + std::string file_; + int line_; + int index_; + int write_fd_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(InternalRunDeathTestFlag); +}; + +// Returns a newly created InternalRunDeathTestFlag object with fields +// initialized from the GTEST_FLAG(internal_run_death_test) flag if +// the flag is specified; otherwise returns NULL. +InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag(); + +#endif // GTEST_HAS_DEATH_TEST + +} // namespace internal +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-filepath.h b/deps/googletest/include/gtest/internal/gtest-filepath.h new file mode 100644 index 000000000..ae38d95bf --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-filepath.h @@ -0,0 +1,211 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Google Test filepath utilities +// +// This header file declares classes and functions used internally by +// Google Test. They are subject to change without notice. +// +// This file is #included in gtest/internal/gtest-internal.h. +// Do not include this header file separately! + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_ + +#include "gtest/internal/gtest-string.h" + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +namespace testing { +namespace internal { + +// FilePath - a class for file and directory pathname manipulation which +// handles platform-specific conventions (like the pathname separator). +// Used for helper functions for naming files in a directory for xml output. +// Except for Set methods, all methods are const or static, which provides an +// "immutable value object" -- useful for peace of mind. +// A FilePath with a value ending in a path separator ("like/this/") represents +// a directory, otherwise it is assumed to represent a file. In either case, +// it may or may not represent an actual file or directory in the file system. +// Names are NOT checked for syntax correctness -- no checking for illegal +// characters, malformed paths, etc. + +class GTEST_API_ FilePath { + public: + FilePath() : pathname_("") { } + FilePath(const FilePath& rhs) : pathname_(rhs.pathname_) { } + + explicit FilePath(const std::string& pathname) : pathname_(pathname) { + Normalize(); + } + + FilePath& operator=(const FilePath& rhs) { + Set(rhs); + return *this; + } + + void Set(const FilePath& rhs) { + pathname_ = rhs.pathname_; + } + + const std::string& string() const { return pathname_; } + const char* c_str() const { return pathname_.c_str(); } + + // Returns the current working directory, or "" if unsuccessful. + static FilePath GetCurrentDir(); + + // Given directory = "dir", base_name = "test", number = 0, + // extension = "xml", returns "dir/test.xml". If number is greater + // than zero (e.g., 12), returns "dir/test_12.xml". + // On Windows platform, uses \ as the separator rather than /. + static FilePath MakeFileName(const FilePath& directory, + const FilePath& base_name, + int number, + const char* extension); + + // Given directory = "dir", relative_path = "test.xml", + // returns "dir/test.xml". + // On Windows, uses \ as the separator rather than /. + static FilePath ConcatPaths(const FilePath& directory, + const FilePath& relative_path); + + // Returns a pathname for a file that does not currently exist. The pathname + // will be directory/base_name.extension or + // directory/base_name_.extension if directory/base_name.extension + // already exists. The number will be incremented until a pathname is found + // that does not already exist. + // Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'. + // There could be a race condition if two or more processes are calling this + // function at the same time -- they could both pick the same filename. + static FilePath GenerateUniqueFileName(const FilePath& directory, + const FilePath& base_name, + const char* extension); + + // Returns true iff the path is "". + bool IsEmpty() const { return pathname_.empty(); } + + // If input name has a trailing separator character, removes it and returns + // the name, otherwise return the name string unmodified. + // On Windows platform, uses \ as the separator, other platforms use /. + FilePath RemoveTrailingPathSeparator() const; + + // Returns a copy of the FilePath with the directory part removed. + // Example: FilePath("path/to/file").RemoveDirectoryName() returns + // FilePath("file"). If there is no directory part ("just_a_file"), it returns + // the FilePath unmodified. If there is no file part ("just_a_dir/") it + // returns an empty FilePath (""). + // On Windows platform, '\' is the path separator, otherwise it is '/'. + FilePath RemoveDirectoryName() const; + + // RemoveFileName returns the directory path with the filename removed. + // Example: FilePath("path/to/file").RemoveFileName() returns "path/to/". + // If the FilePath is "a_file" or "/a_file", RemoveFileName returns + // FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does + // not have a file, like "just/a/dir/", it returns the FilePath unmodified. + // On Windows platform, '\' is the path separator, otherwise it is '/'. + FilePath RemoveFileName() const; + + // Returns a copy of the FilePath with the case-insensitive extension removed. + // Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns + // FilePath("dir/file"). If a case-insensitive extension is not + // found, returns a copy of the original FilePath. + FilePath RemoveExtension(const char* extension) const; + + // Creates directories so that path exists. Returns true if successful or if + // the directories already exist; returns false if unable to create + // directories for any reason. Will also return false if the FilePath does + // not represent a directory (that is, it doesn't end with a path separator). + bool CreateDirectoriesRecursively() const; + + // Create the directory so that path exists. Returns true if successful or + // if the directory already exists; returns false if unable to create the + // directory for any reason, including if the parent directory does not + // exist. Not named "CreateDirectory" because that's a macro on Windows. + bool CreateFolder() const; + + // Returns true if FilePath describes something in the file-system, + // either a file, directory, or whatever, and that something exists. + bool FileOrDirectoryExists() const; + + // Returns true if pathname describes a directory in the file-system + // that exists. + bool DirectoryExists() const; + + // Returns true if FilePath ends with a path separator, which indicates that + // it is intended to represent a directory. Returns false otherwise. + // This does NOT check that a directory (or file) actually exists. + bool IsDirectory() const; + + // Returns true if pathname describes a root directory. (Windows has one + // root directory per disk drive.) + bool IsRootDirectory() const; + + // Returns true if pathname describes an absolute path. + bool IsAbsolutePath() const; + + private: + // Replaces multiple consecutive separators with a single separator. + // For example, "bar///foo" becomes "bar/foo". Does not eliminate other + // redundancies that might be in a pathname involving "." or "..". + // + // A pathname with multiple consecutive separators may occur either through + // user error or as a result of some scripts or APIs that generate a pathname + // with a trailing separator. On other platforms the same API or script + // may NOT generate a pathname with a trailing "/". Then elsewhere that + // pathname may have another "/" and pathname components added to it, + // without checking for the separator already being there. + // The script language and operating system may allow paths like "foo//bar" + // but some of the functions in FilePath will not handle that correctly. In + // particular, RemoveTrailingPathSeparator() only removes one separator, and + // it is called in CreateDirectoriesRecursively() assuming that it will change + // a pathname from directory syntax (trailing separator) to filename syntax. + // + // On Windows this method also replaces the alternate path separator '/' with + // the primary path separator '\\', so that for example "bar\\/\\foo" becomes + // "bar\\foo". + + void Normalize(); + + // Returns a pointer to the last occurence of a valid path separator in + // the FilePath. On Windows, for example, both '/' and '\' are valid path + // separators. Returns NULL if no path separator was found. + const char* FindLastPathSeparator() const; + + std::string pathname_; +}; // class FilePath + +} // namespace internal +} // namespace testing + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-internal.h b/deps/googletest/include/gtest/internal/gtest-internal.h new file mode 100644 index 000000000..273266081 --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-internal.h @@ -0,0 +1,1428 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file declares functions and macros used internally by +// Google Test. They are subject to change without notice. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_ + +#include "gtest/internal/gtest-port.h" + +#if GTEST_OS_LINUX +# include +# include +# include +# include +#endif // GTEST_OS_LINUX + +#if GTEST_HAS_EXCEPTIONS +# include +#endif + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "gtest/gtest-message.h" +#include "gtest/internal/gtest-filepath.h" +#include "gtest/internal/gtest-string.h" +#include "gtest/internal/gtest-type-util.h" + +// Due to C++ preprocessor weirdness, we need double indirection to +// concatenate two tokens when one of them is __LINE__. Writing +// +// foo ## __LINE__ +// +// will result in the token foo__LINE__, instead of foo followed by +// the current line number. For more details, see +// http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.6 +#define GTEST_CONCAT_TOKEN_(foo, bar) GTEST_CONCAT_TOKEN_IMPL_(foo, bar) +#define GTEST_CONCAT_TOKEN_IMPL_(foo, bar) foo ## bar + +// Stringifies its argument. +#define GTEST_STRINGIFY_(name) #name + +namespace proto2 { class Message; } + +namespace testing { + +// Forward declarations. + +class AssertionResult; // Result of an assertion. +class Message; // Represents a failure message. +class Test; // Represents a test. +class TestInfo; // Information about a test. +class TestPartResult; // Result of a test part. +class UnitTest; // A collection of test suites. + +template +::std::string PrintToString(const T& value); + +namespace internal { + +struct TraceInfo; // Information about a trace point. +class TestInfoImpl; // Opaque implementation of TestInfo +class UnitTestImpl; // Opaque implementation of UnitTest + +// The text used in failure messages to indicate the start of the +// stack trace. +GTEST_API_ extern const char kStackTraceMarker[]; + +// An IgnoredValue object can be implicitly constructed from ANY value. +class IgnoredValue { + struct Sink {}; + public: + // This constructor template allows any value to be implicitly + // converted to IgnoredValue. The object has no data member and + // doesn't try to remember anything about the argument. We + // deliberately omit the 'explicit' keyword in order to allow the + // conversion to be implicit. + // Disable the conversion if T already has a magical conversion operator. + // Otherwise we get ambiguity. + template ::value, + int>::type = 0> + IgnoredValue(const T& /* ignored */) {} // NOLINT(runtime/explicit) +}; + +// Appends the user-supplied message to the Google-Test-generated message. +GTEST_API_ std::string AppendUserMessage( + const std::string& gtest_msg, const Message& user_msg); + +#if GTEST_HAS_EXCEPTIONS + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4275 \ +/* an exported class was derived from a class that was not exported */) + +// This exception is thrown by (and only by) a failed Google Test +// assertion when GTEST_FLAG(throw_on_failure) is true (if exceptions +// are enabled). We derive it from std::runtime_error, which is for +// errors presumably detectable only at run time. Since +// std::runtime_error inherits from std::exception, many testing +// frameworks know how to extract and print the message inside it. +class GTEST_API_ GoogleTestFailureException : public ::std::runtime_error { + public: + explicit GoogleTestFailureException(const TestPartResult& failure); +}; + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4275 + +#endif // GTEST_HAS_EXCEPTIONS + +namespace edit_distance { +// Returns the optimal edits to go from 'left' to 'right'. +// All edits cost the same, with replace having lower priority than +// add/remove. +// Simple implementation of the Wagner-Fischer algorithm. +// See http://en.wikipedia.org/wiki/Wagner-Fischer_algorithm +enum EditType { kMatch, kAdd, kRemove, kReplace }; +GTEST_API_ std::vector CalculateOptimalEdits( + const std::vector& left, const std::vector& right); + +// Same as above, but the input is represented as strings. +GTEST_API_ std::vector CalculateOptimalEdits( + const std::vector& left, + const std::vector& right); + +// Create a diff of the input strings in Unified diff format. +GTEST_API_ std::string CreateUnifiedDiff(const std::vector& left, + const std::vector& right, + size_t context = 2); + +} // namespace edit_distance + +// Calculate the diff between 'left' and 'right' and return it in unified diff +// format. +// If not null, stores in 'total_line_count' the total number of lines found +// in left + right. +GTEST_API_ std::string DiffStrings(const std::string& left, + const std::string& right, + size_t* total_line_count); + +// Constructs and returns the message for an equality assertion +// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure. +// +// The first four parameters are the expressions used in the assertion +// and their values, as strings. For example, for ASSERT_EQ(foo, bar) +// where foo is 5 and bar is 6, we have: +// +// expected_expression: "foo" +// actual_expression: "bar" +// expected_value: "5" +// actual_value: "6" +// +// The ignoring_case parameter is true iff the assertion is a +// *_STRCASEEQ*. When it's true, the string " (ignoring case)" will +// be inserted into the message. +GTEST_API_ AssertionResult EqFailure(const char* expected_expression, + const char* actual_expression, + const std::string& expected_value, + const std::string& actual_value, + bool ignoring_case); + +// Constructs a failure message for Boolean assertions such as EXPECT_TRUE. +GTEST_API_ std::string GetBoolAssertionFailureMessage( + const AssertionResult& assertion_result, + const char* expression_text, + const char* actual_predicate_value, + const char* expected_predicate_value); + +// This template class represents an IEEE floating-point number +// (either single-precision or double-precision, depending on the +// template parameters). +// +// The purpose of this class is to do more sophisticated number +// comparison. (Due to round-off error, etc, it's very unlikely that +// two floating-points will be equal exactly. Hence a naive +// comparison by the == operation often doesn't work.) +// +// Format of IEEE floating-point: +// +// The most-significant bit being the leftmost, an IEEE +// floating-point looks like +// +// sign_bit exponent_bits fraction_bits +// +// Here, sign_bit is a single bit that designates the sign of the +// number. +// +// For float, there are 8 exponent bits and 23 fraction bits. +// +// For double, there are 11 exponent bits and 52 fraction bits. +// +// More details can be found at +// http://en.wikipedia.org/wiki/IEEE_floating-point_standard. +// +// Template parameter: +// +// RawType: the raw floating-point type (either float or double) +template +class FloatingPoint { + public: + // Defines the unsigned integer type that has the same size as the + // floating point number. + typedef typename TypeWithSize::UInt Bits; + + // Constants. + + // # of bits in a number. + static const size_t kBitCount = 8*sizeof(RawType); + + // # of fraction bits in a number. + static const size_t kFractionBitCount = + std::numeric_limits::digits - 1; + + // # of exponent bits in a number. + static const size_t kExponentBitCount = kBitCount - 1 - kFractionBitCount; + + // The mask for the sign bit. + static const Bits kSignBitMask = static_cast(1) << (kBitCount - 1); + + // The mask for the fraction bits. + static const Bits kFractionBitMask = + ~static_cast(0) >> (kExponentBitCount + 1); + + // The mask for the exponent bits. + static const Bits kExponentBitMask = ~(kSignBitMask | kFractionBitMask); + + // How many ULP's (Units in the Last Place) we want to tolerate when + // comparing two numbers. The larger the value, the more error we + // allow. A 0 value means that two numbers must be exactly the same + // to be considered equal. + // + // The maximum error of a single floating-point operation is 0.5 + // units in the last place. On Intel CPU's, all floating-point + // calculations are done with 80-bit precision, while double has 64 + // bits. Therefore, 4 should be enough for ordinary use. + // + // See the following article for more details on ULP: + // http://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/ + static const size_t kMaxUlps = 4; + + // Constructs a FloatingPoint from a raw floating-point number. + // + // On an Intel CPU, passing a non-normalized NAN (Not a Number) + // around may change its bits, although the new value is guaranteed + // to be also a NAN. Therefore, don't expect this constructor to + // preserve the bits in x when x is a NAN. + explicit FloatingPoint(const RawType& x) { u_.value_ = x; } + + // Static methods + + // Reinterprets a bit pattern as a floating-point number. + // + // This function is needed to test the AlmostEquals() method. + static RawType ReinterpretBits(const Bits bits) { + FloatingPoint fp(0); + fp.u_.bits_ = bits; + return fp.u_.value_; + } + + // Returns the floating-point number that represent positive infinity. + static RawType Infinity() { + return ReinterpretBits(kExponentBitMask); + } + + // Returns the maximum representable finite floating-point number. + static RawType Max(); + + // Non-static methods + + // Returns the bits that represents this number. + const Bits &bits() const { return u_.bits_; } + + // Returns the exponent bits of this number. + Bits exponent_bits() const { return kExponentBitMask & u_.bits_; } + + // Returns the fraction bits of this number. + Bits fraction_bits() const { return kFractionBitMask & u_.bits_; } + + // Returns the sign bit of this number. + Bits sign_bit() const { return kSignBitMask & u_.bits_; } + + // Returns true iff this is NAN (not a number). + bool is_nan() const { + // It's a NAN if the exponent bits are all ones and the fraction + // bits are not entirely zeros. + return (exponent_bits() == kExponentBitMask) && (fraction_bits() != 0); + } + + // Returns true iff this number is at most kMaxUlps ULP's away from + // rhs. In particular, this function: + // + // - returns false if either number is (or both are) NAN. + // - treats really large numbers as almost equal to infinity. + // - thinks +0.0 and -0.0 are 0 DLP's apart. + bool AlmostEquals(const FloatingPoint& rhs) const { + // The IEEE standard says that any comparison operation involving + // a NAN must return false. + if (is_nan() || rhs.is_nan()) return false; + + return DistanceBetweenSignAndMagnitudeNumbers(u_.bits_, rhs.u_.bits_) + <= kMaxUlps; + } + + private: + // The data type used to store the actual floating-point number. + union FloatingPointUnion { + RawType value_; // The raw floating-point number. + Bits bits_; // The bits that represent the number. + }; + + // Converts an integer from the sign-and-magnitude representation to + // the biased representation. More precisely, let N be 2 to the + // power of (kBitCount - 1), an integer x is represented by the + // unsigned number x + N. + // + // For instance, + // + // -N + 1 (the most negative number representable using + // sign-and-magnitude) is represented by 1; + // 0 is represented by N; and + // N - 1 (the biggest number representable using + // sign-and-magnitude) is represented by 2N - 1. + // + // Read http://en.wikipedia.org/wiki/Signed_number_representations + // for more details on signed number representations. + static Bits SignAndMagnitudeToBiased(const Bits &sam) { + if (kSignBitMask & sam) { + // sam represents a negative number. + return ~sam + 1; + } else { + // sam represents a positive number. + return kSignBitMask | sam; + } + } + + // Given two numbers in the sign-and-magnitude representation, + // returns the distance between them as an unsigned number. + static Bits DistanceBetweenSignAndMagnitudeNumbers(const Bits &sam1, + const Bits &sam2) { + const Bits biased1 = SignAndMagnitudeToBiased(sam1); + const Bits biased2 = SignAndMagnitudeToBiased(sam2); + return (biased1 >= biased2) ? (biased1 - biased2) : (biased2 - biased1); + } + + FloatingPointUnion u_; +}; + +// We cannot use std::numeric_limits::max() as it clashes with the max() +// macro defined by . +template <> +inline float FloatingPoint::Max() { return FLT_MAX; } +template <> +inline double FloatingPoint::Max() { return DBL_MAX; } + +// Typedefs the instances of the FloatingPoint template class that we +// care to use. +typedef FloatingPoint Float; +typedef FloatingPoint Double; + +// In order to catch the mistake of putting tests that use different +// test fixture classes in the same test suite, we need to assign +// unique IDs to fixture classes and compare them. The TypeId type is +// used to hold such IDs. The user should treat TypeId as an opaque +// type: the only operation allowed on TypeId values is to compare +// them for equality using the == operator. +typedef const void* TypeId; + +template +class TypeIdHelper { + public: + // dummy_ must not have a const type. Otherwise an overly eager + // compiler (e.g. MSVC 7.1 & 8.0) may try to merge + // TypeIdHelper::dummy_ for different Ts as an "optimization". + static bool dummy_; +}; + +template +bool TypeIdHelper::dummy_ = false; + +// GetTypeId() returns the ID of type T. Different values will be +// returned for different types. Calling the function twice with the +// same type argument is guaranteed to return the same ID. +template +TypeId GetTypeId() { + // The compiler is required to allocate a different + // TypeIdHelper::dummy_ variable for each T used to instantiate + // the template. Therefore, the address of dummy_ is guaranteed to + // be unique. + return &(TypeIdHelper::dummy_); +} + +// Returns the type ID of ::testing::Test. Always call this instead +// of GetTypeId< ::testing::Test>() to get the type ID of +// ::testing::Test, as the latter may give the wrong result due to a +// suspected linker bug when compiling Google Test as a Mac OS X +// framework. +GTEST_API_ TypeId GetTestTypeId(); + +// Defines the abstract factory interface that creates instances +// of a Test object. +class TestFactoryBase { + public: + virtual ~TestFactoryBase() {} + + // Creates a test instance to run. The instance is both created and destroyed + // within TestInfoImpl::Run() + virtual Test* CreateTest() = 0; + + protected: + TestFactoryBase() {} + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestFactoryBase); +}; + +// This class provides implementation of TeastFactoryBase interface. +// It is used in TEST and TEST_F macros. +template +class TestFactoryImpl : public TestFactoryBase { + public: + Test* CreateTest() override { return new TestClass; } +}; + +#if GTEST_OS_WINDOWS + +// Predicate-formatters for implementing the HRESULT checking macros +// {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED} +// We pass a long instead of HRESULT to avoid causing an +// include dependency for the HRESULT type. +GTEST_API_ AssertionResult IsHRESULTSuccess(const char* expr, + long hr); // NOLINT +GTEST_API_ AssertionResult IsHRESULTFailure(const char* expr, + long hr); // NOLINT + +#endif // GTEST_OS_WINDOWS + +// Types of SetUpTestSuite() and TearDownTestSuite() functions. +using SetUpTestSuiteFunc = void (*)(); +using TearDownTestSuiteFunc = void (*)(); + +struct CodeLocation { + CodeLocation(const std::string& a_file, int a_line) + : file(a_file), line(a_line) {} + + std::string file; + int line; +}; + +// Helper to identify which setup function for TestCase / TestSuite to call. +// Only one function is allowed, either TestCase or TestSute but not both. + +// Utility functions to help SuiteApiResolver +using SetUpTearDownSuiteFuncType = void (*)(); + +inline SetUpTearDownSuiteFuncType GetNotDefaultOrNull( + SetUpTearDownSuiteFuncType a, SetUpTearDownSuiteFuncType def) { + return a == def ? nullptr : a; +} + +template +// Note that SuiteApiResolver inherits from T because +// SetUpTestSuite()/TearDownTestSuite() could be protected. Ths way +// SuiteApiResolver can access them. +struct SuiteApiResolver : T { + // testing::Test is only forward declared at this point. So we make it a + // dependend class for the compiler to be OK with it. + using Test = + typename std::conditional::type; + + static SetUpTearDownSuiteFuncType GetSetUpCaseOrSuite(const char* filename, + int line_num) { + SetUpTearDownSuiteFuncType test_case_fp = + GetNotDefaultOrNull(&T::SetUpTestCase, &Test::SetUpTestCase); + SetUpTearDownSuiteFuncType test_suite_fp = + GetNotDefaultOrNull(&T::SetUpTestSuite, &Test::SetUpTestSuite); + + GTEST_CHECK_(!test_case_fp || !test_suite_fp) + << "Test can not provide both SetUpTestSuite and SetUpTestCase, please " + "make sure there is only one present at " + << filename << ":" << line_num; + + return test_case_fp != nullptr ? test_case_fp : test_suite_fp; + } + + static SetUpTearDownSuiteFuncType GetTearDownCaseOrSuite(const char* filename, + int line_num) { + SetUpTearDownSuiteFuncType test_case_fp = + GetNotDefaultOrNull(&T::TearDownTestCase, &Test::TearDownTestCase); + SetUpTearDownSuiteFuncType test_suite_fp = + GetNotDefaultOrNull(&T::TearDownTestSuite, &Test::TearDownTestSuite); + + GTEST_CHECK_(!test_case_fp || !test_suite_fp) + << "Test can not provide both TearDownTestSuite and TearDownTestCase," + " please make sure there is only one present at" + << filename << ":" << line_num; + + return test_case_fp != nullptr ? test_case_fp : test_suite_fp; + } +}; + +// Creates a new TestInfo object and registers it with Google Test; +// returns the created object. +// +// Arguments: +// +// test_suite_name: name of the test suite +// name: name of the test +// type_param the name of the test's type parameter, or NULL if +// this is not a typed or a type-parameterized test. +// value_param text representation of the test's value parameter, +// or NULL if this is not a type-parameterized test. +// code_location: code location where the test is defined +// fixture_class_id: ID of the test fixture class +// set_up_tc: pointer to the function that sets up the test suite +// tear_down_tc: pointer to the function that tears down the test suite +// factory: pointer to the factory that creates a test object. +// The newly created TestInfo instance will assume +// ownership of the factory object. +GTEST_API_ TestInfo* MakeAndRegisterTestInfo( + const char* test_suite_name, const char* name, const char* type_param, + const char* value_param, CodeLocation code_location, + TypeId fixture_class_id, SetUpTestSuiteFunc set_up_tc, + TearDownTestSuiteFunc tear_down_tc, TestFactoryBase* factory); + +// If *pstr starts with the given prefix, modifies *pstr to be right +// past the prefix and returns true; otherwise leaves *pstr unchanged +// and returns false. None of pstr, *pstr, and prefix can be NULL. +GTEST_API_ bool SkipPrefix(const char* prefix, const char** pstr); + +#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +// State of the definition of a type-parameterized test suite. +class GTEST_API_ TypedTestSuitePState { + public: + TypedTestSuitePState() : registered_(false) {} + + // Adds the given test name to defined_test_names_ and return true + // if the test suite hasn't been registered; otherwise aborts the + // program. + bool AddTestName(const char* file, int line, const char* case_name, + const char* test_name) { + if (registered_) { + fprintf(stderr, + "%s Test %s must be defined before " + "REGISTER_TYPED_TEST_SUITE_P(%s, ...).\n", + FormatFileLocation(file, line).c_str(), test_name, case_name); + fflush(stderr); + posix::Abort(); + } + registered_tests_.insert( + ::std::make_pair(test_name, CodeLocation(file, line))); + return true; + } + + bool TestExists(const std::string& test_name) const { + return registered_tests_.count(test_name) > 0; + } + + const CodeLocation& GetCodeLocation(const std::string& test_name) const { + RegisteredTestsMap::const_iterator it = registered_tests_.find(test_name); + GTEST_CHECK_(it != registered_tests_.end()); + return it->second; + } + + // Verifies that registered_tests match the test names in + // defined_test_names_; returns registered_tests if successful, or + // aborts the program otherwise. + const char* VerifyRegisteredTestNames( + const char* file, int line, const char* registered_tests); + + private: + typedef ::std::map RegisteredTestsMap; + + bool registered_; + RegisteredTestsMap registered_tests_; +}; + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +using TypedTestCasePState = TypedTestSuitePState; +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +// Skips to the first non-space char after the first comma in 'str'; +// returns NULL if no comma is found in 'str'. +inline const char* SkipComma(const char* str) { + const char* comma = strchr(str, ','); + if (comma == nullptr) { + return nullptr; + } + while (IsSpace(*(++comma))) {} + return comma; +} + +// Returns the prefix of 'str' before the first comma in it; returns +// the entire string if it contains no comma. +inline std::string GetPrefixUntilComma(const char* str) { + const char* comma = strchr(str, ','); + return comma == nullptr ? str : std::string(str, comma); +} + +// Splits a given string on a given delimiter, populating a given +// vector with the fields. +void SplitString(const ::std::string& str, char delimiter, + ::std::vector< ::std::string>* dest); + +// The default argument to the template below for the case when the user does +// not provide a name generator. +struct DefaultNameGenerator { + template + static std::string GetName(int i) { + return StreamableToString(i); + } +}; + +template +struct NameGeneratorSelector { + typedef Provided type; +}; + +template +void GenerateNamesRecursively(Types0, std::vector*, int) {} + +template +void GenerateNamesRecursively(Types, std::vector* result, int i) { + result->push_back(NameGenerator::template GetName(i)); + GenerateNamesRecursively(typename Types::Tail(), result, + i + 1); +} + +template +std::vector GenerateNames() { + std::vector result; + GenerateNamesRecursively(Types(), &result, 0); + return result; +} + +// TypeParameterizedTest::Register() +// registers a list of type-parameterized tests with Google Test. The +// return value is insignificant - we just need to return something +// such that we can call this function in a namespace scope. +// +// Implementation note: The GTEST_TEMPLATE_ macro declares a template +// template parameter. It's defined in gtest-type-util.h. +template +class TypeParameterizedTest { + public: + // 'index' is the index of the test in the type list 'Types' + // specified in INSTANTIATE_TYPED_TEST_SUITE_P(Prefix, TestSuite, + // Types). Valid values for 'index' are [0, N - 1] where N is the + // length of Types. + static bool Register(const char* prefix, const CodeLocation& code_location, + const char* case_name, const char* test_names, int index, + const std::vector& type_names = + GenerateNames()) { + typedef typename Types::Head Type; + typedef Fixture FixtureClass; + typedef typename GTEST_BIND_(TestSel, Type) TestClass; + + // First, registers the first type-parameterized test in the type + // list. + MakeAndRegisterTestInfo( + (std::string(prefix) + (prefix[0] == '\0' ? "" : "/") + case_name + + "/" + type_names[static_cast(index)]) + .c_str(), + StripTrailingSpaces(GetPrefixUntilComma(test_names)).c_str(), + GetTypeName().c_str(), + nullptr, // No value parameter. + code_location, GetTypeId(), + SuiteApiResolver::GetSetUpCaseOrSuite( + code_location.file.c_str(), code_location.line), + SuiteApiResolver::GetTearDownCaseOrSuite( + code_location.file.c_str(), code_location.line), + new TestFactoryImpl); + + // Next, recurses (at compile time) with the tail of the type list. + return TypeParameterizedTest::Register(prefix, + code_location, + case_name, + test_names, + index + 1, + type_names); + } +}; + +// The base case for the compile time recursion. +template +class TypeParameterizedTest { + public: + static bool Register(const char* /*prefix*/, const CodeLocation&, + const char* /*case_name*/, const char* /*test_names*/, + int /*index*/, + const std::vector& = + std::vector() /*type_names*/) { + return true; + } +}; + +// TypeParameterizedTestSuite::Register() +// registers *all combinations* of 'Tests' and 'Types' with Google +// Test. The return value is insignificant - we just need to return +// something such that we can call this function in a namespace scope. +template +class TypeParameterizedTestSuite { + public: + static bool Register(const char* prefix, CodeLocation code_location, + const TypedTestSuitePState* state, const char* case_name, + const char* test_names, + const std::vector& type_names = + GenerateNames()) { + std::string test_name = StripTrailingSpaces( + GetPrefixUntilComma(test_names)); + if (!state->TestExists(test_name)) { + fprintf(stderr, "Failed to get code location for test %s.%s at %s.", + case_name, test_name.c_str(), + FormatFileLocation(code_location.file.c_str(), + code_location.line).c_str()); + fflush(stderr); + posix::Abort(); + } + const CodeLocation& test_location = state->GetCodeLocation(test_name); + + typedef typename Tests::Head Head; + + // First, register the first test in 'Test' for each type in 'Types'. + TypeParameterizedTest::Register( + prefix, test_location, case_name, test_names, 0, type_names); + + // Next, recurses (at compile time) with the tail of the test list. + return TypeParameterizedTestSuite::Register(prefix, code_location, + state, case_name, + SkipComma(test_names), + type_names); + } +}; + +// The base case for the compile time recursion. +template +class TypeParameterizedTestSuite { + public: + static bool Register(const char* /*prefix*/, const CodeLocation&, + const TypedTestSuitePState* /*state*/, + const char* /*case_name*/, const char* /*test_names*/, + const std::vector& = + std::vector() /*type_names*/) { + return true; + } +}; + +#endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P + +// Returns the current OS stack trace as an std::string. +// +// The maximum number of stack frames to be included is specified by +// the gtest_stack_trace_depth flag. The skip_count parameter +// specifies the number of top frames to be skipped, which doesn't +// count against the number of frames to be included. +// +// For example, if Foo() calls Bar(), which in turn calls +// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in +// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't. +GTEST_API_ std::string GetCurrentOsStackTraceExceptTop( + UnitTest* unit_test, int skip_count); + +// Helpers for suppressing warnings on unreachable code or constant +// condition. + +// Always returns true. +GTEST_API_ bool AlwaysTrue(); + +// Always returns false. +inline bool AlwaysFalse() { return !AlwaysTrue(); } + +// Helper for suppressing false warning from Clang on a const char* +// variable declared in a conditional expression always being NULL in +// the else branch. +struct GTEST_API_ ConstCharPtr { + ConstCharPtr(const char* str) : value(str) {} + operator bool() const { return true; } + const char* value; +}; + +// A simple Linear Congruential Generator for generating random +// numbers with a uniform distribution. Unlike rand() and srand(), it +// doesn't use global state (and therefore can't interfere with user +// code). Unlike rand_r(), it's portable. An LCG isn't very random, +// but it's good enough for our purposes. +class GTEST_API_ Random { + public: + static const UInt32 kMaxRange = 1u << 31; + + explicit Random(UInt32 seed) : state_(seed) {} + + void Reseed(UInt32 seed) { state_ = seed; } + + // Generates a random number from [0, range). Crashes if 'range' is + // 0 or greater than kMaxRange. + UInt32 Generate(UInt32 range); + + private: + UInt32 state_; + GTEST_DISALLOW_COPY_AND_ASSIGN_(Random); +}; + +// Defining a variable of type CompileAssertTypesEqual will cause a +// compiler error iff T1 and T2 are different types. +template +struct CompileAssertTypesEqual; + +template +struct CompileAssertTypesEqual { +}; + +// Removes the reference from a type if it is a reference type, +// otherwise leaves it unchanged. This is the same as +// tr1::remove_reference, which is not widely available yet. +template +struct RemoveReference { typedef T type; }; // NOLINT +template +struct RemoveReference { typedef T type; }; // NOLINT + +// A handy wrapper around RemoveReference that works when the argument +// T depends on template parameters. +#define GTEST_REMOVE_REFERENCE_(T) \ + typename ::testing::internal::RemoveReference::type + +// Removes const from a type if it is a const type, otherwise leaves +// it unchanged. This is the same as tr1::remove_const, which is not +// widely available yet. +template +struct RemoveConst { typedef T type; }; // NOLINT +template +struct RemoveConst { typedef T type; }; // NOLINT + +// MSVC 8.0, Sun C++, and IBM XL C++ have a bug which causes the above +// definition to fail to remove the const in 'const int[3]' and 'const +// char[3][4]'. The following specialization works around the bug. +template +struct RemoveConst { + typedef typename RemoveConst::type type[N]; +}; + +// A handy wrapper around RemoveConst that works when the argument +// T depends on template parameters. +#define GTEST_REMOVE_CONST_(T) \ + typename ::testing::internal::RemoveConst::type + +// Turns const U&, U&, const U, and U all into U. +#define GTEST_REMOVE_REFERENCE_AND_CONST_(T) \ + GTEST_REMOVE_CONST_(GTEST_REMOVE_REFERENCE_(T)) + +// IsAProtocolMessage::value is a compile-time bool constant that's +// true iff T is type proto2::Message or a subclass of it. +template +struct IsAProtocolMessage + : public bool_constant< + std::is_convertible::value> { +}; + +// When the compiler sees expression IsContainerTest(0), if C is an +// STL-style container class, the first overload of IsContainerTest +// will be viable (since both C::iterator* and C::const_iterator* are +// valid types and NULL can be implicitly converted to them). It will +// be picked over the second overload as 'int' is a perfect match for +// the type of argument 0. If C::iterator or C::const_iterator is not +// a valid type, the first overload is not viable, and the second +// overload will be picked. Therefore, we can determine whether C is +// a container class by checking the type of IsContainerTest(0). +// The value of the expression is insignificant. +// +// In C++11 mode we check the existence of a const_iterator and that an +// iterator is properly implemented for the container. +// +// For pre-C++11 that we look for both C::iterator and C::const_iterator. +// The reason is that C++ injects the name of a class as a member of the +// class itself (e.g. you can refer to class iterator as either +// 'iterator' or 'iterator::iterator'). If we look for C::iterator +// only, for example, we would mistakenly think that a class named +// iterator is an STL container. +// +// Also note that the simpler approach of overloading +// IsContainerTest(typename C::const_iterator*) and +// IsContainerTest(...) doesn't work with Visual Age C++ and Sun C++. +typedef int IsContainer; +template ().begin()), + class = decltype(::std::declval().end()), + class = decltype(++::std::declval()), + class = decltype(*::std::declval()), + class = typename C::const_iterator> +IsContainer IsContainerTest(int /* dummy */) { + return 0; +} + +typedef char IsNotContainer; +template +IsNotContainer IsContainerTest(long /* dummy */) { return '\0'; } + +// Trait to detect whether a type T is a hash table. +// The heuristic used is that the type contains an inner type `hasher` and does +// not contain an inner type `reverse_iterator`. +// If the container is iterable in reverse, then order might actually matter. +template +struct IsHashTable { + private: + template + static char test(typename U::hasher*, typename U::reverse_iterator*); + template + static int test(typename U::hasher*, ...); + template + static char test(...); + + public: + static const bool value = sizeof(test(nullptr, nullptr)) == sizeof(int); +}; + +template +const bool IsHashTable::value; + +template (0)) == sizeof(IsContainer)> +struct IsRecursiveContainerImpl; + +template +struct IsRecursiveContainerImpl : public false_type {}; + +// Since the IsRecursiveContainerImpl depends on the IsContainerTest we need to +// obey the same inconsistencies as the IsContainerTest, namely check if +// something is a container is relying on only const_iterator in C++11 and +// is relying on both const_iterator and iterator otherwise +template +struct IsRecursiveContainerImpl { + using value_type = decltype(*std::declval()); + using type = + is_same::type>::type, + C>; +}; + +// IsRecursiveContainer is a unary compile-time predicate that +// evaluates whether C is a recursive container type. A recursive container +// type is a container type whose value_type is equal to the container type +// itself. An example for a recursive container type is +// boost::filesystem::path, whose iterator has a value_type that is equal to +// boost::filesystem::path. +template +struct IsRecursiveContainer : public IsRecursiveContainerImpl::type {}; + +// EnableIf::type is void when 'Cond' is true, and +// undefined when 'Cond' is false. To use SFINAE to make a function +// overload only apply when a particular expression is true, add +// "typename EnableIf::type* = 0" as the last parameter. +template struct EnableIf; +template<> struct EnableIf { typedef void type; }; // NOLINT + +// Utilities for native arrays. + +// ArrayEq() compares two k-dimensional native arrays using the +// elements' operator==, where k can be any integer >= 0. When k is +// 0, ArrayEq() degenerates into comparing a single pair of values. + +template +bool ArrayEq(const T* lhs, size_t size, const U* rhs); + +// This generic version is used when k is 0. +template +inline bool ArrayEq(const T& lhs, const U& rhs) { return lhs == rhs; } + +// This overload is used when k >= 1. +template +inline bool ArrayEq(const T(&lhs)[N], const U(&rhs)[N]) { + return internal::ArrayEq(lhs, N, rhs); +} + +// This helper reduces code bloat. If we instead put its logic inside +// the previous ArrayEq() function, arrays with different sizes would +// lead to different copies of the template code. +template +bool ArrayEq(const T* lhs, size_t size, const U* rhs) { + for (size_t i = 0; i != size; i++) { + if (!internal::ArrayEq(lhs[i], rhs[i])) + return false; + } + return true; +} + +// Finds the first element in the iterator range [begin, end) that +// equals elem. Element may be a native array type itself. +template +Iter ArrayAwareFind(Iter begin, Iter end, const Element& elem) { + for (Iter it = begin; it != end; ++it) { + if (internal::ArrayEq(*it, elem)) + return it; + } + return end; +} + +// CopyArray() copies a k-dimensional native array using the elements' +// operator=, where k can be any integer >= 0. When k is 0, +// CopyArray() degenerates into copying a single value. + +template +void CopyArray(const T* from, size_t size, U* to); + +// This generic version is used when k is 0. +template +inline void CopyArray(const T& from, U* to) { *to = from; } + +// This overload is used when k >= 1. +template +inline void CopyArray(const T(&from)[N], U(*to)[N]) { + internal::CopyArray(from, N, *to); +} + +// This helper reduces code bloat. If we instead put its logic inside +// the previous CopyArray() function, arrays with different sizes +// would lead to different copies of the template code. +template +void CopyArray(const T* from, size_t size, U* to) { + for (size_t i = 0; i != size; i++) { + internal::CopyArray(from[i], to + i); + } +} + +// The relation between an NativeArray object (see below) and the +// native array it represents. +// We use 2 different structs to allow non-copyable types to be used, as long +// as RelationToSourceReference() is passed. +struct RelationToSourceReference {}; +struct RelationToSourceCopy {}; + +// Adapts a native array to a read-only STL-style container. Instead +// of the complete STL container concept, this adaptor only implements +// members useful for Google Mock's container matchers. New members +// should be added as needed. To simplify the implementation, we only +// support Element being a raw type (i.e. having no top-level const or +// reference modifier). It's the client's responsibility to satisfy +// this requirement. Element can be an array type itself (hence +// multi-dimensional arrays are supported). +template +class NativeArray { + public: + // STL-style container typedefs. + typedef Element value_type; + typedef Element* iterator; + typedef const Element* const_iterator; + + // Constructs from a native array. References the source. + NativeArray(const Element* array, size_t count, RelationToSourceReference) { + InitRef(array, count); + } + + // Constructs from a native array. Copies the source. + NativeArray(const Element* array, size_t count, RelationToSourceCopy) { + InitCopy(array, count); + } + + // Copy constructor. + NativeArray(const NativeArray& rhs) { + (this->*rhs.clone_)(rhs.array_, rhs.size_); + } + + ~NativeArray() { + if (clone_ != &NativeArray::InitRef) + delete[] array_; + } + + // STL-style container methods. + size_t size() const { return size_; } + const_iterator begin() const { return array_; } + const_iterator end() const { return array_ + size_; } + bool operator==(const NativeArray& rhs) const { + return size() == rhs.size() && + ArrayEq(begin(), size(), rhs.begin()); + } + + private: + enum { + kCheckTypeIsNotConstOrAReference = StaticAssertTypeEqHelper< + Element, GTEST_REMOVE_REFERENCE_AND_CONST_(Element)>::value + }; + + // Initializes this object with a copy of the input. + void InitCopy(const Element* array, size_t a_size) { + Element* const copy = new Element[a_size]; + CopyArray(array, a_size, copy); + array_ = copy; + size_ = a_size; + clone_ = &NativeArray::InitCopy; + } + + // Initializes this object with a reference of the input. + void InitRef(const Element* array, size_t a_size) { + array_ = array; + size_ = a_size; + clone_ = &NativeArray::InitRef; + } + + const Element* array_; + size_t size_; + void (NativeArray::*clone_)(const Element*, size_t); + + GTEST_DISALLOW_ASSIGN_(NativeArray); +}; + +// Backport of std::index_sequence. +template +struct IndexSequence { + using type = IndexSequence; +}; + +// Double the IndexSequence, and one if plus_one is true. +template +struct DoubleSequence; +template +struct DoubleSequence, sizeofT> { + using type = IndexSequence; +}; +template +struct DoubleSequence, sizeofT> { + using type = IndexSequence; +}; + +// Backport of std::make_index_sequence. +// It uses O(ln(N)) instantiation depth. +template +struct MakeIndexSequence + : DoubleSequence::type, + N / 2>::type {}; + +template <> +struct MakeIndexSequence<0> : IndexSequence<> {}; + +// FIXME: This implementation of ElemFromList is O(1) in instantiation depth, +// but it is O(N^2) in total instantiations. Not sure if this is the best +// tradeoff, as it will make it somewhat slow to compile. +template +struct ElemFromListImpl {}; + +template +struct ElemFromListImpl { + using type = T; +}; + +// Get the Nth element from T... +// It uses O(1) instantiation depth. +template +struct ElemFromList; + +template +struct ElemFromList, T...> + : ElemFromListImpl... {}; + +template +class FlatTuple; + +template +struct FlatTupleElemBase; + +template +struct FlatTupleElemBase, I> { + using value_type = + typename ElemFromList::type, + T...>::type; + FlatTupleElemBase() = default; + explicit FlatTupleElemBase(value_type t) : value(std::move(t)) {} + value_type value; +}; + +template +struct FlatTupleBase; + +template +struct FlatTupleBase, IndexSequence> + : FlatTupleElemBase, Idx>... { + using Indices = IndexSequence; + FlatTupleBase() = default; + explicit FlatTupleBase(T... t) + : FlatTupleElemBase, Idx>(std::move(t))... {} +}; + +// Analog to std::tuple but with different tradeoffs. +// This class minimizes the template instantiation depth, thus allowing more +// elements that std::tuple would. std::tuple has been seen to require an +// instantiation depth of more than 10x the number of elements in some +// implementations. +// FlatTuple and ElemFromList are not recursive and have a fixed depth +// regardless of T... +// MakeIndexSequence, on the other hand, it is recursive but with an +// instantiation depth of O(ln(N)). +template +class FlatTuple + : private FlatTupleBase, + typename MakeIndexSequence::type> { + using Indices = typename FlatTuple::FlatTupleBase::Indices; + + public: + FlatTuple() = default; + explicit FlatTuple(T... t) : FlatTuple::FlatTupleBase(std::move(t)...) {} + + template + const typename ElemFromList::type& Get() const { + return static_cast*>(this)->value; + } + + template + typename ElemFromList::type& Get() { + return static_cast*>(this)->value; + } +}; + +// Utility functions to be called with static_assert to induce deprecation +// warnings. +GTEST_INTERNAL_DEPRECATED( + "INSTANTIATE_TEST_CASE_P is deprecated, please use " + "INSTANTIATE_TEST_SUITE_P") +constexpr bool InstantiateTestCase_P_IsDeprecated() { return true; } + +GTEST_INTERNAL_DEPRECATED( + "TYPED_TEST_CASE_P is deprecated, please use " + "TYPED_TEST_SUITE_P") +constexpr bool TypedTestCase_P_IsDeprecated() { return true; } + +GTEST_INTERNAL_DEPRECATED( + "TYPED_TEST_CASE is deprecated, please use " + "TYPED_TEST_SUITE") +constexpr bool TypedTestCaseIsDeprecated() { return true; } + +GTEST_INTERNAL_DEPRECATED( + "REGISTER_TYPED_TEST_CASE_P is deprecated, please use " + "REGISTER_TYPED_TEST_SUITE_P") +constexpr bool RegisterTypedTestCase_P_IsDeprecated() { return true; } + +GTEST_INTERNAL_DEPRECATED( + "INSTANTIATE_TYPED_TEST_CASE_P is deprecated, please use " + "INSTANTIATE_TYPED_TEST_SUITE_P") +constexpr bool InstantiateTypedTestCase_P_IsDeprecated() { return true; } + +} // namespace internal +} // namespace testing + +#define GTEST_MESSAGE_AT_(file, line, message, result_type) \ + ::testing::internal::AssertHelper(result_type, file, line, message) \ + = ::testing::Message() + +#define GTEST_MESSAGE_(message, result_type) \ + GTEST_MESSAGE_AT_(__FILE__, __LINE__, message, result_type) + +#define GTEST_FATAL_FAILURE_(message) \ + return GTEST_MESSAGE_(message, ::testing::TestPartResult::kFatalFailure) + +#define GTEST_NONFATAL_FAILURE_(message) \ + GTEST_MESSAGE_(message, ::testing::TestPartResult::kNonFatalFailure) + +#define GTEST_SUCCESS_(message) \ + GTEST_MESSAGE_(message, ::testing::TestPartResult::kSuccess) + +#define GTEST_SKIP_(message) \ + return GTEST_MESSAGE_(message, ::testing::TestPartResult::kSkip) + +// Suppress MSVC warning 4072 (unreachable code) for the code following +// statement if it returns or throws (or doesn't return or throw in some +// situations). +#define GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) \ + if (::testing::internal::AlwaysTrue()) { statement; } + +#define GTEST_TEST_THROW_(statement, expected_exception, fail) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::ConstCharPtr gtest_msg = "") { \ + bool gtest_caught_expected = false; \ + try { \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + } \ + catch (expected_exception const&) { \ + gtest_caught_expected = true; \ + } \ + catch (...) { \ + gtest_msg.value = \ + "Expected: " #statement " throws an exception of type " \ + #expected_exception ".\n Actual: it throws a different type."; \ + goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \ + } \ + if (!gtest_caught_expected) { \ + gtest_msg.value = \ + "Expected: " #statement " throws an exception of type " \ + #expected_exception ".\n Actual: it throws nothing."; \ + goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \ + } \ + } else \ + GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__): \ + fail(gtest_msg.value) + +#define GTEST_TEST_NO_THROW_(statement, fail) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::AlwaysTrue()) { \ + try { \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + } \ + catch (...) { \ + goto GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__); \ + } \ + } else \ + GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__): \ + fail("Expected: " #statement " doesn't throw an exception.\n" \ + " Actual: it throws.") + +#define GTEST_TEST_ANY_THROW_(statement, fail) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::AlwaysTrue()) { \ + bool gtest_caught_any = false; \ + try { \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + } \ + catch (...) { \ + gtest_caught_any = true; \ + } \ + if (!gtest_caught_any) { \ + goto GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__); \ + } \ + } else \ + GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__): \ + fail("Expected: " #statement " throws an exception.\n" \ + " Actual: it doesn't.") + + +// Implements Boolean test assertions such as EXPECT_TRUE. expression can be +// either a boolean expression or an AssertionResult. text is a textual +// represenation of expression as it was passed into the EXPECT_TRUE. +#define GTEST_TEST_BOOLEAN_(expression, text, actual, expected, fail) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (const ::testing::AssertionResult gtest_ar_ = \ + ::testing::AssertionResult(expression)) \ + ; \ + else \ + fail(::testing::internal::GetBoolAssertionFailureMessage(\ + gtest_ar_, text, #actual, #expected).c_str()) + +#define GTEST_TEST_NO_FATAL_FAILURE_(statement, fail) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::AlwaysTrue()) { \ + ::testing::internal::HasNewFatalFailureHelper gtest_fatal_failure_checker; \ + GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \ + if (gtest_fatal_failure_checker.has_new_fatal_failure()) { \ + goto GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__); \ + } \ + } else \ + GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__): \ + fail("Expected: " #statement " doesn't generate new fatal " \ + "failures in the current thread.\n" \ + " Actual: it does.") + +// Expands to the name of the class that implements the given test. +#define GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) \ + test_suite_name##_##test_name##_Test + +// Helper macro for defining tests. +#define GTEST_TEST_(test_suite_name, test_name, parent_class, parent_id) \ + class GTEST_TEST_CLASS_NAME_(test_suite_name, test_name) \ + : public parent_class { \ + public: \ + GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)() {} \ + \ + private: \ + virtual void TestBody(); \ + static ::testing::TestInfo* const test_info_ GTEST_ATTRIBUTE_UNUSED_; \ + GTEST_DISALLOW_COPY_AND_ASSIGN_(GTEST_TEST_CLASS_NAME_(test_suite_name, \ + test_name)); \ + }; \ + \ + ::testing::TestInfo* const GTEST_TEST_CLASS_NAME_(test_suite_name, \ + test_name)::test_info_ = \ + ::testing::internal::MakeAndRegisterTestInfo( \ + #test_suite_name, #test_name, nullptr, nullptr, \ + ::testing::internal::CodeLocation(__FILE__, __LINE__), (parent_id), \ + ::testing::internal::SuiteApiResolver< \ + parent_class>::GetSetUpCaseOrSuite(__FILE__, __LINE__), \ + ::testing::internal::SuiteApiResolver< \ + parent_class>::GetTearDownCaseOrSuite(__FILE__, __LINE__), \ + new ::testing::internal::TestFactoryImpl); \ + void GTEST_TEST_CLASS_NAME_(test_suite_name, test_name)::TestBody() + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-param-util.h b/deps/googletest/include/gtest/internal/gtest-param-util.h new file mode 100644 index 000000000..e900b3ffb --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-param-util.h @@ -0,0 +1,880 @@ +// Copyright 2008 Google Inc. +// All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +// Type and function utilities for implementing parameterized tests. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_ + +#include + +#include +#include +#include +#include +#include +#include +#include + +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-port.h" +#include "gtest/gtest-printers.h" + +namespace testing { +// Input to a parameterized test name generator, describing a test parameter. +// Consists of the parameter value and the integer parameter index. +template +struct TestParamInfo { + TestParamInfo(const ParamType& a_param, size_t an_index) : + param(a_param), + index(an_index) {} + ParamType param; + size_t index; +}; + +// A builtin parameterized test name generator which returns the result of +// testing::PrintToString. +struct PrintToStringParamName { + template + std::string operator()(const TestParamInfo& info) const { + return PrintToString(info.param); + } +}; + +namespace internal { + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// Utility Functions + +// Outputs a message explaining invalid registration of different +// fixture class for the same test suite. This may happen when +// TEST_P macro is used to define two tests with the same name +// but in different namespaces. +GTEST_API_ void ReportInvalidTestSuiteType(const char* test_suite_name, + CodeLocation code_location); + +template class ParamGeneratorInterface; +template class ParamGenerator; + +// Interface for iterating over elements provided by an implementation +// of ParamGeneratorInterface. +template +class ParamIteratorInterface { + public: + virtual ~ParamIteratorInterface() {} + // A pointer to the base generator instance. + // Used only for the purposes of iterator comparison + // to make sure that two iterators belong to the same generator. + virtual const ParamGeneratorInterface* BaseGenerator() const = 0; + // Advances iterator to point to the next element + // provided by the generator. The caller is responsible + // for not calling Advance() on an iterator equal to + // BaseGenerator()->End(). + virtual void Advance() = 0; + // Clones the iterator object. Used for implementing copy semantics + // of ParamIterator. + virtual ParamIteratorInterface* Clone() const = 0; + // Dereferences the current iterator and provides (read-only) access + // to the pointed value. It is the caller's responsibility not to call + // Current() on an iterator equal to BaseGenerator()->End(). + // Used for implementing ParamGenerator::operator*(). + virtual const T* Current() const = 0; + // Determines whether the given iterator and other point to the same + // element in the sequence generated by the generator. + // Used for implementing ParamGenerator::operator==(). + virtual bool Equals(const ParamIteratorInterface& other) const = 0; +}; + +// Class iterating over elements provided by an implementation of +// ParamGeneratorInterface. It wraps ParamIteratorInterface +// and implements the const forward iterator concept. +template +class ParamIterator { + public: + typedef T value_type; + typedef const T& reference; + typedef ptrdiff_t difference_type; + + // ParamIterator assumes ownership of the impl_ pointer. + ParamIterator(const ParamIterator& other) : impl_(other.impl_->Clone()) {} + ParamIterator& operator=(const ParamIterator& other) { + if (this != &other) + impl_.reset(other.impl_->Clone()); + return *this; + } + + const T& operator*() const { return *impl_->Current(); } + const T* operator->() const { return impl_->Current(); } + // Prefix version of operator++. + ParamIterator& operator++() { + impl_->Advance(); + return *this; + } + // Postfix version of operator++. + ParamIterator operator++(int /*unused*/) { + ParamIteratorInterface* clone = impl_->Clone(); + impl_->Advance(); + return ParamIterator(clone); + } + bool operator==(const ParamIterator& other) const { + return impl_.get() == other.impl_.get() || impl_->Equals(*other.impl_); + } + bool operator!=(const ParamIterator& other) const { + return !(*this == other); + } + + private: + friend class ParamGenerator; + explicit ParamIterator(ParamIteratorInterface* impl) : impl_(impl) {} + std::unique_ptr > impl_; +}; + +// ParamGeneratorInterface is the binary interface to access generators +// defined in other translation units. +template +class ParamGeneratorInterface { + public: + typedef T ParamType; + + virtual ~ParamGeneratorInterface() {} + + // Generator interface definition + virtual ParamIteratorInterface* Begin() const = 0; + virtual ParamIteratorInterface* End() const = 0; +}; + +// Wraps ParamGeneratorInterface and provides general generator syntax +// compatible with the STL Container concept. +// This class implements copy initialization semantics and the contained +// ParamGeneratorInterface instance is shared among all copies +// of the original object. This is possible because that instance is immutable. +template +class ParamGenerator { + public: + typedef ParamIterator iterator; + + explicit ParamGenerator(ParamGeneratorInterface* impl) : impl_(impl) {} + ParamGenerator(const ParamGenerator& other) : impl_(other.impl_) {} + + ParamGenerator& operator=(const ParamGenerator& other) { + impl_ = other.impl_; + return *this; + } + + iterator begin() const { return iterator(impl_->Begin()); } + iterator end() const { return iterator(impl_->End()); } + + private: + std::shared_ptr > impl_; +}; + +// Generates values from a range of two comparable values. Can be used to +// generate sequences of user-defined types that implement operator+() and +// operator<(). +// This class is used in the Range() function. +template +class RangeGenerator : public ParamGeneratorInterface { + public: + RangeGenerator(T begin, T end, IncrementT step) + : begin_(begin), end_(end), + step_(step), end_index_(CalculateEndIndex(begin, end, step)) {} + ~RangeGenerator() override {} + + ParamIteratorInterface* Begin() const override { + return new Iterator(this, begin_, 0, step_); + } + ParamIteratorInterface* End() const override { + return new Iterator(this, end_, end_index_, step_); + } + + private: + class Iterator : public ParamIteratorInterface { + public: + Iterator(const ParamGeneratorInterface* base, T value, int index, + IncrementT step) + : base_(base), value_(value), index_(index), step_(step) {} + ~Iterator() override {} + + const ParamGeneratorInterface* BaseGenerator() const override { + return base_; + } + void Advance() override { + value_ = static_cast(value_ + step_); + index_++; + } + ParamIteratorInterface* Clone() const override { + return new Iterator(*this); + } + const T* Current() const override { return &value_; } + bool Equals(const ParamIteratorInterface& other) const override { + // Having the same base generator guarantees that the other + // iterator is of the same type and we can downcast. + GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) + << "The program attempted to compare iterators " + << "from different generators." << std::endl; + const int other_index = + CheckedDowncastToActualType(&other)->index_; + return index_ == other_index; + } + + private: + Iterator(const Iterator& other) + : ParamIteratorInterface(), + base_(other.base_), value_(other.value_), index_(other.index_), + step_(other.step_) {} + + // No implementation - assignment is unsupported. + void operator=(const Iterator& other); + + const ParamGeneratorInterface* const base_; + T value_; + int index_; + const IncrementT step_; + }; // class RangeGenerator::Iterator + + static int CalculateEndIndex(const T& begin, + const T& end, + const IncrementT& step) { + int end_index = 0; + for (T i = begin; i < end; i = static_cast(i + step)) + end_index++; + return end_index; + } + + // No implementation - assignment is unsupported. + void operator=(const RangeGenerator& other); + + const T begin_; + const T end_; + const IncrementT step_; + // The index for the end() iterator. All the elements in the generated + // sequence are indexed (0-based) to aid iterator comparison. + const int end_index_; +}; // class RangeGenerator + + +// Generates values from a pair of STL-style iterators. Used in the +// ValuesIn() function. The elements are copied from the source range +// since the source can be located on the stack, and the generator +// is likely to persist beyond that stack frame. +template +class ValuesInIteratorRangeGenerator : public ParamGeneratorInterface { + public: + template + ValuesInIteratorRangeGenerator(ForwardIterator begin, ForwardIterator end) + : container_(begin, end) {} + ~ValuesInIteratorRangeGenerator() override {} + + ParamIteratorInterface* Begin() const override { + return new Iterator(this, container_.begin()); + } + ParamIteratorInterface* End() const override { + return new Iterator(this, container_.end()); + } + + private: + typedef typename ::std::vector ContainerType; + + class Iterator : public ParamIteratorInterface { + public: + Iterator(const ParamGeneratorInterface* base, + typename ContainerType::const_iterator iterator) + : base_(base), iterator_(iterator) {} + ~Iterator() override {} + + const ParamGeneratorInterface* BaseGenerator() const override { + return base_; + } + void Advance() override { + ++iterator_; + value_.reset(); + } + ParamIteratorInterface* Clone() const override { + return new Iterator(*this); + } + // We need to use cached value referenced by iterator_ because *iterator_ + // can return a temporary object (and of type other then T), so just + // having "return &*iterator_;" doesn't work. + // value_ is updated here and not in Advance() because Advance() + // can advance iterator_ beyond the end of the range, and we cannot + // detect that fact. The client code, on the other hand, is + // responsible for not calling Current() on an out-of-range iterator. + const T* Current() const override { + if (value_.get() == nullptr) value_.reset(new T(*iterator_)); + return value_.get(); + } + bool Equals(const ParamIteratorInterface& other) const override { + // Having the same base generator guarantees that the other + // iterator is of the same type and we can downcast. + GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) + << "The program attempted to compare iterators " + << "from different generators." << std::endl; + return iterator_ == + CheckedDowncastToActualType(&other)->iterator_; + } + + private: + Iterator(const Iterator& other) + // The explicit constructor call suppresses a false warning + // emitted by gcc when supplied with the -Wextra option. + : ParamIteratorInterface(), + base_(other.base_), + iterator_(other.iterator_) {} + + const ParamGeneratorInterface* const base_; + typename ContainerType::const_iterator iterator_; + // A cached value of *iterator_. We keep it here to allow access by + // pointer in the wrapping iterator's operator->(). + // value_ needs to be mutable to be accessed in Current(). + // Use of std::unique_ptr helps manage cached value's lifetime, + // which is bound by the lifespan of the iterator itself. + mutable std::unique_ptr value_; + }; // class ValuesInIteratorRangeGenerator::Iterator + + // No implementation - assignment is unsupported. + void operator=(const ValuesInIteratorRangeGenerator& other); + + const ContainerType container_; +}; // class ValuesInIteratorRangeGenerator + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// Default parameterized test name generator, returns a string containing the +// integer test parameter index. +template +std::string DefaultParamName(const TestParamInfo& info) { + Message name_stream; + name_stream << info.index; + return name_stream.GetString(); +} + +template +void TestNotEmpty() { + static_assert(sizeof(T) == 0, "Empty arguments are not allowed."); +} +template +void TestNotEmpty(const T&) {} + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// Stores a parameter value and later creates tests parameterized with that +// value. +template +class ParameterizedTestFactory : public TestFactoryBase { + public: + typedef typename TestClass::ParamType ParamType; + explicit ParameterizedTestFactory(ParamType parameter) : + parameter_(parameter) {} + Test* CreateTest() override { + TestClass::SetParam(¶meter_); + return new TestClass(); + } + + private: + const ParamType parameter_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestFactory); +}; + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// TestMetaFactoryBase is a base class for meta-factories that create +// test factories for passing into MakeAndRegisterTestInfo function. +template +class TestMetaFactoryBase { + public: + virtual ~TestMetaFactoryBase() {} + + virtual TestFactoryBase* CreateTestFactory(ParamType parameter) = 0; +}; + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// TestMetaFactory creates test factories for passing into +// MakeAndRegisterTestInfo function. Since MakeAndRegisterTestInfo receives +// ownership of test factory pointer, same factory object cannot be passed +// into that method twice. But ParameterizedTestSuiteInfo is going to call +// it for each Test/Parameter value combination. Thus it needs meta factory +// creator class. +template +class TestMetaFactory + : public TestMetaFactoryBase { + public: + using ParamType = typename TestSuite::ParamType; + + TestMetaFactory() {} + + TestFactoryBase* CreateTestFactory(ParamType parameter) override { + return new ParameterizedTestFactory(parameter); + } + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestMetaFactory); +}; + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// ParameterizedTestSuiteInfoBase is a generic interface +// to ParameterizedTestSuiteInfo classes. ParameterizedTestSuiteInfoBase +// accumulates test information provided by TEST_P macro invocations +// and generators provided by INSTANTIATE_TEST_SUITE_P macro invocations +// and uses that information to register all resulting test instances +// in RegisterTests method. The ParameterizeTestSuiteRegistry class holds +// a collection of pointers to the ParameterizedTestSuiteInfo objects +// and calls RegisterTests() on each of them when asked. +class ParameterizedTestSuiteInfoBase { + public: + virtual ~ParameterizedTestSuiteInfoBase() {} + + // Base part of test suite name for display purposes. + virtual const std::string& GetTestSuiteName() const = 0; + // Test case id to verify identity. + virtual TypeId GetTestSuiteTypeId() const = 0; + // UnitTest class invokes this method to register tests in this + // test suite right before running them in RUN_ALL_TESTS macro. + // This method should not be called more than once on any single + // instance of a ParameterizedTestSuiteInfoBase derived class. + virtual void RegisterTests() = 0; + + protected: + ParameterizedTestSuiteInfoBase() {} + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestSuiteInfoBase); +}; + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// ParameterizedTestSuiteInfo accumulates tests obtained from TEST_P +// macro invocations for a particular test suite and generators +// obtained from INSTANTIATE_TEST_SUITE_P macro invocations for that +// test suite. It registers tests with all values generated by all +// generators when asked. +template +class ParameterizedTestSuiteInfo : public ParameterizedTestSuiteInfoBase { + public: + // ParamType and GeneratorCreationFunc are private types but are required + // for declarations of public methods AddTestPattern() and + // AddTestSuiteInstantiation(). + using ParamType = typename TestSuite::ParamType; + // A function that returns an instance of appropriate generator type. + typedef ParamGenerator(GeneratorCreationFunc)(); + using ParamNameGeneratorFunc = std::string(const TestParamInfo&); + + explicit ParameterizedTestSuiteInfo(const char* name, + CodeLocation code_location) + : test_suite_name_(name), code_location_(code_location) {} + + // Test case base name for display purposes. + const std::string& GetTestSuiteName() const override { + return test_suite_name_; + } + // Test case id to verify identity. + TypeId GetTestSuiteTypeId() const override { return GetTypeId(); } + // TEST_P macro uses AddTestPattern() to record information + // about a single test in a LocalTestInfo structure. + // test_suite_name is the base name of the test suite (without invocation + // prefix). test_base_name is the name of an individual test without + // parameter index. For the test SequenceA/FooTest.DoBar/1 FooTest is + // test suite base name and DoBar is test base name. + void AddTestPattern(const char* test_suite_name, const char* test_base_name, + TestMetaFactoryBase* meta_factory) { + tests_.push_back(std::shared_ptr( + new TestInfo(test_suite_name, test_base_name, meta_factory))); + } + // INSTANTIATE_TEST_SUITE_P macro uses AddGenerator() to record information + // about a generator. + int AddTestSuiteInstantiation(const std::string& instantiation_name, + GeneratorCreationFunc* func, + ParamNameGeneratorFunc* name_func, + const char* file, int line) { + instantiations_.push_back( + InstantiationInfo(instantiation_name, func, name_func, file, line)); + return 0; // Return value used only to run this method in namespace scope. + } + // UnitTest class invokes this method to register tests in this test suite + // test suites right before running tests in RUN_ALL_TESTS macro. + // This method should not be called more than once on any single + // instance of a ParameterizedTestSuiteInfoBase derived class. + // UnitTest has a guard to prevent from calling this method more than once. + void RegisterTests() override { + for (typename TestInfoContainer::iterator test_it = tests_.begin(); + test_it != tests_.end(); ++test_it) { + std::shared_ptr test_info = *test_it; + for (typename InstantiationContainer::iterator gen_it = + instantiations_.begin(); gen_it != instantiations_.end(); + ++gen_it) { + const std::string& instantiation_name = gen_it->name; + ParamGenerator generator((*gen_it->generator)()); + ParamNameGeneratorFunc* name_func = gen_it->name_func; + const char* file = gen_it->file; + int line = gen_it->line; + + std::string test_suite_name; + if ( !instantiation_name.empty() ) + test_suite_name = instantiation_name + "/"; + test_suite_name += test_info->test_suite_base_name; + + size_t i = 0; + std::set test_param_names; + for (typename ParamGenerator::iterator param_it = + generator.begin(); + param_it != generator.end(); ++param_it, ++i) { + Message test_name_stream; + + std::string param_name = name_func( + TestParamInfo(*param_it, i)); + + GTEST_CHECK_(IsValidParamName(param_name)) + << "Parameterized test name '" << param_name + << "' is invalid, in " << file + << " line " << line << std::endl; + + GTEST_CHECK_(test_param_names.count(param_name) == 0) + << "Duplicate parameterized test name '" << param_name + << "', in " << file << " line " << line << std::endl; + + test_param_names.insert(param_name); + + test_name_stream << test_info->test_base_name << "/" << param_name; + MakeAndRegisterTestInfo( + test_suite_name.c_str(), test_name_stream.GetString().c_str(), + nullptr, // No type parameter. + PrintToString(*param_it).c_str(), code_location_, + GetTestSuiteTypeId(), + SuiteApiResolver::GetSetUpCaseOrSuite(file, line), + SuiteApiResolver::GetTearDownCaseOrSuite(file, line), + test_info->test_meta_factory->CreateTestFactory(*param_it)); + } // for param_it + } // for gen_it + } // for test_it + } // RegisterTests + + private: + // LocalTestInfo structure keeps information about a single test registered + // with TEST_P macro. + struct TestInfo { + TestInfo(const char* a_test_suite_base_name, const char* a_test_base_name, + TestMetaFactoryBase* a_test_meta_factory) + : test_suite_base_name(a_test_suite_base_name), + test_base_name(a_test_base_name), + test_meta_factory(a_test_meta_factory) {} + + const std::string test_suite_base_name; + const std::string test_base_name; + const std::unique_ptr > test_meta_factory; + }; + using TestInfoContainer = ::std::vector >; + // Records data received from INSTANTIATE_TEST_SUITE_P macros: + // + struct InstantiationInfo { + InstantiationInfo(const std::string &name_in, + GeneratorCreationFunc* generator_in, + ParamNameGeneratorFunc* name_func_in, + const char* file_in, + int line_in) + : name(name_in), + generator(generator_in), + name_func(name_func_in), + file(file_in), + line(line_in) {} + + std::string name; + GeneratorCreationFunc* generator; + ParamNameGeneratorFunc* name_func; + const char* file; + int line; + }; + typedef ::std::vector InstantiationContainer; + + static bool IsValidParamName(const std::string& name) { + // Check for empty string + if (name.empty()) + return false; + + // Check for invalid characters + for (std::string::size_type index = 0; index < name.size(); ++index) { + if (!isalnum(name[index]) && name[index] != '_') + return false; + } + + return true; + } + + const std::string test_suite_name_; + CodeLocation code_location_; + TestInfoContainer tests_; + InstantiationContainer instantiations_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestSuiteInfo); +}; // class ParameterizedTestSuiteInfo + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +template +using ParameterizedTestCaseInfo = ParameterizedTestSuiteInfo; +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// ParameterizedTestSuiteRegistry contains a map of +// ParameterizedTestSuiteInfoBase classes accessed by test suite names. TEST_P +// and INSTANTIATE_TEST_SUITE_P macros use it to locate their corresponding +// ParameterizedTestSuiteInfo descriptors. +class ParameterizedTestSuiteRegistry { + public: + ParameterizedTestSuiteRegistry() {} + ~ParameterizedTestSuiteRegistry() { + for (auto& test_suite_info : test_suite_infos_) { + delete test_suite_info; + } + } + + // Looks up or creates and returns a structure containing information about + // tests and instantiations of a particular test suite. + template + ParameterizedTestSuiteInfo* GetTestSuitePatternHolder( + const char* test_suite_name, CodeLocation code_location) { + ParameterizedTestSuiteInfo* typed_test_info = nullptr; + for (auto& test_suite_info : test_suite_infos_) { + if (test_suite_info->GetTestSuiteName() == test_suite_name) { + if (test_suite_info->GetTestSuiteTypeId() != GetTypeId()) { + // Complain about incorrect usage of Google Test facilities + // and terminate the program since we cannot guaranty correct + // test suite setup and tear-down in this case. + ReportInvalidTestSuiteType(test_suite_name, code_location); + posix::Abort(); + } else { + // At this point we are sure that the object we found is of the same + // type we are looking for, so we downcast it to that type + // without further checks. + typed_test_info = CheckedDowncastToActualType< + ParameterizedTestSuiteInfo >(test_suite_info); + } + break; + } + } + if (typed_test_info == nullptr) { + typed_test_info = new ParameterizedTestSuiteInfo( + test_suite_name, code_location); + test_suite_infos_.push_back(typed_test_info); + } + return typed_test_info; + } + void RegisterTests() { + for (auto& test_suite_info : test_suite_infos_) { + test_suite_info->RegisterTests(); + } + } +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + template + ParameterizedTestCaseInfo* GetTestCasePatternHolder( + const char* test_case_name, CodeLocation code_location) { + return GetTestSuitePatternHolder(test_case_name, code_location); + } + +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + private: + using TestSuiteInfoContainer = ::std::vector; + + TestSuiteInfoContainer test_suite_infos_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestSuiteRegistry); +}; + +} // namespace internal + +// Forward declarations of ValuesIn(), which is implemented in +// include/gtest/gtest-param-test.h. +template +internal::ParamGenerator ValuesIn( + const Container& container); + +namespace internal { +// Used in the Values() function to provide polymorphic capabilities. + +template +class ValueArray { + public: + ValueArray(Ts... v) : v_{std::move(v)...} {} + + template + operator ParamGenerator() const { // NOLINT + return ValuesIn(MakeVector(MakeIndexSequence())); + } + + private: + template + std::vector MakeVector(IndexSequence) const { + return std::vector{static_cast(v_.template Get())...}; + } + + FlatTuple v_; +}; + +template +class CartesianProductGenerator + : public ParamGeneratorInterface<::std::tuple> { + public: + typedef ::std::tuple ParamType; + + CartesianProductGenerator(const std::tuple...>& g) + : generators_(g) {} + ~CartesianProductGenerator() override {} + + ParamIteratorInterface* Begin() const override { + return new Iterator(this, generators_, false); + } + ParamIteratorInterface* End() const override { + return new Iterator(this, generators_, true); + } + + private: + template + class IteratorImpl; + template + class IteratorImpl> + : public ParamIteratorInterface { + public: + IteratorImpl(const ParamGeneratorInterface* base, + const std::tuple...>& generators, bool is_end) + : base_(base), + begin_(std::get(generators).begin()...), + end_(std::get(generators).end()...), + current_(is_end ? end_ : begin_) { + ComputeCurrentValue(); + } + ~IteratorImpl() override {} + + const ParamGeneratorInterface* BaseGenerator() const override { + return base_; + } + // Advance should not be called on beyond-of-range iterators + // so no component iterators must be beyond end of range, either. + void Advance() override { + assert(!AtEnd()); + // Advance the last iterator. + ++std::get(current_); + // if that reaches end, propagate that up. + AdvanceIfEnd(); + ComputeCurrentValue(); + } + ParamIteratorInterface* Clone() const override { + return new IteratorImpl(*this); + } + + const ParamType* Current() const override { return current_value_.get(); } + + bool Equals(const ParamIteratorInterface& other) const override { + // Having the same base generator guarantees that the other + // iterator is of the same type and we can downcast. + GTEST_CHECK_(BaseGenerator() == other.BaseGenerator()) + << "The program attempted to compare iterators " + << "from different generators." << std::endl; + const IteratorImpl* typed_other = + CheckedDowncastToActualType(&other); + + // We must report iterators equal if they both point beyond their + // respective ranges. That can happen in a variety of fashions, + // so we have to consult AtEnd(). + if (AtEnd() && typed_other->AtEnd()) return true; + + bool same = true; + bool dummy[] = { + (same = same && std::get(current_) == + std::get(typed_other->current_))...}; + (void)dummy; + return same; + } + + private: + template + void AdvanceIfEnd() { + if (std::get(current_) != std::get(end_)) return; + + bool last = ThisI == 0; + if (last) { + // We are done. Nothing else to propagate. + return; + } + + constexpr size_t NextI = ThisI - (ThisI != 0); + std::get(current_) = std::get(begin_); + ++std::get(current_); + AdvanceIfEnd(); + } + + void ComputeCurrentValue() { + if (!AtEnd()) + current_value_ = std::make_shared(*std::get(current_)...); + } + bool AtEnd() const { + bool at_end = false; + bool dummy[] = { + (at_end = at_end || std::get(current_) == std::get(end_))...}; + (void)dummy; + return at_end; + } + + const ParamGeneratorInterface* const base_; + std::tuple::iterator...> begin_; + std::tuple::iterator...> end_; + std::tuple::iterator...> current_; + std::shared_ptr current_value_; + }; + + using Iterator = IteratorImpl::type>; + + std::tuple...> generators_; +}; + +template +class CartesianProductHolder { + public: + CartesianProductHolder(const Gen&... g) : generators_(g...) {} + template + operator ParamGenerator<::std::tuple>() const { + return ParamGenerator<::std::tuple>( + new CartesianProductGenerator(generators_)); + } + + private: + std::tuple generators_; +}; + +} // namespace internal +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-port-arch.h b/deps/googletest/include/gtest/internal/gtest-port-arch.h new file mode 100644 index 000000000..cece93dba --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-port-arch.h @@ -0,0 +1,107 @@ +// Copyright 2015, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file defines the GTEST_OS_* macro. +// It is separate from gtest-port.h so that custom/gtest-port.h can include it. + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_ + +// Determines the platform on which Google Test is compiled. +#ifdef __CYGWIN__ +# define GTEST_OS_CYGWIN 1 +# elif defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__) +# define GTEST_OS_WINDOWS_MINGW 1 +# define GTEST_OS_WINDOWS 1 +#elif defined _WIN32 +# define GTEST_OS_WINDOWS 1 +# ifdef _WIN32_WCE +# define GTEST_OS_WINDOWS_MOBILE 1 +# elif defined(WINAPI_FAMILY) +# include +# if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP) +# define GTEST_OS_WINDOWS_DESKTOP 1 +# elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_PHONE_APP) +# define GTEST_OS_WINDOWS_PHONE 1 +# elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_APP) +# define GTEST_OS_WINDOWS_RT 1 +# elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_TV_TITLE) +# define GTEST_OS_WINDOWS_PHONE 1 +# define GTEST_OS_WINDOWS_TV_TITLE 1 +# else + // WINAPI_FAMILY defined but no known partition matched. + // Default to desktop. +# define GTEST_OS_WINDOWS_DESKTOP 1 +# endif +# else +# define GTEST_OS_WINDOWS_DESKTOP 1 +# endif // _WIN32_WCE +#elif defined __OS2__ +# define GTEST_OS_OS2 1 +#elif defined __APPLE__ +# define GTEST_OS_MAC 1 +# if TARGET_OS_IPHONE +# define GTEST_OS_IOS 1 +# endif +#elif defined __DragonFly__ +# define GTEST_OS_DRAGONFLY 1 +#elif defined __FreeBSD__ +# define GTEST_OS_FREEBSD 1 +#elif defined __Fuchsia__ +# define GTEST_OS_FUCHSIA 1 +#elif defined(__GLIBC__) && defined(__FreeBSD_kernel__) +# define GTEST_OS_GNU_KFREEBSD 1 +#elif defined __linux__ +# define GTEST_OS_LINUX 1 +# if defined __ANDROID__ +# define GTEST_OS_LINUX_ANDROID 1 +# endif +#elif defined __MVS__ +# define GTEST_OS_ZOS 1 +#elif defined(__sun) && defined(__SVR4) +# define GTEST_OS_SOLARIS 1 +#elif defined(_AIX) +# define GTEST_OS_AIX 1 +#elif defined(__hpux) +# define GTEST_OS_HPUX 1 +#elif defined __native_client__ +# define GTEST_OS_NACL 1 +#elif defined __NetBSD__ +# define GTEST_OS_NETBSD 1 +#elif defined __OpenBSD__ +# define GTEST_OS_OPENBSD 1 +#elif defined __QNX__ +# define GTEST_OS_QNX 1 +#elif defined(__HAIKU__) +#define GTEST_OS_HAIKU 1 +#endif // __CYGWIN__ + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-port.h b/deps/googletest/include/gtest/internal/gtest-port.h new file mode 100644 index 000000000..31ec8fe9b --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-port.h @@ -0,0 +1,2320 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Low-level types and utilities for porting Google Test to various +// platforms. All macros ending with _ and symbols defined in an +// internal namespace are subject to change without notice. Code +// outside Google Test MUST NOT USE THEM DIRECTLY. Macros that don't +// end with _ are part of Google Test's public API and can be used by +// code outside Google Test. +// +// This file is fundamental to Google Test. All other Google Test source +// files are expected to #include this. Therefore, it cannot #include +// any other Google Test header. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_ + +// Environment-describing macros +// ----------------------------- +// +// Google Test can be used in many different environments. Macros in +// this section tell Google Test what kind of environment it is being +// used in, such that Google Test can provide environment-specific +// features and implementations. +// +// Google Test tries to automatically detect the properties of its +// environment, so users usually don't need to worry about these +// macros. However, the automatic detection is not perfect. +// Sometimes it's necessary for a user to define some of the following +// macros in the build script to override Google Test's decisions. +// +// If the user doesn't define a macro in the list, Google Test will +// provide a default definition. After this header is #included, all +// macros in this list will be defined to either 1 or 0. +// +// Notes to maintainers: +// - Each macro here is a user-tweakable knob; do not grow the list +// lightly. +// - Use #if to key off these macros. Don't use #ifdef or "#if +// defined(...)", which will not work as these macros are ALWAYS +// defined. +// +// GTEST_HAS_CLONE - Define it to 1/0 to indicate that clone(2) +// is/isn't available. +// GTEST_HAS_EXCEPTIONS - Define it to 1/0 to indicate that exceptions +// are enabled. +// GTEST_HAS_POSIX_RE - Define it to 1/0 to indicate that POSIX regular +// expressions are/aren't available. +// GTEST_HAS_PTHREAD - Define it to 1/0 to indicate that +// is/isn't available. +// GTEST_HAS_RTTI - Define it to 1/0 to indicate that RTTI is/isn't +// enabled. +// GTEST_HAS_STD_WSTRING - Define it to 1/0 to indicate that +// std::wstring does/doesn't work (Google Test can +// be used where std::wstring is unavailable). +// GTEST_HAS_SEH - Define it to 1/0 to indicate whether the +// compiler supports Microsoft's "Structured +// Exception Handling". +// GTEST_HAS_STREAM_REDIRECTION +// - Define it to 1/0 to indicate whether the +// platform supports I/O stream redirection using +// dup() and dup2(). +// GTEST_LINKED_AS_SHARED_LIBRARY +// - Define to 1 when compiling tests that use +// Google Test as a shared library (known as +// DLL on Windows). +// GTEST_CREATE_SHARED_LIBRARY +// - Define to 1 when compiling Google Test itself +// as a shared library. +// GTEST_DEFAULT_DEATH_TEST_STYLE +// - The default value of --gtest_death_test_style. +// The legacy default has been "fast" in the open +// source version since 2008. The recommended value +// is "threadsafe", and can be set in +// custom/gtest-port.h. + +// Platform-indicating macros +// -------------------------- +// +// Macros indicating the platform on which Google Test is being used +// (a macro is defined to 1 if compiled on the given platform; +// otherwise UNDEFINED -- it's never defined to 0.). Google Test +// defines these macros automatically. Code outside Google Test MUST +// NOT define them. +// +// GTEST_OS_AIX - IBM AIX +// GTEST_OS_CYGWIN - Cygwin +// GTEST_OS_DRAGONFLY - DragonFlyBSD +// GTEST_OS_FREEBSD - FreeBSD +// GTEST_OS_FUCHSIA - Fuchsia +// GTEST_OS_GNU_KFREEBSD - GNU/kFreeBSD +// GTEST_OS_HAIKU - Haiku +// GTEST_OS_HPUX - HP-UX +// GTEST_OS_LINUX - Linux +// GTEST_OS_LINUX_ANDROID - Google Android +// GTEST_OS_MAC - Mac OS X +// GTEST_OS_IOS - iOS +// GTEST_OS_NACL - Google Native Client (NaCl) +// GTEST_OS_NETBSD - NetBSD +// GTEST_OS_OPENBSD - OpenBSD +// GTEST_OS_OS2 - OS/2 +// GTEST_OS_QNX - QNX +// GTEST_OS_SOLARIS - Sun Solaris +// GTEST_OS_WINDOWS - Windows (Desktop, MinGW, or Mobile) +// GTEST_OS_WINDOWS_DESKTOP - Windows Desktop +// GTEST_OS_WINDOWS_MINGW - MinGW +// GTEST_OS_WINDOWS_MOBILE - Windows Mobile +// GTEST_OS_WINDOWS_PHONE - Windows Phone +// GTEST_OS_WINDOWS_RT - Windows Store App/WinRT +// GTEST_OS_ZOS - z/OS +// +// Among the platforms, Cygwin, Linux, Mac OS X, and Windows have the +// most stable support. Since core members of the Google Test project +// don't have access to other platforms, support for them may be less +// stable. If you notice any problems on your platform, please notify +// googletestframework@googlegroups.com (patches for fixing them are +// even more welcome!). +// +// It is possible that none of the GTEST_OS_* macros are defined. + +// Feature-indicating macros +// ------------------------- +// +// Macros indicating which Google Test features are available (a macro +// is defined to 1 if the corresponding feature is supported; +// otherwise UNDEFINED -- it's never defined to 0.). Google Test +// defines these macros automatically. Code outside Google Test MUST +// NOT define them. +// +// These macros are public so that portable tests can be written. +// Such tests typically surround code using a feature with an #if +// which controls that code. For example: +// +// #if GTEST_HAS_DEATH_TEST +// EXPECT_DEATH(DoSomethingDeadly()); +// #endif +// +// GTEST_HAS_DEATH_TEST - death tests +// GTEST_HAS_TYPED_TEST - typed tests +// GTEST_HAS_TYPED_TEST_P - type-parameterized tests +// GTEST_IS_THREADSAFE - Google Test is thread-safe. +// GOOGLETEST_CM0007 DO NOT DELETE +// GTEST_USES_POSIX_RE - enhanced POSIX regex is used. Do not confuse with +// GTEST_HAS_POSIX_RE (see above) which users can +// define themselves. +// GTEST_USES_SIMPLE_RE - our own simple regex is used; +// the above RE\b(s) are mutually exclusive. + +// Misc public macros +// ------------------ +// +// GTEST_FLAG(flag_name) - references the variable corresponding to +// the given Google Test flag. + +// Internal utilities +// ------------------ +// +// The following macros and utilities are for Google Test's INTERNAL +// use only. Code outside Google Test MUST NOT USE THEM DIRECTLY. +// +// Macros for basic C++ coding: +// GTEST_AMBIGUOUS_ELSE_BLOCKER_ - for disabling a gcc warning. +// GTEST_ATTRIBUTE_UNUSED_ - declares that a class' instances or a +// variable don't have to be used. +// GTEST_DISALLOW_ASSIGN_ - disables operator=. +// GTEST_DISALLOW_COPY_AND_ASSIGN_ - disables copy ctor and operator=. +// GTEST_MUST_USE_RESULT_ - declares that a function's result must be used. +// GTEST_INTENTIONAL_CONST_COND_PUSH_ - start code section where MSVC C4127 is +// suppressed (constant conditional). +// GTEST_INTENTIONAL_CONST_COND_POP_ - finish code section where MSVC C4127 +// is suppressed. +// +// Synchronization: +// Mutex, MutexLock, ThreadLocal, GetThreadCount() +// - synchronization primitives. +// +// Template meta programming: +// IteratorTraits - partial implementation of std::iterator_traits, which +// is not available in libCstd when compiled with Sun C++. +// +// +// Regular expressions: +// RE - a simple regular expression class using the POSIX +// Extended Regular Expression syntax on UNIX-like platforms +// GOOGLETEST_CM0008 DO NOT DELETE +// or a reduced regular exception syntax on other +// platforms, including Windows. +// Logging: +// GTEST_LOG_() - logs messages at the specified severity level. +// LogToStderr() - directs all log messages to stderr. +// FlushInfoLog() - flushes informational log messages. +// +// Stdout and stderr capturing: +// CaptureStdout() - starts capturing stdout. +// GetCapturedStdout() - stops capturing stdout and returns the captured +// string. +// CaptureStderr() - starts capturing stderr. +// GetCapturedStderr() - stops capturing stderr and returns the captured +// string. +// +// Integer types: +// TypeWithSize - maps an integer to a int type. +// Int32, UInt32, Int64, UInt64, TimeInMillis +// - integers of known sizes. +// BiggestInt - the biggest signed integer type. +// +// Command-line utilities: +// GTEST_DECLARE_*() - declares a flag. +// GTEST_DEFINE_*() - defines a flag. +// GetInjectableArgvs() - returns the command line as a vector of strings. +// +// Environment variable utilities: +// GetEnv() - gets the value of an environment variable. +// BoolFromGTestEnv() - parses a bool environment variable. +// Int32FromGTestEnv() - parses an Int32 environment variable. +// StringFromGTestEnv() - parses a string environment variable. +// +// Deprecation warnings: +// GTEST_INTERNAL_DEPRECATED(message) - attribute marking a function as +// deprecated; calling a marked function +// should generate a compiler warning + +#include // for isspace, etc +#include // for ptrdiff_t +#include +#include +#include +#include +#include + +#ifndef _WIN32_WCE +# include +# include +#endif // !_WIN32_WCE + +#if defined __APPLE__ +# include +# include +#endif + +#include // NOLINT +#include // NOLINT +#include // NOLINT +#include // NOLINT +#include +#include +#include // NOLINT + +#include "gtest/internal/gtest-port-arch.h" +#include "gtest/internal/custom/gtest-port.h" + +#if !defined(GTEST_DEV_EMAIL_) +# define GTEST_DEV_EMAIL_ "googletestframework@@googlegroups.com" +# define GTEST_FLAG_PREFIX_ "gtest_" +# define GTEST_FLAG_PREFIX_DASH_ "gtest-" +# define GTEST_FLAG_PREFIX_UPPER_ "GTEST_" +# define GTEST_NAME_ "Google Test" +# define GTEST_PROJECT_URL_ "https://github.com/google/googletest/" +#endif // !defined(GTEST_DEV_EMAIL_) + +#if !defined(GTEST_INIT_GOOGLE_TEST_NAME_) +# define GTEST_INIT_GOOGLE_TEST_NAME_ "testing::InitGoogleTest" +#endif // !defined(GTEST_INIT_GOOGLE_TEST_NAME_) + +// Determines the version of gcc that is used to compile this. +#ifdef __GNUC__ +// 40302 means version 4.3.2. +# define GTEST_GCC_VER_ \ + (__GNUC__*10000 + __GNUC_MINOR__*100 + __GNUC_PATCHLEVEL__) +#endif // __GNUC__ + +// Macros for disabling Microsoft Visual C++ warnings. +// +// GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 4385) +// /* code that triggers warnings C4800 and C4385 */ +// GTEST_DISABLE_MSC_WARNINGS_POP_() +#if defined(_MSC_VER) +# define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings) \ + __pragma(warning(push)) \ + __pragma(warning(disable: warnings)) +# define GTEST_DISABLE_MSC_WARNINGS_POP_() \ + __pragma(warning(pop)) +#else +// Not all compilers are MSVC +# define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings) +# define GTEST_DISABLE_MSC_WARNINGS_POP_() +#endif + +// Clang on Windows does not understand MSVC's pragma warning. +// We need clang-specific way to disable function deprecation warning. +#ifdef __clang__ +# define GTEST_DISABLE_MSC_DEPRECATED_PUSH_() \ + _Pragma("clang diagnostic push") \ + _Pragma("clang diagnostic ignored \"-Wdeprecated-declarations\"") \ + _Pragma("clang diagnostic ignored \"-Wdeprecated-implementations\"") +#define GTEST_DISABLE_MSC_DEPRECATED_POP_() \ + _Pragma("clang diagnostic pop") +#else +# define GTEST_DISABLE_MSC_DEPRECATED_PUSH_() \ + GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996) +# define GTEST_DISABLE_MSC_DEPRECATED_POP_() \ + GTEST_DISABLE_MSC_WARNINGS_POP_() +#endif + +// Brings in definitions for functions used in the testing::internal::posix +// namespace (read, write, close, chdir, isatty, stat). We do not currently +// use them on Windows Mobile. +#if GTEST_OS_WINDOWS +# if !GTEST_OS_WINDOWS_MOBILE +# include +# include +# endif +// In order to avoid having to include , use forward declaration +#if GTEST_OS_WINDOWS_MINGW && !defined(__MINGW64_VERSION_MAJOR) +// MinGW defined _CRITICAL_SECTION and _RTL_CRITICAL_SECTION as two +// separate (equivalent) structs, instead of using typedef +typedef struct _CRITICAL_SECTION GTEST_CRITICAL_SECTION; +#else +// Assume CRITICAL_SECTION is a typedef of _RTL_CRITICAL_SECTION. +// This assumption is verified by +// WindowsTypesTest.CRITICAL_SECTIONIs_RTL_CRITICAL_SECTION. +typedef struct _RTL_CRITICAL_SECTION GTEST_CRITICAL_SECTION; +#endif +#else +// This assumes that non-Windows OSes provide unistd.h. For OSes where this +// is not the case, we need to include headers that provide the functions +// mentioned above. +# include +# include +#endif // GTEST_OS_WINDOWS + +#if GTEST_OS_LINUX_ANDROID +// Used to define __ANDROID_API__ matching the target NDK API level. +# include // NOLINT +#endif + +// Defines this to true iff Google Test can use POSIX regular expressions. +#ifndef GTEST_HAS_POSIX_RE +# if GTEST_OS_LINUX_ANDROID +// On Android, is only available starting with Gingerbread. +# define GTEST_HAS_POSIX_RE (__ANDROID_API__ >= 9) +# else +# define GTEST_HAS_POSIX_RE (!GTEST_OS_WINDOWS) +# endif +#endif + +#if GTEST_USES_PCRE +// The appropriate headers have already been included. + +#elif GTEST_HAS_POSIX_RE + +// On some platforms, needs someone to define size_t, and +// won't compile otherwise. We can #include it here as we already +// included , which is guaranteed to define size_t through +// . +# include // NOLINT + +# define GTEST_USES_POSIX_RE 1 + +#elif GTEST_OS_WINDOWS + +// is not available on Windows. Use our own simple regex +// implementation instead. +# define GTEST_USES_SIMPLE_RE 1 + +#else + +// may not be available on this platform. Use our own +// simple regex implementation instead. +# define GTEST_USES_SIMPLE_RE 1 + +#endif // GTEST_USES_PCRE + +#ifndef GTEST_HAS_EXCEPTIONS +// The user didn't tell us whether exceptions are enabled, so we need +// to figure it out. +# if defined(_MSC_VER) && defined(_CPPUNWIND) +// MSVC defines _CPPUNWIND to 1 iff exceptions are enabled. +# define GTEST_HAS_EXCEPTIONS 1 +# elif defined(__BORLANDC__) +// C++Builder's implementation of the STL uses the _HAS_EXCEPTIONS +// macro to enable exceptions, so we'll do the same. +// Assumes that exceptions are enabled by default. +# ifndef _HAS_EXCEPTIONS +# define _HAS_EXCEPTIONS 1 +# endif // _HAS_EXCEPTIONS +# define GTEST_HAS_EXCEPTIONS _HAS_EXCEPTIONS +# elif defined(__clang__) +// clang defines __EXCEPTIONS iff exceptions are enabled before clang 220714, +// but iff cleanups are enabled after that. In Obj-C++ files, there can be +// cleanups for ObjC exceptions which also need cleanups, even if C++ exceptions +// are disabled. clang has __has_feature(cxx_exceptions) which checks for C++ +// exceptions starting at clang r206352, but which checked for cleanups prior to +// that. To reliably check for C++ exception availability with clang, check for +// __EXCEPTIONS && __has_feature(cxx_exceptions). +# define GTEST_HAS_EXCEPTIONS (__EXCEPTIONS && __has_feature(cxx_exceptions)) +# elif defined(__GNUC__) && __EXCEPTIONS +// gcc defines __EXCEPTIONS to 1 iff exceptions are enabled. +# define GTEST_HAS_EXCEPTIONS 1 +# elif defined(__SUNPRO_CC) +// Sun Pro CC supports exceptions. However, there is no compile-time way of +// detecting whether they are enabled or not. Therefore, we assume that +// they are enabled unless the user tells us otherwise. +# define GTEST_HAS_EXCEPTIONS 1 +# elif defined(__IBMCPP__) && __EXCEPTIONS +// xlC defines __EXCEPTIONS to 1 iff exceptions are enabled. +# define GTEST_HAS_EXCEPTIONS 1 +# elif defined(__HP_aCC) +// Exception handling is in effect by default in HP aCC compiler. It has to +// be turned of by +noeh compiler option if desired. +# define GTEST_HAS_EXCEPTIONS 1 +# else +// For other compilers, we assume exceptions are disabled to be +// conservative. +# define GTEST_HAS_EXCEPTIONS 0 +# endif // defined(_MSC_VER) || defined(__BORLANDC__) +#endif // GTEST_HAS_EXCEPTIONS + +#if !defined(GTEST_HAS_STD_STRING) +// Even though we don't use this macro any longer, we keep it in case +// some clients still depend on it. +# define GTEST_HAS_STD_STRING 1 +#elif !GTEST_HAS_STD_STRING +// The user told us that ::std::string isn't available. +# error "::std::string isn't available." +#endif // !defined(GTEST_HAS_STD_STRING) + +#ifndef GTEST_HAS_STD_WSTRING +// The user didn't tell us whether ::std::wstring is available, so we need +// to figure it out. +// Cygwin 1.7 and below doesn't support ::std::wstring. +// Solaris' libc++ doesn't support it either. Android has +// no support for it at least as recent as Froyo (2.2). +#define GTEST_HAS_STD_WSTRING \ + (!(GTEST_OS_LINUX_ANDROID || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS || \ + GTEST_OS_HAIKU)) + +#endif // GTEST_HAS_STD_WSTRING + +// Determines whether RTTI is available. +#ifndef GTEST_HAS_RTTI +// The user didn't tell us whether RTTI is enabled, so we need to +// figure it out. + +# ifdef _MSC_VER + +# ifdef _CPPRTTI // MSVC defines this macro iff RTTI is enabled. +# define GTEST_HAS_RTTI 1 +# else +# define GTEST_HAS_RTTI 0 +# endif + +// Starting with version 4.3.2, gcc defines __GXX_RTTI iff RTTI is enabled. +# elif defined(__GNUC__) + +# ifdef __GXX_RTTI +// When building against STLport with the Android NDK and with +// -frtti -fno-exceptions, the build fails at link time with undefined +// references to __cxa_bad_typeid. Note sure if STL or toolchain bug, +// so disable RTTI when detected. +# if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR) && \ + !defined(__EXCEPTIONS) +# define GTEST_HAS_RTTI 0 +# else +# define GTEST_HAS_RTTI 1 +# endif // GTEST_OS_LINUX_ANDROID && __STLPORT_MAJOR && !__EXCEPTIONS +# else +# define GTEST_HAS_RTTI 0 +# endif // __GXX_RTTI + +// Clang defines __GXX_RTTI starting with version 3.0, but its manual recommends +// using has_feature instead. has_feature(cxx_rtti) is supported since 2.7, the +// first version with C++ support. +# elif defined(__clang__) + +# define GTEST_HAS_RTTI __has_feature(cxx_rtti) + +// Starting with version 9.0 IBM Visual Age defines __RTTI_ALL__ to 1 if +// both the typeid and dynamic_cast features are present. +# elif defined(__IBMCPP__) && (__IBMCPP__ >= 900) + +# ifdef __RTTI_ALL__ +# define GTEST_HAS_RTTI 1 +# else +# define GTEST_HAS_RTTI 0 +# endif + +# else + +// For all other compilers, we assume RTTI is enabled. +# define GTEST_HAS_RTTI 1 + +# endif // _MSC_VER + +#endif // GTEST_HAS_RTTI + +// It's this header's responsibility to #include when RTTI +// is enabled. +#if GTEST_HAS_RTTI +# include +#endif + +// Determines whether Google Test can use the pthreads library. +#ifndef GTEST_HAS_PTHREAD +// The user didn't tell us explicitly, so we make reasonable assumptions about +// which platforms have pthreads support. +// +// To disable threading support in Google Test, add -DGTEST_HAS_PTHREAD=0 +// to your compiler flags. +#define GTEST_HAS_PTHREAD \ + (GTEST_OS_LINUX || GTEST_OS_MAC || GTEST_OS_HPUX || GTEST_OS_QNX || \ + GTEST_OS_FREEBSD || GTEST_OS_NACL || GTEST_OS_NETBSD || GTEST_OS_FUCHSIA || \ + GTEST_OS_DRAGONFLY || GTEST_OS_GNU_KFREEBSD || GTEST_OS_OPENBSD || \ + GTEST_OS_HAIKU) +#endif // GTEST_HAS_PTHREAD + +#if GTEST_HAS_PTHREAD +// gtest-port.h guarantees to #include when GTEST_HAS_PTHREAD is +// true. +# include // NOLINT + +// For timespec and nanosleep, used below. +# include // NOLINT +#endif + +// Determines whether clone(2) is supported. +// Usually it will only be available on Linux, excluding +// Linux on the Itanium architecture. +// Also see http://linux.die.net/man/2/clone. +#ifndef GTEST_HAS_CLONE +// The user didn't tell us, so we need to figure it out. + +# if GTEST_OS_LINUX && !defined(__ia64__) +# if GTEST_OS_LINUX_ANDROID +// On Android, clone() became available at different API levels for each 32-bit +// architecture. +# if defined(__LP64__) || \ + (defined(__arm__) && __ANDROID_API__ >= 9) || \ + (defined(__mips__) && __ANDROID_API__ >= 12) || \ + (defined(__i386__) && __ANDROID_API__ >= 17) +# define GTEST_HAS_CLONE 1 +# else +# define GTEST_HAS_CLONE 0 +# endif +# else +# define GTEST_HAS_CLONE 1 +# endif +# else +# define GTEST_HAS_CLONE 0 +# endif // GTEST_OS_LINUX && !defined(__ia64__) + +#endif // GTEST_HAS_CLONE + +// Determines whether to support stream redirection. This is used to test +// output correctness and to implement death tests. +#ifndef GTEST_HAS_STREAM_REDIRECTION +// By default, we assume that stream redirection is supported on all +// platforms except known mobile ones. +# if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT +# define GTEST_HAS_STREAM_REDIRECTION 0 +# else +# define GTEST_HAS_STREAM_REDIRECTION 1 +# endif // !GTEST_OS_WINDOWS_MOBILE +#endif // GTEST_HAS_STREAM_REDIRECTION + +// Determines whether to support death tests. +// pops up a dialog window that cannot be suppressed programmatically. +#if (GTEST_OS_LINUX || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS || \ + (GTEST_OS_MAC && !GTEST_OS_IOS) || \ + (GTEST_OS_WINDOWS_DESKTOP && _MSC_VER) || GTEST_OS_WINDOWS_MINGW || \ + GTEST_OS_AIX || GTEST_OS_HPUX || GTEST_OS_OPENBSD || GTEST_OS_QNX || \ + GTEST_OS_FREEBSD || GTEST_OS_NETBSD || GTEST_OS_FUCHSIA || \ + GTEST_OS_DRAGONFLY || GTEST_OS_GNU_KFREEBSD || GTEST_OS_HAIKU) +# define GTEST_HAS_DEATH_TEST 1 +#endif + +// Determines whether to support type-driven tests. + +// Typed tests need and variadic macros, which GCC, VC++ 8.0, +// Sun Pro CC, IBM Visual Age, and HP aCC support. +#if defined(__GNUC__) || defined(_MSC_VER) || defined(__SUNPRO_CC) || \ + defined(__IBMCPP__) || defined(__HP_aCC) +# define GTEST_HAS_TYPED_TEST 1 +# define GTEST_HAS_TYPED_TEST_P 1 +#endif + +// Determines whether the system compiler uses UTF-16 for encoding wide strings. +#define GTEST_WIDE_STRING_USES_UTF16_ \ + (GTEST_OS_WINDOWS || GTEST_OS_CYGWIN || GTEST_OS_AIX || GTEST_OS_OS2) + +// Determines whether test results can be streamed to a socket. +#if GTEST_OS_LINUX || GTEST_OS_GNU_KFREEBSD || GTEST_OS_DRAGONFLY || \ + GTEST_OS_FREEBSD || GTEST_OS_NETBSD || GTEST_OS_OPENBSD +# define GTEST_CAN_STREAM_RESULTS_ 1 +#endif + +// Defines some utility macros. + +// The GNU compiler emits a warning if nested "if" statements are followed by +// an "else" statement and braces are not used to explicitly disambiguate the +// "else" binding. This leads to problems with code like: +// +// if (gate) +// ASSERT_*(condition) << "Some message"; +// +// The "switch (0) case 0:" idiom is used to suppress this. +#ifdef __INTEL_COMPILER +# define GTEST_AMBIGUOUS_ELSE_BLOCKER_ +#else +# define GTEST_AMBIGUOUS_ELSE_BLOCKER_ switch (0) case 0: default: // NOLINT +#endif + +// Use this annotation at the end of a struct/class definition to +// prevent the compiler from optimizing away instances that are never +// used. This is useful when all interesting logic happens inside the +// c'tor and / or d'tor. Example: +// +// struct Foo { +// Foo() { ... } +// } GTEST_ATTRIBUTE_UNUSED_; +// +// Also use it after a variable or parameter declaration to tell the +// compiler the variable/parameter does not have to be used. +#if defined(__GNUC__) && !defined(COMPILER_ICC) +# define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused)) +#elif defined(__clang__) +# if __has_attribute(unused) +# define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused)) +# endif +#endif +#ifndef GTEST_ATTRIBUTE_UNUSED_ +# define GTEST_ATTRIBUTE_UNUSED_ +#endif + +// Use this annotation before a function that takes a printf format string. +#if (defined(__GNUC__) || defined(__clang__)) && !defined(COMPILER_ICC) +# if defined(__MINGW_PRINTF_FORMAT) +// MinGW has two different printf implementations. Ensure the format macro +// matches the selected implementation. See +// https://sourceforge.net/p/mingw-w64/wiki2/gnu%20printf/. +# define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) \ + __attribute__((__format__(__MINGW_PRINTF_FORMAT, string_index, \ + first_to_check))) +# else +# define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) \ + __attribute__((__format__(__printf__, string_index, first_to_check))) +# endif +#else +# define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) +#endif + + +// A macro to disallow operator= +// This should be used in the private: declarations for a class. +#define GTEST_DISALLOW_ASSIGN_(type) \ + void operator=(type const &) = delete + +// A macro to disallow copy constructor and operator= +// This should be used in the private: declarations for a class. +#define GTEST_DISALLOW_COPY_AND_ASSIGN_(type) \ + type(type const &) = delete; \ + GTEST_DISALLOW_ASSIGN_(type) + +// Tell the compiler to warn about unused return values for functions declared +// with this macro. The macro should be used on function declarations +// following the argument list: +// +// Sprocket* AllocateSprocket() GTEST_MUST_USE_RESULT_; +#if defined(__GNUC__) && !defined(COMPILER_ICC) +# define GTEST_MUST_USE_RESULT_ __attribute__ ((warn_unused_result)) +#else +# define GTEST_MUST_USE_RESULT_ +#endif // __GNUC__ && !COMPILER_ICC + +// MS C++ compiler emits warning when a conditional expression is compile time +// constant. In some contexts this warning is false positive and needs to be +// suppressed. Use the following two macros in such cases: +// +// GTEST_INTENTIONAL_CONST_COND_PUSH_() +// while (true) { +// GTEST_INTENTIONAL_CONST_COND_POP_() +// } +# define GTEST_INTENTIONAL_CONST_COND_PUSH_() \ + GTEST_DISABLE_MSC_WARNINGS_PUSH_(4127) +# define GTEST_INTENTIONAL_CONST_COND_POP_() \ + GTEST_DISABLE_MSC_WARNINGS_POP_() + +// Determine whether the compiler supports Microsoft's Structured Exception +// Handling. This is supported by several Windows compilers but generally +// does not exist on any other system. +#ifndef GTEST_HAS_SEH +// The user didn't tell us, so we need to figure it out. + +# if defined(_MSC_VER) || defined(__BORLANDC__) +// These two compilers are known to support SEH. +# define GTEST_HAS_SEH 1 +# else +// Assume no SEH. +# define GTEST_HAS_SEH 0 +# endif + +#endif // GTEST_HAS_SEH + +#ifndef GTEST_IS_THREADSAFE + +#define GTEST_IS_THREADSAFE \ + (GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ || \ + (GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT) || \ + GTEST_HAS_PTHREAD) + +#endif // GTEST_IS_THREADSAFE + +// GTEST_API_ qualifies all symbols that must be exported. The definitions below +// are guarded by #ifndef to give embedders a chance to define GTEST_API_ in +// gtest/internal/custom/gtest-port.h +#ifndef GTEST_API_ + +#ifdef _MSC_VER +# if GTEST_LINKED_AS_SHARED_LIBRARY +# define GTEST_API_ __declspec(dllimport) +# elif GTEST_CREATE_SHARED_LIBRARY +# define GTEST_API_ __declspec(dllexport) +# endif +#elif __GNUC__ >= 4 || defined(__clang__) +# define GTEST_API_ __attribute__((visibility ("default"))) +#endif // _MSC_VER + +#endif // GTEST_API_ + +#ifndef GTEST_API_ +# define GTEST_API_ +#endif // GTEST_API_ + +#ifndef GTEST_DEFAULT_DEATH_TEST_STYLE +# define GTEST_DEFAULT_DEATH_TEST_STYLE "fast" +#endif // GTEST_DEFAULT_DEATH_TEST_STYLE + +#ifdef __GNUC__ +// Ask the compiler to never inline a given function. +# define GTEST_NO_INLINE_ __attribute__((noinline)) +#else +# define GTEST_NO_INLINE_ +#endif + +// _LIBCPP_VERSION is defined by the libc++ library from the LLVM project. +#if !defined(GTEST_HAS_CXXABI_H_) +# if defined(__GLIBCXX__) || (defined(_LIBCPP_VERSION) && !defined(_MSC_VER)) +# define GTEST_HAS_CXXABI_H_ 1 +# else +# define GTEST_HAS_CXXABI_H_ 0 +# endif +#endif + +// A function level attribute to disable checking for use of uninitialized +// memory when built with MemorySanitizer. +#if defined(__clang__) +# if __has_feature(memory_sanitizer) +# define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ \ + __attribute__((no_sanitize_memory)) +# else +# define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ +# endif // __has_feature(memory_sanitizer) +#else +# define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ +#endif // __clang__ + +// A function level attribute to disable AddressSanitizer instrumentation. +#if defined(__clang__) +# if __has_feature(address_sanitizer) +# define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ \ + __attribute__((no_sanitize_address)) +# else +# define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ +# endif // __has_feature(address_sanitizer) +#else +# define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ +#endif // __clang__ + +// A function level attribute to disable HWAddressSanitizer instrumentation. +#if defined(__clang__) +# if __has_feature(hwaddress_sanitizer) +# define GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ \ + __attribute__((no_sanitize("hwaddress"))) +# else +# define GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +# endif // __has_feature(hwaddress_sanitizer) +#else +# define GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +#endif // __clang__ + +// A function level attribute to disable ThreadSanitizer instrumentation. +#if defined(__clang__) +# if __has_feature(thread_sanitizer) +# define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ \ + __attribute__((no_sanitize_thread)) +# else +# define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ +# endif // __has_feature(thread_sanitizer) +#else +# define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ +#endif // __clang__ + +namespace testing { + +class Message; + +// Legacy imports for backwards compatibility. +// New code should use std:: names directly. +using std::get; +using std::make_tuple; +using std::tuple; +using std::tuple_element; +using std::tuple_size; + +namespace internal { + +// A secret type that Google Test users don't know about. It has no +// definition on purpose. Therefore it's impossible to create a +// Secret object, which is what we want. +class Secret; + +// The GTEST_COMPILE_ASSERT_ is a legacy macro used to verify that a compile +// time expression is true (in new code, use static_assert instead). For +// example, you could use it to verify the size of a static array: +// +// GTEST_COMPILE_ASSERT_(GTEST_ARRAY_SIZE_(names) == NUM_NAMES, +// names_incorrect_size); +// +// The second argument to the macro must be a valid C++ identifier. If the +// expression is false, compiler will issue an error containing this identifier. +#define GTEST_COMPILE_ASSERT_(expr, msg) static_assert(expr, #msg) + +// StaticAssertTypeEqHelper is used by StaticAssertTypeEq defined in gtest.h. +// +// This template is declared, but intentionally undefined. +template +struct StaticAssertTypeEqHelper; + +template +struct StaticAssertTypeEqHelper { + enum { value = true }; +}; + +// Same as std::is_same<>. +template +struct IsSame { + enum { value = false }; +}; +template +struct IsSame { + enum { value = true }; +}; + +// Evaluates to the number of elements in 'array'. +#define GTEST_ARRAY_SIZE_(array) (sizeof(array) / sizeof(array[0])) + +// A helper for suppressing warnings on constant condition. It just +// returns 'condition'. +GTEST_API_ bool IsTrue(bool condition); + +// Defines RE. + +#if GTEST_USES_PCRE +// if used, PCRE is injected by custom/gtest-port.h +#elif GTEST_USES_POSIX_RE || GTEST_USES_SIMPLE_RE + +// A simple C++ wrapper for . It uses the POSIX Extended +// Regular Expression syntax. +class GTEST_API_ RE { + public: + // A copy constructor is required by the Standard to initialize object + // references from r-values. + RE(const RE& other) { Init(other.pattern()); } + + // Constructs an RE from a string. + RE(const ::std::string& regex) { Init(regex.c_str()); } // NOLINT + + RE(const char* regex) { Init(regex); } // NOLINT + ~RE(); + + // Returns the string representation of the regex. + const char* pattern() const { return pattern_; } + + // FullMatch(str, re) returns true iff regular expression re matches + // the entire str. + // PartialMatch(str, re) returns true iff regular expression re + // matches a substring of str (including str itself). + static bool FullMatch(const ::std::string& str, const RE& re) { + return FullMatch(str.c_str(), re); + } + static bool PartialMatch(const ::std::string& str, const RE& re) { + return PartialMatch(str.c_str(), re); + } + + static bool FullMatch(const char* str, const RE& re); + static bool PartialMatch(const char* str, const RE& re); + + private: + void Init(const char* regex); + const char* pattern_; + bool is_valid_; + +# if GTEST_USES_POSIX_RE + + regex_t full_regex_; // For FullMatch(). + regex_t partial_regex_; // For PartialMatch(). + +# else // GTEST_USES_SIMPLE_RE + + const char* full_pattern_; // For FullMatch(); + +# endif + + GTEST_DISALLOW_ASSIGN_(RE); +}; + +#endif // GTEST_USES_PCRE + +// Formats a source file path and a line number as they would appear +// in an error message from the compiler used to compile this code. +GTEST_API_ ::std::string FormatFileLocation(const char* file, int line); + +// Formats a file location for compiler-independent XML output. +// Although this function is not platform dependent, we put it next to +// FormatFileLocation in order to contrast the two functions. +GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(const char* file, + int line); + +// Defines logging utilities: +// GTEST_LOG_(severity) - logs messages at the specified severity level. The +// message itself is streamed into the macro. +// LogToStderr() - directs all log messages to stderr. +// FlushInfoLog() - flushes informational log messages. + +enum GTestLogSeverity { + GTEST_INFO, + GTEST_WARNING, + GTEST_ERROR, + GTEST_FATAL +}; + +// Formats log entry severity, provides a stream object for streaming the +// log message, and terminates the message with a newline when going out of +// scope. +class GTEST_API_ GTestLog { + public: + GTestLog(GTestLogSeverity severity, const char* file, int line); + + // Flushes the buffers and, if severity is GTEST_FATAL, aborts the program. + ~GTestLog(); + + ::std::ostream& GetStream() { return ::std::cerr; } + + private: + const GTestLogSeverity severity_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestLog); +}; + +#if !defined(GTEST_LOG_) + +# define GTEST_LOG_(severity) \ + ::testing::internal::GTestLog(::testing::internal::GTEST_##severity, \ + __FILE__, __LINE__).GetStream() + +inline void LogToStderr() {} +inline void FlushInfoLog() { fflush(nullptr); } + +#endif // !defined(GTEST_LOG_) + +#if !defined(GTEST_CHECK_) +// INTERNAL IMPLEMENTATION - DO NOT USE. +// +// GTEST_CHECK_ is an all-mode assert. It aborts the program if the condition +// is not satisfied. +// Synopsys: +// GTEST_CHECK_(boolean_condition); +// or +// GTEST_CHECK_(boolean_condition) << "Additional message"; +// +// This checks the condition and if the condition is not satisfied +// it prints message about the condition violation, including the +// condition itself, plus additional message streamed into it, if any, +// and then it aborts the program. It aborts the program irrespective of +// whether it is built in the debug mode or not. +# define GTEST_CHECK_(condition) \ + GTEST_AMBIGUOUS_ELSE_BLOCKER_ \ + if (::testing::internal::IsTrue(condition)) \ + ; \ + else \ + GTEST_LOG_(FATAL) << "Condition " #condition " failed. " +#endif // !defined(GTEST_CHECK_) + +// An all-mode assert to verify that the given POSIX-style function +// call returns 0 (indicating success). Known limitation: this +// doesn't expand to a balanced 'if' statement, so enclose the macro +// in {} if you need to use it as the only statement in an 'if' +// branch. +#define GTEST_CHECK_POSIX_SUCCESS_(posix_call) \ + if (const int gtest_error = (posix_call)) \ + GTEST_LOG_(FATAL) << #posix_call << "failed with error " \ + << gtest_error + +// Adds reference to a type if it is not a reference type, +// otherwise leaves it unchanged. This is the same as +// tr1::add_reference, which is not widely available yet. +template +struct AddReference { typedef T& type; }; // NOLINT +template +struct AddReference { typedef T& type; }; // NOLINT + +// A handy wrapper around AddReference that works when the argument T +// depends on template parameters. +#define GTEST_ADD_REFERENCE_(T) \ + typename ::testing::internal::AddReference::type + +// Transforms "T" into "const T&" according to standard reference collapsing +// rules (this is only needed as a backport for C++98 compilers that do not +// support reference collapsing). Specifically, it transforms: +// +// char ==> const char& +// const char ==> const char& +// char& ==> char& +// const char& ==> const char& +// +// Note that the non-const reference will not have "const" added. This is +// standard, and necessary so that "T" can always bind to "const T&". +template +struct ConstRef { typedef const T& type; }; +template +struct ConstRef { typedef T& type; }; + +// The argument T must depend on some template parameters. +#define GTEST_REFERENCE_TO_CONST_(T) \ + typename ::testing::internal::ConstRef::type + +// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE. +// +// Use ImplicitCast_ as a safe version of static_cast for upcasting in +// the type hierarchy (e.g. casting a Foo* to a SuperclassOfFoo* or a +// const Foo*). When you use ImplicitCast_, the compiler checks that +// the cast is safe. Such explicit ImplicitCast_s are necessary in +// surprisingly many situations where C++ demands an exact type match +// instead of an argument type convertable to a target type. +// +// The syntax for using ImplicitCast_ is the same as for static_cast: +// +// ImplicitCast_(expr) +// +// ImplicitCast_ would have been part of the C++ standard library, +// but the proposal was submitted too late. It will probably make +// its way into the language in the future. +// +// This relatively ugly name is intentional. It prevents clashes with +// similar functions users may have (e.g., implicit_cast). The internal +// namespace alone is not enough because the function can be found by ADL. +template +inline To ImplicitCast_(To x) { return x; } + +// When you upcast (that is, cast a pointer from type Foo to type +// SuperclassOfFoo), it's fine to use ImplicitCast_<>, since upcasts +// always succeed. When you downcast (that is, cast a pointer from +// type Foo to type SubclassOfFoo), static_cast<> isn't safe, because +// how do you know the pointer is really of type SubclassOfFoo? It +// could be a bare Foo, or of type DifferentSubclassOfFoo. Thus, +// when you downcast, you should use this macro. In debug mode, we +// use dynamic_cast<> to double-check the downcast is legal (we die +// if it's not). In normal mode, we do the efficient static_cast<> +// instead. Thus, it's important to test in debug mode to make sure +// the cast is legal! +// This is the only place in the code we should use dynamic_cast<>. +// In particular, you SHOULDN'T be using dynamic_cast<> in order to +// do RTTI (eg code like this: +// if (dynamic_cast(foo)) HandleASubclass1Object(foo); +// if (dynamic_cast(foo)) HandleASubclass2Object(foo); +// You should design the code some other way not to need this. +// +// This relatively ugly name is intentional. It prevents clashes with +// similar functions users may have (e.g., down_cast). The internal +// namespace alone is not enough because the function can be found by ADL. +template // use like this: DownCast_(foo); +inline To DownCast_(From* f) { // so we only accept pointers + // Ensures that To is a sub-type of From *. This test is here only + // for compile-time type checking, and has no overhead in an + // optimized build at run-time, as it will be optimized away + // completely. + GTEST_INTENTIONAL_CONST_COND_PUSH_() + if (false) { + GTEST_INTENTIONAL_CONST_COND_POP_() + const To to = nullptr; + ::testing::internal::ImplicitCast_(to); + } + +#if GTEST_HAS_RTTI + // RTTI: debug mode only! + GTEST_CHECK_(f == nullptr || dynamic_cast(f) != nullptr); +#endif + return static_cast(f); +} + +// Downcasts the pointer of type Base to Derived. +// Derived must be a subclass of Base. The parameter MUST +// point to a class of type Derived, not any subclass of it. +// When RTTI is available, the function performs a runtime +// check to enforce this. +template +Derived* CheckedDowncastToActualType(Base* base) { +#if GTEST_HAS_RTTI + GTEST_CHECK_(typeid(*base) == typeid(Derived)); +#endif + +#if GTEST_HAS_DOWNCAST_ + return ::down_cast(base); +#elif GTEST_HAS_RTTI + return dynamic_cast(base); // NOLINT +#else + return static_cast(base); // Poor man's downcast. +#endif +} + +#if GTEST_HAS_STREAM_REDIRECTION + +// Defines the stderr capturer: +// CaptureStdout - starts capturing stdout. +// GetCapturedStdout - stops capturing stdout and returns the captured string. +// CaptureStderr - starts capturing stderr. +// GetCapturedStderr - stops capturing stderr and returns the captured string. +// +GTEST_API_ void CaptureStdout(); +GTEST_API_ std::string GetCapturedStdout(); +GTEST_API_ void CaptureStderr(); +GTEST_API_ std::string GetCapturedStderr(); + +#endif // GTEST_HAS_STREAM_REDIRECTION +// Returns the size (in bytes) of a file. +GTEST_API_ size_t GetFileSize(FILE* file); + +// Reads the entire content of a file as a string. +GTEST_API_ std::string ReadEntireFile(FILE* file); + +// All command line arguments. +GTEST_API_ std::vector GetArgvs(); + +#if GTEST_HAS_DEATH_TEST + +std::vector GetInjectableArgvs(); +// Deprecated: pass the args vector by value instead. +void SetInjectableArgvs(const std::vector* new_argvs); +void SetInjectableArgvs(const std::vector& new_argvs); +void ClearInjectableArgvs(); + +#endif // GTEST_HAS_DEATH_TEST + +// Defines synchronization primitives. +#if GTEST_IS_THREADSAFE +# if GTEST_HAS_PTHREAD +// Sleeps for (roughly) n milliseconds. This function is only for testing +// Google Test's own constructs. Don't use it in user tests, either +// directly or indirectly. +inline void SleepMilliseconds(int n) { + const timespec time = { + 0, // 0 seconds. + n * 1000L * 1000L, // And n ms. + }; + nanosleep(&time, nullptr); +} +# endif // GTEST_HAS_PTHREAD + +# if GTEST_HAS_NOTIFICATION_ +// Notification has already been imported into the namespace. +// Nothing to do here. + +# elif GTEST_HAS_PTHREAD +// Allows a controller thread to pause execution of newly created +// threads until notified. Instances of this class must be created +// and destroyed in the controller thread. +// +// This class is only for testing Google Test's own constructs. Do not +// use it in user tests, either directly or indirectly. +class Notification { + public: + Notification() : notified_(false) { + GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, nullptr)); + } + ~Notification() { + pthread_mutex_destroy(&mutex_); + } + + // Notifies all threads created with this notification to start. Must + // be called from the controller thread. + void Notify() { + pthread_mutex_lock(&mutex_); + notified_ = true; + pthread_mutex_unlock(&mutex_); + } + + // Blocks until the controller thread notifies. Must be called from a test + // thread. + void WaitForNotification() { + for (;;) { + pthread_mutex_lock(&mutex_); + const bool notified = notified_; + pthread_mutex_unlock(&mutex_); + if (notified) + break; + SleepMilliseconds(10); + } + } + + private: + pthread_mutex_t mutex_; + bool notified_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification); +}; + +# elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT + +GTEST_API_ void SleepMilliseconds(int n); + +// Provides leak-safe Windows kernel handle ownership. +// Used in death tests and in threading support. +class GTEST_API_ AutoHandle { + public: + // Assume that Win32 HANDLE type is equivalent to void*. Doing so allows us to + // avoid including in this header file. Including is + // undesirable because it defines a lot of symbols and macros that tend to + // conflict with client code. This assumption is verified by + // WindowsTypesTest.HANDLEIsVoidStar. + typedef void* Handle; + AutoHandle(); + explicit AutoHandle(Handle handle); + + ~AutoHandle(); + + Handle Get() const; + void Reset(); + void Reset(Handle handle); + + private: + // Returns true iff the handle is a valid handle object that can be closed. + bool IsCloseable() const; + + Handle handle_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(AutoHandle); +}; + +// Allows a controller thread to pause execution of newly created +// threads until notified. Instances of this class must be created +// and destroyed in the controller thread. +// +// This class is only for testing Google Test's own constructs. Do not +// use it in user tests, either directly or indirectly. +class GTEST_API_ Notification { + public: + Notification(); + void Notify(); + void WaitForNotification(); + + private: + AutoHandle event_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification); +}; +# endif // GTEST_HAS_NOTIFICATION_ + +// On MinGW, we can have both GTEST_OS_WINDOWS and GTEST_HAS_PTHREAD +// defined, but we don't want to use MinGW's pthreads implementation, which +// has conformance problems with some versions of the POSIX standard. +# if GTEST_HAS_PTHREAD && !GTEST_OS_WINDOWS_MINGW + +// As a C-function, ThreadFuncWithCLinkage cannot be templated itself. +// Consequently, it cannot select a correct instantiation of ThreadWithParam +// in order to call its Run(). Introducing ThreadWithParamBase as a +// non-templated base class for ThreadWithParam allows us to bypass this +// problem. +class ThreadWithParamBase { + public: + virtual ~ThreadWithParamBase() {} + virtual void Run() = 0; +}; + +// pthread_create() accepts a pointer to a function type with the C linkage. +// According to the Standard (7.5/1), function types with different linkages +// are different even if they are otherwise identical. Some compilers (for +// example, SunStudio) treat them as different types. Since class methods +// cannot be defined with C-linkage we need to define a free C-function to +// pass into pthread_create(). +extern "C" inline void* ThreadFuncWithCLinkage(void* thread) { + static_cast(thread)->Run(); + return nullptr; +} + +// Helper class for testing Google Test's multi-threading constructs. +// To use it, write: +// +// void ThreadFunc(int param) { /* Do things with param */ } +// Notification thread_can_start; +// ... +// // The thread_can_start parameter is optional; you can supply NULL. +// ThreadWithParam thread(&ThreadFunc, 5, &thread_can_start); +// thread_can_start.Notify(); +// +// These classes are only for testing Google Test's own constructs. Do +// not use them in user tests, either directly or indirectly. +template +class ThreadWithParam : public ThreadWithParamBase { + public: + typedef void UserThreadFunc(T); + + ThreadWithParam(UserThreadFunc* func, T param, Notification* thread_can_start) + : func_(func), + param_(param), + thread_can_start_(thread_can_start), + finished_(false) { + ThreadWithParamBase* const base = this; + // The thread can be created only after all fields except thread_ + // have been initialized. + GTEST_CHECK_POSIX_SUCCESS_( + pthread_create(&thread_, nullptr, &ThreadFuncWithCLinkage, base)); + } + ~ThreadWithParam() override { Join(); } + + void Join() { + if (!finished_) { + GTEST_CHECK_POSIX_SUCCESS_(pthread_join(thread_, nullptr)); + finished_ = true; + } + } + + void Run() override { + if (thread_can_start_ != nullptr) thread_can_start_->WaitForNotification(); + func_(param_); + } + + private: + UserThreadFunc* const func_; // User-supplied thread function. + const T param_; // User-supplied parameter to the thread function. + // When non-NULL, used to block execution until the controller thread + // notifies. + Notification* const thread_can_start_; + bool finished_; // true iff we know that the thread function has finished. + pthread_t thread_; // The native thread object. + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam); +}; +# endif // !GTEST_OS_WINDOWS && GTEST_HAS_PTHREAD || + // GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ + +# if GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ +// Mutex and ThreadLocal have already been imported into the namespace. +// Nothing to do here. + +# elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT + +// Mutex implements mutex on Windows platforms. It is used in conjunction +// with class MutexLock: +// +// Mutex mutex; +// ... +// MutexLock lock(&mutex); // Acquires the mutex and releases it at the +// // end of the current scope. +// +// A static Mutex *must* be defined or declared using one of the following +// macros: +// GTEST_DEFINE_STATIC_MUTEX_(g_some_mutex); +// GTEST_DECLARE_STATIC_MUTEX_(g_some_mutex); +// +// (A non-static Mutex is defined/declared in the usual way). +class GTEST_API_ Mutex { + public: + enum MutexType { kStatic = 0, kDynamic = 1 }; + // We rely on kStaticMutex being 0 as it is to what the linker initializes + // type_ in static mutexes. critical_section_ will be initialized lazily + // in ThreadSafeLazyInit(). + enum StaticConstructorSelector { kStaticMutex = 0 }; + + // This constructor intentionally does nothing. It relies on type_ being + // statically initialized to 0 (effectively setting it to kStatic) and on + // ThreadSafeLazyInit() to lazily initialize the rest of the members. + explicit Mutex(StaticConstructorSelector /*dummy*/) {} + + Mutex(); + ~Mutex(); + + void Lock(); + + void Unlock(); + + // Does nothing if the current thread holds the mutex. Otherwise, crashes + // with high probability. + void AssertHeld(); + + private: + // Initializes owner_thread_id_ and critical_section_ in static mutexes. + void ThreadSafeLazyInit(); + + // Per https://blogs.msdn.microsoft.com/oldnewthing/20040223-00/?p=40503, + // we assume that 0 is an invalid value for thread IDs. + unsigned int owner_thread_id_; + + // For static mutexes, we rely on these members being initialized to zeros + // by the linker. + MutexType type_; + long critical_section_init_phase_; // NOLINT + GTEST_CRITICAL_SECTION* critical_section_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex); +}; + +# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \ + extern ::testing::internal::Mutex mutex + +# define GTEST_DEFINE_STATIC_MUTEX_(mutex) \ + ::testing::internal::Mutex mutex(::testing::internal::Mutex::kStaticMutex) + +// We cannot name this class MutexLock because the ctor declaration would +// conflict with a macro named MutexLock, which is defined on some +// platforms. That macro is used as a defensive measure to prevent against +// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than +// "MutexLock l(&mu)". Hence the typedef trick below. +class GTestMutexLock { + public: + explicit GTestMutexLock(Mutex* mutex) + : mutex_(mutex) { mutex_->Lock(); } + + ~GTestMutexLock() { mutex_->Unlock(); } + + private: + Mutex* const mutex_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock); +}; + +typedef GTestMutexLock MutexLock; + +// Base class for ValueHolder. Allows a caller to hold and delete a value +// without knowing its type. +class ThreadLocalValueHolderBase { + public: + virtual ~ThreadLocalValueHolderBase() {} +}; + +// Provides a way for a thread to send notifications to a ThreadLocal +// regardless of its parameter type. +class ThreadLocalBase { + public: + // Creates a new ValueHolder object holding a default value passed to + // this ThreadLocal's constructor and returns it. It is the caller's + // responsibility not to call this when the ThreadLocal instance already + // has a value on the current thread. + virtual ThreadLocalValueHolderBase* NewValueForCurrentThread() const = 0; + + protected: + ThreadLocalBase() {} + virtual ~ThreadLocalBase() {} + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocalBase); +}; + +// Maps a thread to a set of ThreadLocals that have values instantiated on that +// thread and notifies them when the thread exits. A ThreadLocal instance is +// expected to persist until all threads it has values on have terminated. +class GTEST_API_ ThreadLocalRegistry { + public: + // Registers thread_local_instance as having value on the current thread. + // Returns a value that can be used to identify the thread from other threads. + static ThreadLocalValueHolderBase* GetValueOnCurrentThread( + const ThreadLocalBase* thread_local_instance); + + // Invoked when a ThreadLocal instance is destroyed. + static void OnThreadLocalDestroyed( + const ThreadLocalBase* thread_local_instance); +}; + +class GTEST_API_ ThreadWithParamBase { + public: + void Join(); + + protected: + class Runnable { + public: + virtual ~Runnable() {} + virtual void Run() = 0; + }; + + ThreadWithParamBase(Runnable *runnable, Notification* thread_can_start); + virtual ~ThreadWithParamBase(); + + private: + AutoHandle thread_; +}; + +// Helper class for testing Google Test's multi-threading constructs. +template +class ThreadWithParam : public ThreadWithParamBase { + public: + typedef void UserThreadFunc(T); + + ThreadWithParam(UserThreadFunc* func, T param, Notification* thread_can_start) + : ThreadWithParamBase(new RunnableImpl(func, param), thread_can_start) { + } + virtual ~ThreadWithParam() {} + + private: + class RunnableImpl : public Runnable { + public: + RunnableImpl(UserThreadFunc* func, T param) + : func_(func), + param_(param) { + } + virtual ~RunnableImpl() {} + virtual void Run() { + func_(param_); + } + + private: + UserThreadFunc* const func_; + const T param_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(RunnableImpl); + }; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam); +}; + +// Implements thread-local storage on Windows systems. +// +// // Thread 1 +// ThreadLocal tl(100); // 100 is the default value for each thread. +// +// // Thread 2 +// tl.set(150); // Changes the value for thread 2 only. +// EXPECT_EQ(150, tl.get()); +// +// // Thread 1 +// EXPECT_EQ(100, tl.get()); // In thread 1, tl has the original value. +// tl.set(200); +// EXPECT_EQ(200, tl.get()); +// +// The template type argument T must have a public copy constructor. +// In addition, the default ThreadLocal constructor requires T to have +// a public default constructor. +// +// The users of a TheadLocal instance have to make sure that all but one +// threads (including the main one) using that instance have exited before +// destroying it. Otherwise, the per-thread objects managed for them by the +// ThreadLocal instance are not guaranteed to be destroyed on all platforms. +// +// Google Test only uses global ThreadLocal objects. That means they +// will die after main() has returned. Therefore, no per-thread +// object managed by Google Test will be leaked as long as all threads +// using Google Test have exited when main() returns. +template +class ThreadLocal : public ThreadLocalBase { + public: + ThreadLocal() : default_factory_(new DefaultValueHolderFactory()) {} + explicit ThreadLocal(const T& value) + : default_factory_(new InstanceValueHolderFactory(value)) {} + + ~ThreadLocal() { ThreadLocalRegistry::OnThreadLocalDestroyed(this); } + + T* pointer() { return GetOrCreateValue(); } + const T* pointer() const { return GetOrCreateValue(); } + const T& get() const { return *pointer(); } + void set(const T& value) { *pointer() = value; } + + private: + // Holds a value of T. Can be deleted via its base class without the caller + // knowing the type of T. + class ValueHolder : public ThreadLocalValueHolderBase { + public: + ValueHolder() : value_() {} + explicit ValueHolder(const T& value) : value_(value) {} + + T* pointer() { return &value_; } + + private: + T value_; + GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder); + }; + + + T* GetOrCreateValue() const { + return static_cast( + ThreadLocalRegistry::GetValueOnCurrentThread(this))->pointer(); + } + + virtual ThreadLocalValueHolderBase* NewValueForCurrentThread() const { + return default_factory_->MakeNewHolder(); + } + + class ValueHolderFactory { + public: + ValueHolderFactory() {} + virtual ~ValueHolderFactory() {} + virtual ValueHolder* MakeNewHolder() const = 0; + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolderFactory); + }; + + class DefaultValueHolderFactory : public ValueHolderFactory { + public: + DefaultValueHolderFactory() {} + virtual ValueHolder* MakeNewHolder() const { return new ValueHolder(); } + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultValueHolderFactory); + }; + + class InstanceValueHolderFactory : public ValueHolderFactory { + public: + explicit InstanceValueHolderFactory(const T& value) : value_(value) {} + virtual ValueHolder* MakeNewHolder() const { + return new ValueHolder(value_); + } + + private: + const T value_; // The value for each thread. + + GTEST_DISALLOW_COPY_AND_ASSIGN_(InstanceValueHolderFactory); + }; + + std::unique_ptr default_factory_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal); +}; + +# elif GTEST_HAS_PTHREAD + +// MutexBase and Mutex implement mutex on pthreads-based platforms. +class MutexBase { + public: + // Acquires this mutex. + void Lock() { + GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_lock(&mutex_)); + owner_ = pthread_self(); + has_owner_ = true; + } + + // Releases this mutex. + void Unlock() { + // Since the lock is being released the owner_ field should no longer be + // considered valid. We don't protect writing to has_owner_ here, as it's + // the caller's responsibility to ensure that the current thread holds the + // mutex when this is called. + has_owner_ = false; + GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_unlock(&mutex_)); + } + + // Does nothing if the current thread holds the mutex. Otherwise, crashes + // with high probability. + void AssertHeld() const { + GTEST_CHECK_(has_owner_ && pthread_equal(owner_, pthread_self())) + << "The current thread is not holding the mutex @" << this; + } + + // A static mutex may be used before main() is entered. It may even + // be used before the dynamic initialization stage. Therefore we + // must be able to initialize a static mutex object at link time. + // This means MutexBase has to be a POD and its member variables + // have to be public. + public: + pthread_mutex_t mutex_; // The underlying pthread mutex. + // has_owner_ indicates whether the owner_ field below contains a valid thread + // ID and is therefore safe to inspect (e.g., to use in pthread_equal()). All + // accesses to the owner_ field should be protected by a check of this field. + // An alternative might be to memset() owner_ to all zeros, but there's no + // guarantee that a zero'd pthread_t is necessarily invalid or even different + // from pthread_self(). + bool has_owner_; + pthread_t owner_; // The thread holding the mutex. +}; + +// Forward-declares a static mutex. +# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \ + extern ::testing::internal::MutexBase mutex + +// Defines and statically (i.e. at link time) initializes a static mutex. +// The initialization list here does not explicitly initialize each field, +// instead relying on default initialization for the unspecified fields. In +// particular, the owner_ field (a pthread_t) is not explicitly initialized. +// This allows initialization to work whether pthread_t is a scalar or struct. +// The flag -Wmissing-field-initializers must not be specified for this to work. +#define GTEST_DEFINE_STATIC_MUTEX_(mutex) \ + ::testing::internal::MutexBase mutex = {PTHREAD_MUTEX_INITIALIZER, false, 0} + +// The Mutex class can only be used for mutexes created at runtime. It +// shares its API with MutexBase otherwise. +class Mutex : public MutexBase { + public: + Mutex() { + GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, nullptr)); + has_owner_ = false; + } + ~Mutex() { + GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_destroy(&mutex_)); + } + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex); +}; + +// We cannot name this class MutexLock because the ctor declaration would +// conflict with a macro named MutexLock, which is defined on some +// platforms. That macro is used as a defensive measure to prevent against +// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than +// "MutexLock l(&mu)". Hence the typedef trick below. +class GTestMutexLock { + public: + explicit GTestMutexLock(MutexBase* mutex) + : mutex_(mutex) { mutex_->Lock(); } + + ~GTestMutexLock() { mutex_->Unlock(); } + + private: + MutexBase* const mutex_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock); +}; + +typedef GTestMutexLock MutexLock; + +// Helpers for ThreadLocal. + +// pthread_key_create() requires DeleteThreadLocalValue() to have +// C-linkage. Therefore it cannot be templatized to access +// ThreadLocal. Hence the need for class +// ThreadLocalValueHolderBase. +class ThreadLocalValueHolderBase { + public: + virtual ~ThreadLocalValueHolderBase() {} +}; + +// Called by pthread to delete thread-local data stored by +// pthread_setspecific(). +extern "C" inline void DeleteThreadLocalValue(void* value_holder) { + delete static_cast(value_holder); +} + +// Implements thread-local storage on pthreads-based systems. +template +class GTEST_API_ ThreadLocal { + public: + ThreadLocal() + : key_(CreateKey()), default_factory_(new DefaultValueHolderFactory()) {} + explicit ThreadLocal(const T& value) + : key_(CreateKey()), + default_factory_(new InstanceValueHolderFactory(value)) {} + + ~ThreadLocal() { + // Destroys the managed object for the current thread, if any. + DeleteThreadLocalValue(pthread_getspecific(key_)); + + // Releases resources associated with the key. This will *not* + // delete managed objects for other threads. + GTEST_CHECK_POSIX_SUCCESS_(pthread_key_delete(key_)); + } + + T* pointer() { return GetOrCreateValue(); } + const T* pointer() const { return GetOrCreateValue(); } + const T& get() const { return *pointer(); } + void set(const T& value) { *pointer() = value; } + + private: + // Holds a value of type T. + class ValueHolder : public ThreadLocalValueHolderBase { + public: + ValueHolder() : value_() {} + explicit ValueHolder(const T& value) : value_(value) {} + + T* pointer() { return &value_; } + + private: + T value_; + GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder); + }; + + static pthread_key_t CreateKey() { + pthread_key_t key; + // When a thread exits, DeleteThreadLocalValue() will be called on + // the object managed for that thread. + GTEST_CHECK_POSIX_SUCCESS_( + pthread_key_create(&key, &DeleteThreadLocalValue)); + return key; + } + + T* GetOrCreateValue() const { + ThreadLocalValueHolderBase* const holder = + static_cast(pthread_getspecific(key_)); + if (holder != nullptr) { + return CheckedDowncastToActualType(holder)->pointer(); + } + + ValueHolder* const new_holder = default_factory_->MakeNewHolder(); + ThreadLocalValueHolderBase* const holder_base = new_holder; + GTEST_CHECK_POSIX_SUCCESS_(pthread_setspecific(key_, holder_base)); + return new_holder->pointer(); + } + + class ValueHolderFactory { + public: + ValueHolderFactory() {} + virtual ~ValueHolderFactory() {} + virtual ValueHolder* MakeNewHolder() const = 0; + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolderFactory); + }; + + class DefaultValueHolderFactory : public ValueHolderFactory { + public: + DefaultValueHolderFactory() {} + virtual ValueHolder* MakeNewHolder() const { return new ValueHolder(); } + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultValueHolderFactory); + }; + + class InstanceValueHolderFactory : public ValueHolderFactory { + public: + explicit InstanceValueHolderFactory(const T& value) : value_(value) {} + virtual ValueHolder* MakeNewHolder() const { + return new ValueHolder(value_); + } + + private: + const T value_; // The value for each thread. + + GTEST_DISALLOW_COPY_AND_ASSIGN_(InstanceValueHolderFactory); + }; + + // A key pthreads uses for looking up per-thread values. + const pthread_key_t key_; + std::unique_ptr default_factory_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal); +}; + +# endif // GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ + +#else // GTEST_IS_THREADSAFE + +// A dummy implementation of synchronization primitives (mutex, lock, +// and thread-local variable). Necessary for compiling Google Test where +// mutex is not supported - using Google Test in multiple threads is not +// supported on such platforms. + +class Mutex { + public: + Mutex() {} + void Lock() {} + void Unlock() {} + void AssertHeld() const {} +}; + +# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \ + extern ::testing::internal::Mutex mutex + +# define GTEST_DEFINE_STATIC_MUTEX_(mutex) ::testing::internal::Mutex mutex + +// We cannot name this class MutexLock because the ctor declaration would +// conflict with a macro named MutexLock, which is defined on some +// platforms. That macro is used as a defensive measure to prevent against +// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than +// "MutexLock l(&mu)". Hence the typedef trick below. +class GTestMutexLock { + public: + explicit GTestMutexLock(Mutex*) {} // NOLINT +}; + +typedef GTestMutexLock MutexLock; + +template +class GTEST_API_ ThreadLocal { + public: + ThreadLocal() : value_() {} + explicit ThreadLocal(const T& value) : value_(value) {} + T* pointer() { return &value_; } + const T* pointer() const { return &value_; } + const T& get() const { return value_; } + void set(const T& value) { value_ = value; } + private: + T value_; +}; + +#endif // GTEST_IS_THREADSAFE + +// Returns the number of threads running in the process, or 0 to indicate that +// we cannot detect it. +GTEST_API_ size_t GetThreadCount(); + +template +struct bool_constant { + typedef bool_constant type; + static const bool value = bool_value; +}; +template const bool bool_constant::value; + +typedef bool_constant false_type; +typedef bool_constant true_type; + +template +struct is_same : public false_type {}; + +template +struct is_same : public true_type {}; + +template +struct IteratorTraits { + typedef typename Iterator::value_type value_type; +}; + + +template +struct IteratorTraits { + typedef T value_type; +}; + +template +struct IteratorTraits { + typedef T value_type; +}; + +#if GTEST_OS_WINDOWS +# define GTEST_PATH_SEP_ "\\" +# define GTEST_HAS_ALT_PATH_SEP_ 1 +// The biggest signed integer type the compiler supports. +typedef __int64 BiggestInt; +#else +# define GTEST_PATH_SEP_ "/" +# define GTEST_HAS_ALT_PATH_SEP_ 0 +typedef long long BiggestInt; // NOLINT +#endif // GTEST_OS_WINDOWS + +// Utilities for char. + +// isspace(int ch) and friends accept an unsigned char or EOF. char +// may be signed, depending on the compiler (or compiler flags). +// Therefore we need to cast a char to unsigned char before calling +// isspace(), etc. + +inline bool IsAlpha(char ch) { + return isalpha(static_cast(ch)) != 0; +} +inline bool IsAlNum(char ch) { + return isalnum(static_cast(ch)) != 0; +} +inline bool IsDigit(char ch) { + return isdigit(static_cast(ch)) != 0; +} +inline bool IsLower(char ch) { + return islower(static_cast(ch)) != 0; +} +inline bool IsSpace(char ch) { + return isspace(static_cast(ch)) != 0; +} +inline bool IsUpper(char ch) { + return isupper(static_cast(ch)) != 0; +} +inline bool IsXDigit(char ch) { + return isxdigit(static_cast(ch)) != 0; +} +inline bool IsXDigit(wchar_t ch) { + const unsigned char low_byte = static_cast(ch); + return ch == low_byte && isxdigit(low_byte) != 0; +} + +inline char ToLower(char ch) { + return static_cast(tolower(static_cast(ch))); +} +inline char ToUpper(char ch) { + return static_cast(toupper(static_cast(ch))); +} + +inline std::string StripTrailingSpaces(std::string str) { + std::string::iterator it = str.end(); + while (it != str.begin() && IsSpace(*--it)) + it = str.erase(it); + return str; +} + +// The testing::internal::posix namespace holds wrappers for common +// POSIX functions. These wrappers hide the differences between +// Windows/MSVC and POSIX systems. Since some compilers define these +// standard functions as macros, the wrapper cannot have the same name +// as the wrapped function. + +namespace posix { + +// Functions with a different name on Windows. + +#if GTEST_OS_WINDOWS + +typedef struct _stat StatStruct; + +# ifdef __BORLANDC__ +inline int IsATTY(int fd) { return isatty(fd); } +inline int StrCaseCmp(const char* s1, const char* s2) { + return stricmp(s1, s2); +} +inline char* StrDup(const char* src) { return strdup(src); } +# else // !__BORLANDC__ +# if GTEST_OS_WINDOWS_MOBILE +inline int IsATTY(int /* fd */) { return 0; } +# else +inline int IsATTY(int fd) { return _isatty(fd); } +# endif // GTEST_OS_WINDOWS_MOBILE +inline int StrCaseCmp(const char* s1, const char* s2) { + return _stricmp(s1, s2); +} +inline char* StrDup(const char* src) { return _strdup(src); } +# endif // __BORLANDC__ + +# if GTEST_OS_WINDOWS_MOBILE +inline int FileNo(FILE* file) { return reinterpret_cast(_fileno(file)); } +// Stat(), RmDir(), and IsDir() are not needed on Windows CE at this +// time and thus not defined there. +# else +inline int FileNo(FILE* file) { return _fileno(file); } +inline int Stat(const char* path, StatStruct* buf) { return _stat(path, buf); } +inline int RmDir(const char* dir) { return _rmdir(dir); } +inline bool IsDir(const StatStruct& st) { + return (_S_IFDIR & st.st_mode) != 0; +} +# endif // GTEST_OS_WINDOWS_MOBILE + +#else + +typedef struct stat StatStruct; + +inline int FileNo(FILE* file) { return fileno(file); } +inline int IsATTY(int fd) { return isatty(fd); } +inline int Stat(const char* path, StatStruct* buf) { return stat(path, buf); } +inline int StrCaseCmp(const char* s1, const char* s2) { + return strcasecmp(s1, s2); +} +inline char* StrDup(const char* src) { return strdup(src); } +inline int RmDir(const char* dir) { return rmdir(dir); } +inline bool IsDir(const StatStruct& st) { return S_ISDIR(st.st_mode); } + +#endif // GTEST_OS_WINDOWS + +// Functions deprecated by MSVC 8.0. + +GTEST_DISABLE_MSC_DEPRECATED_PUSH_() + +inline const char* StrNCpy(char* dest, const char* src, size_t n) { + return strncpy(dest, src, n); +} + +// ChDir(), FReopen(), FDOpen(), Read(), Write(), Close(), and +// StrError() aren't needed on Windows CE at this time and thus not +// defined there. + +#if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT +inline int ChDir(const char* dir) { +#if GTEST_OS_WINDOWS + return _chdir(dir); +#else + return chdir(dir); +#endif +} +#endif +inline FILE* FOpen(const char* path, const char* mode) { + return fopen(path, mode); +} +#if !GTEST_OS_WINDOWS_MOBILE +inline FILE *FReopen(const char* path, const char* mode, FILE* stream) { + return freopen(path, mode, stream); +} +inline FILE* FDOpen(int fd, const char* mode) { +#if GTEST_OS_WINDOWS + return _fdopen(fd, mode); +#else + return fdopen(fd, mode); +#endif +} +#endif +inline int FClose(FILE* fp) { return fclose(fp); } +#if !GTEST_OS_WINDOWS_MOBILE +inline int Read(int fd, void* buf, unsigned int count) { +#if GTEST_OS_WINDOWS + return static_cast(_read(fd, buf, count)); +#else + return static_cast(read(fd, buf, count)); +#endif +} +inline int Write(int fd, const void* buf, unsigned int count) { +#if GTEST_OS_WINDOWS + return static_cast(_write(fd, buf, count)); +#else + return static_cast(write(fd, buf, count)); +#endif +} +inline int Close(int fd) { +#if GTEST_OS_WINDOWS + return _close(fd); +#else + return close(fd); +#endif +} +inline const char* StrError(int errnum) { return strerror(errnum); } +#endif +inline const char* GetEnv(const char* name) { +#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT + // We are on Windows CE, which has no environment variables. + static_cast(name); // To prevent 'unused argument' warning. + return nullptr; +#elif defined(__BORLANDC__) || defined(__SunOS_5_8) || defined(__SunOS_5_9) + // Environment variables which we programmatically clear will be set to the + // empty string rather than unset (NULL). Handle that case. + const char* const env = getenv(name); + return (env != nullptr && env[0] != '\0') ? env : nullptr; +#else + return getenv(name); +#endif +} + +GTEST_DISABLE_MSC_DEPRECATED_POP_() + +#if GTEST_OS_WINDOWS_MOBILE +// Windows CE has no C library. The abort() function is used in +// several places in Google Test. This implementation provides a reasonable +// imitation of standard behaviour. +[[noreturn]] void Abort(); +#else +[[noreturn]] inline void Abort() { abort(); } +#endif // GTEST_OS_WINDOWS_MOBILE + +} // namespace posix + +// MSVC "deprecates" snprintf and issues warnings wherever it is used. In +// order to avoid these warnings, we need to use _snprintf or _snprintf_s on +// MSVC-based platforms. We map the GTEST_SNPRINTF_ macro to the appropriate +// function in order to achieve that. We use macro definition here because +// snprintf is a variadic function. +#if _MSC_VER && !GTEST_OS_WINDOWS_MOBILE +// MSVC 2005 and above support variadic macros. +# define GTEST_SNPRINTF_(buffer, size, format, ...) \ + _snprintf_s(buffer, size, size, format, __VA_ARGS__) +#elif defined(_MSC_VER) +// Windows CE does not define _snprintf_s +# define GTEST_SNPRINTF_ _snprintf +#else +# define GTEST_SNPRINTF_ snprintf +#endif + +// The maximum number a BiggestInt can represent. This definition +// works no matter BiggestInt is represented in one's complement or +// two's complement. +// +// We cannot rely on numeric_limits in STL, as __int64 and long long +// are not part of standard C++ and numeric_limits doesn't need to be +// defined for them. +const BiggestInt kMaxBiggestInt = + ~(static_cast(1) << (8*sizeof(BiggestInt) - 1)); + +// This template class serves as a compile-time function from size to +// type. It maps a size in bytes to a primitive type with that +// size. e.g. +// +// TypeWithSize<4>::UInt +// +// is typedef-ed to be unsigned int (unsigned integer made up of 4 +// bytes). +// +// Such functionality should belong to STL, but I cannot find it +// there. +// +// Google Test uses this class in the implementation of floating-point +// comparison. +// +// For now it only handles UInt (unsigned int) as that's all Google Test +// needs. Other types can be easily added in the future if need +// arises. +template +class TypeWithSize { + public: + // This prevents the user from using TypeWithSize with incorrect + // values of N. + typedef void UInt; +}; + +// The specialization for size 4. +template <> +class TypeWithSize<4> { + public: + // unsigned int has size 4 in both gcc and MSVC. + // + // As base/basictypes.h doesn't compile on Windows, we cannot use + // uint32, uint64, and etc here. + typedef int Int; + typedef unsigned int UInt; +}; + +// The specialization for size 8. +template <> +class TypeWithSize<8> { + public: +#if GTEST_OS_WINDOWS + typedef __int64 Int; + typedef unsigned __int64 UInt; +#else + typedef long long Int; // NOLINT + typedef unsigned long long UInt; // NOLINT +#endif // GTEST_OS_WINDOWS +}; + +// Integer types of known sizes. +typedef TypeWithSize<4>::Int Int32; +typedef TypeWithSize<4>::UInt UInt32; +typedef TypeWithSize<8>::Int Int64; +typedef TypeWithSize<8>::UInt UInt64; +typedef TypeWithSize<8>::Int TimeInMillis; // Represents time in milliseconds. + +// Utilities for command line flags and environment variables. + +// Macro for referencing flags. +#if !defined(GTEST_FLAG) +# define GTEST_FLAG(name) FLAGS_gtest_##name +#endif // !defined(GTEST_FLAG) + +#if !defined(GTEST_USE_OWN_FLAGFILE_FLAG_) +# define GTEST_USE_OWN_FLAGFILE_FLAG_ 1 +#endif // !defined(GTEST_USE_OWN_FLAGFILE_FLAG_) + +#if !defined(GTEST_DECLARE_bool_) +# define GTEST_FLAG_SAVER_ ::testing::internal::GTestFlagSaver + +// Macros for declaring flags. +# define GTEST_DECLARE_bool_(name) GTEST_API_ extern bool GTEST_FLAG(name) +# define GTEST_DECLARE_int32_(name) \ + GTEST_API_ extern ::testing::internal::Int32 GTEST_FLAG(name) +# define GTEST_DECLARE_string_(name) \ + GTEST_API_ extern ::std::string GTEST_FLAG(name) + +// Macros for defining flags. +# define GTEST_DEFINE_bool_(name, default_val, doc) \ + GTEST_API_ bool GTEST_FLAG(name) = (default_val) +# define GTEST_DEFINE_int32_(name, default_val, doc) \ + GTEST_API_ ::testing::internal::Int32 GTEST_FLAG(name) = (default_val) +# define GTEST_DEFINE_string_(name, default_val, doc) \ + GTEST_API_ ::std::string GTEST_FLAG(name) = (default_val) + +#endif // !defined(GTEST_DECLARE_bool_) + +// Thread annotations +#if !defined(GTEST_EXCLUSIVE_LOCK_REQUIRED_) +# define GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks) +# define GTEST_LOCK_EXCLUDED_(locks) +#endif // !defined(GTEST_EXCLUSIVE_LOCK_REQUIRED_) + +// Parses 'str' for a 32-bit signed integer. If successful, writes the result +// to *value and returns true; otherwise leaves *value unchanged and returns +// false. +bool ParseInt32(const Message& src_text, const char* str, Int32* value); + +// Parses a bool/Int32/string from the environment variable +// corresponding to the given Google Test flag. +bool BoolFromGTestEnv(const char* flag, bool default_val); +GTEST_API_ Int32 Int32FromGTestEnv(const char* flag, Int32 default_val); +std::string OutputFlagAlsoCheckEnvVar(); +const char* StringFromGTestEnv(const char* flag, const char* default_val); + +} // namespace internal +} // namespace testing + +#if !defined(GTEST_INTERNAL_DEPRECATED) + +// Internal Macro to mark an API deprecated, for googletest usage only +// Usage: class GTEST_INTERNAL_DEPRECATED(message) MyClass or +// GTEST_INTERNAL_DEPRECATED(message) myFunction(); Every usage of +// a deprecated entity will trigger a warning when compiled with +// `-Wdeprecated-declarations` option (clang, gcc, any __GNUC__ compiler). +// For msvc /W3 option will need to be used +// Note that for 'other' compilers this macro evaluates to nothing to prevent +// compilations errors. +#if defined(_MSC_VER) +#define GTEST_INTERNAL_DEPRECATED(message) __declspec(deprecated(message)) +#elif defined(__GNUC__) +#define GTEST_INTERNAL_DEPRECATED(message) __attribute__((deprecated(message))) +#else +#define GTEST_INTERNAL_DEPRECATED(message) +#endif + +#endif // !defined(GTEST_INTERNAL_DEPRECATED) + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-string.h b/deps/googletest/include/gtest/internal/gtest-string.h new file mode 100644 index 000000000..884b1e16f --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-string.h @@ -0,0 +1,170 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This header file declares the String class and functions used internally by +// Google Test. They are subject to change without notice. They should not used +// by code external to Google Test. +// +// This header file is #included by gtest-internal.h. +// It should not be #included by other files. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_ + +#ifdef __BORLANDC__ +// string.h is not guaranteed to provide strcpy on C++ Builder. +# include +#endif + +#include +#include + +#include "gtest/internal/gtest-port.h" + +namespace testing { +namespace internal { + +// String - an abstract class holding static string utilities. +class GTEST_API_ String { + public: + // Static utility methods + + // Clones a 0-terminated C string, allocating memory using new. The + // caller is responsible for deleting the return value using + // delete[]. Returns the cloned string, or NULL if the input is + // NULL. + // + // This is different from strdup() in string.h, which allocates + // memory using malloc(). + static const char* CloneCString(const char* c_str); + +#if GTEST_OS_WINDOWS_MOBILE + // Windows CE does not have the 'ANSI' versions of Win32 APIs. To be + // able to pass strings to Win32 APIs on CE we need to convert them + // to 'Unicode', UTF-16. + + // Creates a UTF-16 wide string from the given ANSI string, allocating + // memory using new. The caller is responsible for deleting the return + // value using delete[]. Returns the wide string, or NULL if the + // input is NULL. + // + // The wide string is created using the ANSI codepage (CP_ACP) to + // match the behaviour of the ANSI versions of Win32 calls and the + // C runtime. + static LPCWSTR AnsiToUtf16(const char* c_str); + + // Creates an ANSI string from the given wide string, allocating + // memory using new. The caller is responsible for deleting the return + // value using delete[]. Returns the ANSI string, or NULL if the + // input is NULL. + // + // The returned string is created using the ANSI codepage (CP_ACP) to + // match the behaviour of the ANSI versions of Win32 calls and the + // C runtime. + static const char* Utf16ToAnsi(LPCWSTR utf16_str); +#endif + + // Compares two C strings. Returns true iff they have the same content. + // + // Unlike strcmp(), this function can handle NULL argument(s). A + // NULL C string is considered different to any non-NULL C string, + // including the empty string. + static bool CStringEquals(const char* lhs, const char* rhs); + + // Converts a wide C string to a String using the UTF-8 encoding. + // NULL will be converted to "(null)". If an error occurred during + // the conversion, "(failed to convert from wide string)" is + // returned. + static std::string ShowWideCString(const wchar_t* wide_c_str); + + // Compares two wide C strings. Returns true iff they have the same + // content. + // + // Unlike wcscmp(), this function can handle NULL argument(s). A + // NULL C string is considered different to any non-NULL C string, + // including the empty string. + static bool WideCStringEquals(const wchar_t* lhs, const wchar_t* rhs); + + // Compares two C strings, ignoring case. Returns true iff they + // have the same content. + // + // Unlike strcasecmp(), this function can handle NULL argument(s). + // A NULL C string is considered different to any non-NULL C string, + // including the empty string. + static bool CaseInsensitiveCStringEquals(const char* lhs, + const char* rhs); + + // Compares two wide C strings, ignoring case. Returns true iff they + // have the same content. + // + // Unlike wcscasecmp(), this function can handle NULL argument(s). + // A NULL C string is considered different to any non-NULL wide C string, + // including the empty string. + // NB: The implementations on different platforms slightly differ. + // On windows, this method uses _wcsicmp which compares according to LC_CTYPE + // environment variable. On GNU platform this method uses wcscasecmp + // which compares according to LC_CTYPE category of the current locale. + // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the + // current locale. + static bool CaseInsensitiveWideCStringEquals(const wchar_t* lhs, + const wchar_t* rhs); + + // Returns true iff the given string ends with the given suffix, ignoring + // case. Any string is considered to end with an empty suffix. + static bool EndsWithCaseInsensitive( + const std::string& str, const std::string& suffix); + + // Formats an int value as "%02d". + static std::string FormatIntWidth2(int value); // "%02d" for width == 2 + + // Formats an int value as "%X". + static std::string FormatHexInt(int value); + + // Formats an int value as "%X". + static std::string FormatHexUInt32(UInt32 value); + + // Formats a byte as "%02X". + static std::string FormatByte(unsigned char value); + + private: + String(); // Not meant to be instantiated. +}; // class String + +// Gets the content of the stringstream's buffer as an std::string. Each '\0' +// character in the buffer is replaced with "\\0". +GTEST_API_ std::string StringStreamToString(::std::stringstream* stream); + +} // namespace internal +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-type-util.h b/deps/googletest/include/gtest/internal/gtest-type-util.h new file mode 100644 index 000000000..4cd1cf3ce --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-type-util.h @@ -0,0 +1,3347 @@ +// This file was GENERATED by command: +// pump.py gtest-type-util.h.pump +// DO NOT EDIT BY HAND!!! + +// Copyright 2008 Google Inc. +// All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// Type utilities needed for implementing typed and type-parameterized +// tests. This file is generated by a SCRIPT. DO NOT EDIT BY HAND! +// +// Currently we support at most 50 types in a list, and at most 50 +// type-parameterized tests in one type-parameterized test suite. +// Please contact googletestframework@googlegroups.com if you need +// more. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ + +#include "gtest/internal/gtest-port.h" + +// #ifdef __GNUC__ is too general here. It is possible to use gcc without using +// libstdc++ (which is where cxxabi.h comes from). +# if GTEST_HAS_CXXABI_H_ +# include +# elif defined(__HP_aCC) +# include +# endif // GTEST_HASH_CXXABI_H_ + +namespace testing { +namespace internal { + +// Canonicalizes a given name with respect to the Standard C++ Library. +// This handles removing the inline namespace within `std` that is +// used by various standard libraries (e.g., `std::__1`). Names outside +// of namespace std are returned unmodified. +inline std::string CanonicalizeForStdLibVersioning(std::string s) { + static const char prefix[] = "std::__"; + if (s.compare(0, strlen(prefix), prefix) == 0) { + std::string::size_type end = s.find("::", strlen(prefix)); + if (end != s.npos) { + // Erase everything between the initial `std` and the second `::`. + s.erase(strlen("std"), end - strlen("std")); + } + } + return s; +} + +// GetTypeName() returns a human-readable name of type T. +// NB: This function is also used in Google Mock, so don't move it inside of +// the typed-test-only section below. +template +std::string GetTypeName() { +# if GTEST_HAS_RTTI + + const char* const name = typeid(T).name(); +# if GTEST_HAS_CXXABI_H_ || defined(__HP_aCC) + int status = 0; + // gcc's implementation of typeid(T).name() mangles the type name, + // so we have to demangle it. +# if GTEST_HAS_CXXABI_H_ + using abi::__cxa_demangle; +# endif // GTEST_HAS_CXXABI_H_ + char* const readable_name = __cxa_demangle(name, nullptr, nullptr, &status); + const std::string name_str(status == 0 ? readable_name : name); + free(readable_name); + return CanonicalizeForStdLibVersioning(name_str); +# else + return name; +# endif // GTEST_HAS_CXXABI_H_ || __HP_aCC + +# else + + return ""; + +# endif // GTEST_HAS_RTTI +} + +#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P + +// AssertyTypeEq::type is defined iff T1 and T2 are the same +// type. This can be used as a compile-time assertion to ensure that +// two types are equal. + +template +struct AssertTypeEq; + +template +struct AssertTypeEq { + typedef bool type; +}; + +// A unique type used as the default value for the arguments of class +// template Types. This allows us to simulate variadic templates +// (e.g. Types, Type, and etc), which C++ doesn't +// support directly. +struct None {}; + +// The following family of struct and struct templates are used to +// represent type lists. In particular, TypesN +// represents a type list with N types (T1, T2, ..., and TN) in it. +// Except for Types0, every struct in the family has two member types: +// Head for the first type in the list, and Tail for the rest of the +// list. + +// The empty type list. +struct Types0 {}; + +// Type lists of length 1, 2, 3, and so on. + +template +struct Types1 { + typedef T1 Head; + typedef Types0 Tail; +}; +template +struct Types2 { + typedef T1 Head; + typedef Types1 Tail; +}; + +template +struct Types3 { + typedef T1 Head; + typedef Types2 Tail; +}; + +template +struct Types4 { + typedef T1 Head; + typedef Types3 Tail; +}; + +template +struct Types5 { + typedef T1 Head; + typedef Types4 Tail; +}; + +template +struct Types6 { + typedef T1 Head; + typedef Types5 Tail; +}; + +template +struct Types7 { + typedef T1 Head; + typedef Types6 Tail; +}; + +template +struct Types8 { + typedef T1 Head; + typedef Types7 Tail; +}; + +template +struct Types9 { + typedef T1 Head; + typedef Types8 Tail; +}; + +template +struct Types10 { + typedef T1 Head; + typedef Types9 Tail; +}; + +template +struct Types11 { + typedef T1 Head; + typedef Types10 Tail; +}; + +template +struct Types12 { + typedef T1 Head; + typedef Types11 Tail; +}; + +template +struct Types13 { + typedef T1 Head; + typedef Types12 Tail; +}; + +template +struct Types14 { + typedef T1 Head; + typedef Types13 Tail; +}; + +template +struct Types15 { + typedef T1 Head; + typedef Types14 Tail; +}; + +template +struct Types16 { + typedef T1 Head; + typedef Types15 Tail; +}; + +template +struct Types17 { + typedef T1 Head; + typedef Types16 Tail; +}; + +template +struct Types18 { + typedef T1 Head; + typedef Types17 Tail; +}; + +template +struct Types19 { + typedef T1 Head; + typedef Types18 Tail; +}; + +template +struct Types20 { + typedef T1 Head; + typedef Types19 Tail; +}; + +template +struct Types21 { + typedef T1 Head; + typedef Types20 Tail; +}; + +template +struct Types22 { + typedef T1 Head; + typedef Types21 Tail; +}; + +template +struct Types23 { + typedef T1 Head; + typedef Types22 Tail; +}; + +template +struct Types24 { + typedef T1 Head; + typedef Types23 Tail; +}; + +template +struct Types25 { + typedef T1 Head; + typedef Types24 Tail; +}; + +template +struct Types26 { + typedef T1 Head; + typedef Types25 Tail; +}; + +template +struct Types27 { + typedef T1 Head; + typedef Types26 Tail; +}; + +template +struct Types28 { + typedef T1 Head; + typedef Types27 Tail; +}; + +template +struct Types29 { + typedef T1 Head; + typedef Types28 Tail; +}; + +template +struct Types30 { + typedef T1 Head; + typedef Types29 Tail; +}; + +template +struct Types31 { + typedef T1 Head; + typedef Types30 Tail; +}; + +template +struct Types32 { + typedef T1 Head; + typedef Types31 Tail; +}; + +template +struct Types33 { + typedef T1 Head; + typedef Types32 Tail; +}; + +template +struct Types34 { + typedef T1 Head; + typedef Types33 Tail; +}; + +template +struct Types35 { + typedef T1 Head; + typedef Types34 Tail; +}; + +template +struct Types36 { + typedef T1 Head; + typedef Types35 Tail; +}; + +template +struct Types37 { + typedef T1 Head; + typedef Types36 Tail; +}; + +template +struct Types38 { + typedef T1 Head; + typedef Types37 Tail; +}; + +template +struct Types39 { + typedef T1 Head; + typedef Types38 Tail; +}; + +template +struct Types40 { + typedef T1 Head; + typedef Types39 Tail; +}; + +template +struct Types41 { + typedef T1 Head; + typedef Types40 Tail; +}; + +template +struct Types42 { + typedef T1 Head; + typedef Types41 Tail; +}; + +template +struct Types43 { + typedef T1 Head; + typedef Types42 Tail; +}; + +template +struct Types44 { + typedef T1 Head; + typedef Types43 Tail; +}; + +template +struct Types45 { + typedef T1 Head; + typedef Types44 Tail; +}; + +template +struct Types46 { + typedef T1 Head; + typedef Types45 Tail; +}; + +template +struct Types47 { + typedef T1 Head; + typedef Types46 Tail; +}; + +template +struct Types48 { + typedef T1 Head; + typedef Types47 Tail; +}; + +template +struct Types49 { + typedef T1 Head; + typedef Types48 Tail; +}; + +template +struct Types50 { + typedef T1 Head; + typedef Types49 Tail; +}; + + +} // namespace internal + +// We don't want to require the users to write TypesN<...> directly, +// as that would require them to count the length. Types<...> is much +// easier to write, but generates horrible messages when there is a +// compiler error, as gcc insists on printing out each template +// argument, even if it has the default value (this means Types +// will appear as Types in the compiler +// errors). +// +// Our solution is to combine the best part of the two approaches: a +// user would write Types, and Google Test will translate +// that to TypesN internally to make error messages +// readable. The translation is done by the 'type' member of the +// Types template. +template +struct Types { + typedef internal::Types50 type; +}; + +template <> +struct Types { + typedef internal::Types0 type; +}; +template +struct Types { + typedef internal::Types1 type; +}; +template +struct Types { + typedef internal::Types2 type; +}; +template +struct Types { + typedef internal::Types3 type; +}; +template +struct Types { + typedef internal::Types4 type; +}; +template +struct Types { + typedef internal::Types5 type; +}; +template +struct Types { + typedef internal::Types6 type; +}; +template +struct Types { + typedef internal::Types7 type; +}; +template +struct Types { + typedef internal::Types8 type; +}; +template +struct Types { + typedef internal::Types9 type; +}; +template +struct Types { + typedef internal::Types10 type; +}; +template +struct Types { + typedef internal::Types11 type; +}; +template +struct Types { + typedef internal::Types12 type; +}; +template +struct Types { + typedef internal::Types13 type; +}; +template +struct Types { + typedef internal::Types14 type; +}; +template +struct Types { + typedef internal::Types15 type; +}; +template +struct Types { + typedef internal::Types16 type; +}; +template +struct Types { + typedef internal::Types17 type; +}; +template +struct Types { + typedef internal::Types18 type; +}; +template +struct Types { + typedef internal::Types19 type; +}; +template +struct Types { + typedef internal::Types20 type; +}; +template +struct Types { + typedef internal::Types21 type; +}; +template +struct Types { + typedef internal::Types22 type; +}; +template +struct Types { + typedef internal::Types23 type; +}; +template +struct Types { + typedef internal::Types24 type; +}; +template +struct Types { + typedef internal::Types25 type; +}; +template +struct Types { + typedef internal::Types26 type; +}; +template +struct Types { + typedef internal::Types27 type; +}; +template +struct Types { + typedef internal::Types28 type; +}; +template +struct Types { + typedef internal::Types29 type; +}; +template +struct Types { + typedef internal::Types30 type; +}; +template +struct Types { + typedef internal::Types31 type; +}; +template +struct Types { + typedef internal::Types32 type; +}; +template +struct Types { + typedef internal::Types33 type; +}; +template +struct Types { + typedef internal::Types34 type; +}; +template +struct Types { + typedef internal::Types35 type; +}; +template +struct Types { + typedef internal::Types36 type; +}; +template +struct Types { + typedef internal::Types37 type; +}; +template +struct Types { + typedef internal::Types38 type; +}; +template +struct Types { + typedef internal::Types39 type; +}; +template +struct Types { + typedef internal::Types40 type; +}; +template +struct Types { + typedef internal::Types41 type; +}; +template +struct Types { + typedef internal::Types42 type; +}; +template +struct Types { + typedef internal::Types43 type; +}; +template +struct Types { + typedef internal::Types44 type; +}; +template +struct Types { + typedef internal::Types45 type; +}; +template +struct Types { + typedef internal::Types46 type; +}; +template +struct Types { + typedef internal::Types47 type; +}; +template +struct Types { + typedef internal::Types48 type; +}; +template +struct Types { + typedef internal::Types49 type; +}; + +namespace internal { + +# define GTEST_TEMPLATE_ template class + +// The template "selector" struct TemplateSel is used to +// represent Tmpl, which must be a class template with one type +// parameter, as a type. TemplateSel::Bind::type is defined +// as the type Tmpl. This allows us to actually instantiate the +// template "selected" by TemplateSel. +// +// This trick is necessary for simulating typedef for class templates, +// which C++ doesn't support directly. +template +struct TemplateSel { + template + struct Bind { + typedef Tmpl type; + }; +}; + +# define GTEST_BIND_(TmplSel, T) \ + TmplSel::template Bind::type + +// A unique struct template used as the default value for the +// arguments of class template Templates. This allows us to simulate +// variadic templates (e.g. Templates, Templates, +// and etc), which C++ doesn't support directly. +template +struct NoneT {}; + +// The following family of struct and struct templates are used to +// represent template lists. In particular, TemplatesN represents a list of N templates (T1, T2, ..., and TN). Except +// for Templates0, every struct in the family has two member types: +// Head for the selector of the first template in the list, and Tail +// for the rest of the list. + +// The empty template list. +struct Templates0 {}; + +// Template lists of length 1, 2, 3, and so on. + +template +struct Templates1 { + typedef TemplateSel Head; + typedef Templates0 Tail; +}; +template +struct Templates2 { + typedef TemplateSel Head; + typedef Templates1 Tail; +}; + +template +struct Templates3 { + typedef TemplateSel Head; + typedef Templates2 Tail; +}; + +template +struct Templates4 { + typedef TemplateSel Head; + typedef Templates3 Tail; +}; + +template +struct Templates5 { + typedef TemplateSel Head; + typedef Templates4 Tail; +}; + +template +struct Templates6 { + typedef TemplateSel Head; + typedef Templates5 Tail; +}; + +template +struct Templates7 { + typedef TemplateSel Head; + typedef Templates6 Tail; +}; + +template +struct Templates8 { + typedef TemplateSel Head; + typedef Templates7 Tail; +}; + +template +struct Templates9 { + typedef TemplateSel Head; + typedef Templates8 Tail; +}; + +template +struct Templates10 { + typedef TemplateSel Head; + typedef Templates9 Tail; +}; + +template +struct Templates11 { + typedef TemplateSel Head; + typedef Templates10 Tail; +}; + +template +struct Templates12 { + typedef TemplateSel Head; + typedef Templates11 Tail; +}; + +template +struct Templates13 { + typedef TemplateSel Head; + typedef Templates12 Tail; +}; + +template +struct Templates14 { + typedef TemplateSel Head; + typedef Templates13 Tail; +}; + +template +struct Templates15 { + typedef TemplateSel Head; + typedef Templates14 Tail; +}; + +template +struct Templates16 { + typedef TemplateSel Head; + typedef Templates15 Tail; +}; + +template +struct Templates17 { + typedef TemplateSel Head; + typedef Templates16 Tail; +}; + +template +struct Templates18 { + typedef TemplateSel Head; + typedef Templates17 Tail; +}; + +template +struct Templates19 { + typedef TemplateSel Head; + typedef Templates18 Tail; +}; + +template +struct Templates20 { + typedef TemplateSel Head; + typedef Templates19 Tail; +}; + +template +struct Templates21 { + typedef TemplateSel Head; + typedef Templates20 Tail; +}; + +template +struct Templates22 { + typedef TemplateSel Head; + typedef Templates21 Tail; +}; + +template +struct Templates23 { + typedef TemplateSel Head; + typedef Templates22 Tail; +}; + +template +struct Templates24 { + typedef TemplateSel Head; + typedef Templates23 Tail; +}; + +template +struct Templates25 { + typedef TemplateSel Head; + typedef Templates24 Tail; +}; + +template +struct Templates26 { + typedef TemplateSel Head; + typedef Templates25 Tail; +}; + +template +struct Templates27 { + typedef TemplateSel Head; + typedef Templates26 Tail; +}; + +template +struct Templates28 { + typedef TemplateSel Head; + typedef Templates27 Tail; +}; + +template +struct Templates29 { + typedef TemplateSel Head; + typedef Templates28 Tail; +}; + +template +struct Templates30 { + typedef TemplateSel Head; + typedef Templates29 Tail; +}; + +template +struct Templates31 { + typedef TemplateSel Head; + typedef Templates30 Tail; +}; + +template +struct Templates32 { + typedef TemplateSel Head; + typedef Templates31 Tail; +}; + +template +struct Templates33 { + typedef TemplateSel Head; + typedef Templates32 Tail; +}; + +template +struct Templates34 { + typedef TemplateSel Head; + typedef Templates33 Tail; +}; + +template +struct Templates35 { + typedef TemplateSel Head; + typedef Templates34 Tail; +}; + +template +struct Templates36 { + typedef TemplateSel Head; + typedef Templates35 Tail; +}; + +template +struct Templates37 { + typedef TemplateSel Head; + typedef Templates36 Tail; +}; + +template +struct Templates38 { + typedef TemplateSel Head; + typedef Templates37 Tail; +}; + +template +struct Templates39 { + typedef TemplateSel Head; + typedef Templates38 Tail; +}; + +template +struct Templates40 { + typedef TemplateSel Head; + typedef Templates39 Tail; +}; + +template +struct Templates41 { + typedef TemplateSel Head; + typedef Templates40 Tail; +}; + +template +struct Templates42 { + typedef TemplateSel Head; + typedef Templates41 Tail; +}; + +template +struct Templates43 { + typedef TemplateSel Head; + typedef Templates42 Tail; +}; + +template +struct Templates44 { + typedef TemplateSel Head; + typedef Templates43 Tail; +}; + +template +struct Templates45 { + typedef TemplateSel Head; + typedef Templates44 Tail; +}; + +template +struct Templates46 { + typedef TemplateSel Head; + typedef Templates45 Tail; +}; + +template +struct Templates47 { + typedef TemplateSel Head; + typedef Templates46 Tail; +}; + +template +struct Templates48 { + typedef TemplateSel Head; + typedef Templates47 Tail; +}; + +template +struct Templates49 { + typedef TemplateSel Head; + typedef Templates48 Tail; +}; + +template +struct Templates50 { + typedef TemplateSel Head; + typedef Templates49 Tail; +}; + + +// We don't want to require the users to write TemplatesN<...> directly, +// as that would require them to count the length. Templates<...> is much +// easier to write, but generates horrible messages when there is a +// compiler error, as gcc insists on printing out each template +// argument, even if it has the default value (this means Templates +// will appear as Templates in the compiler +// errors). +// +// Our solution is to combine the best part of the two approaches: a +// user would write Templates, and Google Test will translate +// that to TemplatesN internally to make error messages +// readable. The translation is done by the 'type' member of the +// Templates template. +template +struct Templates { + typedef Templates50 type; +}; + +template <> +struct Templates { + typedef Templates0 type; +}; +template +struct Templates { + typedef Templates1 type; +}; +template +struct Templates { + typedef Templates2 type; +}; +template +struct Templates { + typedef Templates3 type; +}; +template +struct Templates { + typedef Templates4 type; +}; +template +struct Templates { + typedef Templates5 type; +}; +template +struct Templates { + typedef Templates6 type; +}; +template +struct Templates { + typedef Templates7 type; +}; +template +struct Templates { + typedef Templates8 type; +}; +template +struct Templates { + typedef Templates9 type; +}; +template +struct Templates { + typedef Templates10 type; +}; +template +struct Templates { + typedef Templates11 type; +}; +template +struct Templates { + typedef Templates12 type; +}; +template +struct Templates { + typedef Templates13 type; +}; +template +struct Templates { + typedef Templates14 type; +}; +template +struct Templates { + typedef Templates15 type; +}; +template +struct Templates { + typedef Templates16 type; +}; +template +struct Templates { + typedef Templates17 type; +}; +template +struct Templates { + typedef Templates18 type; +}; +template +struct Templates { + typedef Templates19 type; +}; +template +struct Templates { + typedef Templates20 type; +}; +template +struct Templates { + typedef Templates21 type; +}; +template +struct Templates { + typedef Templates22 type; +}; +template +struct Templates { + typedef Templates23 type; +}; +template +struct Templates { + typedef Templates24 type; +}; +template +struct Templates { + typedef Templates25 type; +}; +template +struct Templates { + typedef Templates26 type; +}; +template +struct Templates { + typedef Templates27 type; +}; +template +struct Templates { + typedef Templates28 type; +}; +template +struct Templates { + typedef Templates29 type; +}; +template +struct Templates { + typedef Templates30 type; +}; +template +struct Templates { + typedef Templates31 type; +}; +template +struct Templates { + typedef Templates32 type; +}; +template +struct Templates { + typedef Templates33 type; +}; +template +struct Templates { + typedef Templates34 type; +}; +template +struct Templates { + typedef Templates35 type; +}; +template +struct Templates { + typedef Templates36 type; +}; +template +struct Templates { + typedef Templates37 type; +}; +template +struct Templates { + typedef Templates38 type; +}; +template +struct Templates { + typedef Templates39 type; +}; +template +struct Templates { + typedef Templates40 type; +}; +template +struct Templates { + typedef Templates41 type; +}; +template +struct Templates { + typedef Templates42 type; +}; +template +struct Templates { + typedef Templates43 type; +}; +template +struct Templates { + typedef Templates44 type; +}; +template +struct Templates { + typedef Templates45 type; +}; +template +struct Templates { + typedef Templates46 type; +}; +template +struct Templates { + typedef Templates47 type; +}; +template +struct Templates { + typedef Templates48 type; +}; +template +struct Templates { + typedef Templates49 type; +}; + +// The TypeList template makes it possible to use either a single type +// or a Types<...> list in TYPED_TEST_SUITE() and +// INSTANTIATE_TYPED_TEST_SUITE_P(). + +template +struct TypeList { + typedef Types1 type; +}; + +template +struct TypeList > { + typedef typename Types::type type; +}; + +#endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P + +} // namespace internal +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ diff --git a/deps/googletest/include/gtest/internal/gtest-type-util.h.pump b/deps/googletest/include/gtest/internal/gtest-type-util.h.pump new file mode 100644 index 000000000..eb014ee1b --- /dev/null +++ b/deps/googletest/include/gtest/internal/gtest-type-util.h.pump @@ -0,0 +1,314 @@ +$$ -*- mode: c++; -*- +$var n = 50 $$ Maximum length of type lists we want to support. +// Copyright 2008 Google Inc. +// All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +// Type utilities needed for implementing typed and type-parameterized +// tests. This file is generated by a SCRIPT. DO NOT EDIT BY HAND! +// +// Currently we support at most $n types in a list, and at most $n +// type-parameterized tests in one type-parameterized test suite. +// Please contact googletestframework@googlegroups.com if you need +// more. + +// GOOGLETEST_CM0001 DO NOT DELETE + +#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ +#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ + +#include "gtest/internal/gtest-port.h" + +// #ifdef __GNUC__ is too general here. It is possible to use gcc without using +// libstdc++ (which is where cxxabi.h comes from). +# if GTEST_HAS_CXXABI_H_ +# include +# elif defined(__HP_aCC) +# include +# endif // GTEST_HASH_CXXABI_H_ + +namespace testing { +namespace internal { + +// Canonicalizes a given name with respect to the Standard C++ Library. +// This handles removing the inline namespace within `std` that is +// used by various standard libraries (e.g., `std::__1`). Names outside +// of namespace std are returned unmodified. +inline std::string CanonicalizeForStdLibVersioning(std::string s) { + static const char prefix[] = "std::__"; + if (s.compare(0, strlen(prefix), prefix) == 0) { + std::string::size_type end = s.find("::", strlen(prefix)); + if (end != s.npos) { + // Erase everything between the initial `std` and the second `::`. + s.erase(strlen("std"), end - strlen("std")); + } + } + return s; +} + +// GetTypeName() returns a human-readable name of type T. +// NB: This function is also used in Google Mock, so don't move it inside of +// the typed-test-only section below. +template +std::string GetTypeName() { +# if GTEST_HAS_RTTI + + const char* const name = typeid(T).name(); +# if GTEST_HAS_CXXABI_H_ || defined(__HP_aCC) + int status = 0; + // gcc's implementation of typeid(T).name() mangles the type name, + // so we have to demangle it. +# if GTEST_HAS_CXXABI_H_ + using abi::__cxa_demangle; +# endif // GTEST_HAS_CXXABI_H_ + char* const readable_name = __cxa_demangle(name, nullptr, nullptr, &status); + const std::string name_str(status == 0 ? readable_name : name); + free(readable_name); + return CanonicalizeForStdLibVersioning(name_str); +# else + return name; +# endif // GTEST_HAS_CXXABI_H_ || __HP_aCC + +# else + + return ""; + +# endif // GTEST_HAS_RTTI +} + +#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P + +// AssertyTypeEq::type is defined iff T1 and T2 are the same +// type. This can be used as a compile-time assertion to ensure that +// two types are equal. + +template +struct AssertTypeEq; + +template +struct AssertTypeEq { + typedef bool type; +}; + +// A unique type used as the default value for the arguments of class +// template Types. This allows us to simulate variadic templates +// (e.g. Types, Type, and etc), which C++ doesn't +// support directly. +struct None {}; + +// The following family of struct and struct templates are used to +// represent type lists. In particular, TypesN +// represents a type list with N types (T1, T2, ..., and TN) in it. +// Except for Types0, every struct in the family has two member types: +// Head for the first type in the list, and Tail for the rest of the +// list. + +// The empty type list. +struct Types0 {}; + +// Type lists of length 1, 2, 3, and so on. + +template +struct Types1 { + typedef T1 Head; + typedef Types0 Tail; +}; + +$range i 2..n + +$for i [[ +$range j 1..i +$range k 2..i +template <$for j, [[typename T$j]]> +struct Types$i { + typedef T1 Head; + typedef Types$(i-1)<$for k, [[T$k]]> Tail; +}; + + +]] + +} // namespace internal + +// We don't want to require the users to write TypesN<...> directly, +// as that would require them to count the length. Types<...> is much +// easier to write, but generates horrible messages when there is a +// compiler error, as gcc insists on printing out each template +// argument, even if it has the default value (this means Types +// will appear as Types in the compiler +// errors). +// +// Our solution is to combine the best part of the two approaches: a +// user would write Types, and Google Test will translate +// that to TypesN internally to make error messages +// readable. The translation is done by the 'type' member of the +// Types template. + +$range i 1..n +template <$for i, [[typename T$i = internal::None]]> +struct Types { + typedef internal::Types$n<$for i, [[T$i]]> type; +}; + +template <> +struct Types<$for i, [[internal::None]]> { + typedef internal::Types0 type; +}; + +$range i 1..n-1 +$for i [[ +$range j 1..i +$range k i+1..n +template <$for j, [[typename T$j]]> +struct Types<$for j, [[T$j]]$for k[[, internal::None]]> { + typedef internal::Types$i<$for j, [[T$j]]> type; +}; + +]] + +namespace internal { + +# define GTEST_TEMPLATE_ template class + +// The template "selector" struct TemplateSel is used to +// represent Tmpl, which must be a class template with one type +// parameter, as a type. TemplateSel::Bind::type is defined +// as the type Tmpl. This allows us to actually instantiate the +// template "selected" by TemplateSel. +// +// This trick is necessary for simulating typedef for class templates, +// which C++ doesn't support directly. +template +struct TemplateSel { + template + struct Bind { + typedef Tmpl type; + }; +}; + +# define GTEST_BIND_(TmplSel, T) \ + TmplSel::template Bind::type + +// A unique struct template used as the default value for the +// arguments of class template Templates. This allows us to simulate +// variadic templates (e.g. Templates, Templates, +// and etc), which C++ doesn't support directly. +template +struct NoneT {}; + +// The following family of struct and struct templates are used to +// represent template lists. In particular, TemplatesN represents a list of N templates (T1, T2, ..., and TN). Except +// for Templates0, every struct in the family has two member types: +// Head for the selector of the first template in the list, and Tail +// for the rest of the list. + +// The empty template list. +struct Templates0 {}; + +// Template lists of length 1, 2, 3, and so on. + +template +struct Templates1 { + typedef TemplateSel Head; + typedef Templates0 Tail; +}; + +$range i 2..n + +$for i [[ +$range j 1..i +$range k 2..i +template <$for j, [[GTEST_TEMPLATE_ T$j]]> +struct Templates$i { + typedef TemplateSel Head; + typedef Templates$(i-1)<$for k, [[T$k]]> Tail; +}; + + +]] + +// We don't want to require the users to write TemplatesN<...> directly, +// as that would require them to count the length. Templates<...> is much +// easier to write, but generates horrible messages when there is a +// compiler error, as gcc insists on printing out each template +// argument, even if it has the default value (this means Templates +// will appear as Templates in the compiler +// errors). +// +// Our solution is to combine the best part of the two approaches: a +// user would write Templates, and Google Test will translate +// that to TemplatesN internally to make error messages +// readable. The translation is done by the 'type' member of the +// Templates template. + +$range i 1..n +template <$for i, [[GTEST_TEMPLATE_ T$i = NoneT]]> +struct Templates { + typedef Templates$n<$for i, [[T$i]]> type; +}; + +template <> +struct Templates<$for i, [[NoneT]]> { + typedef Templates0 type; +}; + +$range i 1..n-1 +$for i [[ +$range j 1..i +$range k i+1..n +template <$for j, [[GTEST_TEMPLATE_ T$j]]> +struct Templates<$for j, [[T$j]]$for k[[, NoneT]]> { + typedef Templates$i<$for j, [[T$j]]> type; +}; + +]] + +// The TypeList template makes it possible to use either a single type +// or a Types<...> list in TYPED_TEST_SUITE() and +// INSTANTIATE_TYPED_TEST_SUITE_P(). + +template +struct TypeList { + typedef Types1 type; +}; + + +$range i 1..n +template <$for i, [[typename T$i]]> +struct TypeList > { + typedef typename Types<$for i, [[T$i]]>::type type; +}; + +#endif // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P + +} // namespace internal +} // namespace testing + +#endif // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_ diff --git a/deps/googletest/src/gtest-all.cc b/deps/googletest/src/gtest-all.cc new file mode 100644 index 000000000..ad292905c --- /dev/null +++ b/deps/googletest/src/gtest-all.cc @@ -0,0 +1,48 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// Google C++ Testing and Mocking Framework (Google Test) +// +// Sometimes it's desirable to build Google Test by compiling a single file. +// This file serves this purpose. + +// This line ensures that gtest.h can be compiled on its own, even +// when it's fused. +#include "gtest/gtest.h" + +// The following lines pull in the real gtest *.cc files. +#include "src/gtest.cc" +#include "src/gtest-death-test.cc" +#include "src/gtest-filepath.cc" +#include "src/gtest-matchers.cc" +#include "src/gtest-port.cc" +#include "src/gtest-printers.cc" +#include "src/gtest-test-part.cc" +#include "src/gtest-typed-test.cc" diff --git a/deps/googletest/src/gtest-death-test.cc b/deps/googletest/src/gtest-death-test.cc new file mode 100644 index 000000000..9bfe31534 --- /dev/null +++ b/deps/googletest/src/gtest-death-test.cc @@ -0,0 +1,1643 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// This file implements death tests. + +#include "gtest/gtest-death-test.h" + +#include + +#include "gtest/internal/gtest-port.h" +#include "gtest/internal/custom/gtest.h" + +#if GTEST_HAS_DEATH_TEST + +# if GTEST_OS_MAC +# include +# endif // GTEST_OS_MAC + +# include +# include +# include + +# if GTEST_OS_LINUX +# include +# endif // GTEST_OS_LINUX + +# include + +# if GTEST_OS_WINDOWS +# include +# else +# include +# include +# endif // GTEST_OS_WINDOWS + +# if GTEST_OS_QNX +# include +# endif // GTEST_OS_QNX + +# if GTEST_OS_FUCHSIA +# include +# include +# include +# include +# include +# include +# include +# include +# include +# include +# endif // GTEST_OS_FUCHSIA + +#endif // GTEST_HAS_DEATH_TEST + +#include "gtest/gtest-message.h" +#include "gtest/internal/gtest-string.h" +#include "src/gtest-internal-inl.h" + +namespace testing { + +// Constants. + +// The default death test style. +// +// This is defined in internal/gtest-port.h as "fast", but can be overridden by +// a definition in internal/custom/gtest-port.h. The recommended value, which is +// used internally at Google, is "threadsafe". +static const char kDefaultDeathTestStyle[] = GTEST_DEFAULT_DEATH_TEST_STYLE; + +GTEST_DEFINE_string_( + death_test_style, + internal::StringFromGTestEnv("death_test_style", kDefaultDeathTestStyle), + "Indicates how to run a death test in a forked child process: " + "\"threadsafe\" (child process re-executes the test binary " + "from the beginning, running only the specific death test) or " + "\"fast\" (child process runs the death test immediately " + "after forking)."); + +GTEST_DEFINE_bool_( + death_test_use_fork, + internal::BoolFromGTestEnv("death_test_use_fork", false), + "Instructs to use fork()/_exit() instead of clone() in death tests. " + "Ignored and always uses fork() on POSIX systems where clone() is not " + "implemented. Useful when running under valgrind or similar tools if " + "those do not support clone(). Valgrind 3.3.1 will just fail if " + "it sees an unsupported combination of clone() flags. " + "It is not recommended to use this flag w/o valgrind though it will " + "work in 99% of the cases. Once valgrind is fixed, this flag will " + "most likely be removed."); + +namespace internal { +GTEST_DEFINE_string_( + internal_run_death_test, "", + "Indicates the file, line number, temporal index of " + "the single death test to run, and a file descriptor to " + "which a success code may be sent, all separated by " + "the '|' characters. This flag is specified if and only if the current " + "process is a sub-process launched for running a thread-safe " + "death test. FOR INTERNAL USE ONLY."); +} // namespace internal + +#if GTEST_HAS_DEATH_TEST + +namespace internal { + +// Valid only for fast death tests. Indicates the code is running in the +// child process of a fast style death test. +# if !GTEST_OS_WINDOWS && !GTEST_OS_FUCHSIA +static bool g_in_fast_death_test_child = false; +# endif + +// Returns a Boolean value indicating whether the caller is currently +// executing in the context of the death test child process. Tools such as +// Valgrind heap checkers may need this to modify their behavior in death +// tests. IMPORTANT: This is an internal utility. Using it may break the +// implementation of death tests. User code MUST NOT use it. +bool InDeathTestChild() { +# if GTEST_OS_WINDOWS || GTEST_OS_FUCHSIA + + // On Windows and Fuchsia, death tests are thread-safe regardless of the value + // of the death_test_style flag. + return !GTEST_FLAG(internal_run_death_test).empty(); + +# else + + if (GTEST_FLAG(death_test_style) == "threadsafe") + return !GTEST_FLAG(internal_run_death_test).empty(); + else + return g_in_fast_death_test_child; +#endif +} + +} // namespace internal + +// ExitedWithCode constructor. +ExitedWithCode::ExitedWithCode(int exit_code) : exit_code_(exit_code) { +} + +// ExitedWithCode function-call operator. +bool ExitedWithCode::operator()(int exit_status) const { +# if GTEST_OS_WINDOWS || GTEST_OS_FUCHSIA + + return exit_status == exit_code_; + +# else + + return WIFEXITED(exit_status) && WEXITSTATUS(exit_status) == exit_code_; + +# endif // GTEST_OS_WINDOWS || GTEST_OS_FUCHSIA +} + +# if !GTEST_OS_WINDOWS && !GTEST_OS_FUCHSIA +// KilledBySignal constructor. +KilledBySignal::KilledBySignal(int signum) : signum_(signum) { +} + +// KilledBySignal function-call operator. +bool KilledBySignal::operator()(int exit_status) const { +# if defined(GTEST_KILLED_BY_SIGNAL_OVERRIDE_) + { + bool result; + if (GTEST_KILLED_BY_SIGNAL_OVERRIDE_(signum_, exit_status, &result)) { + return result; + } + } +# endif // defined(GTEST_KILLED_BY_SIGNAL_OVERRIDE_) + return WIFSIGNALED(exit_status) && WTERMSIG(exit_status) == signum_; +} +# endif // !GTEST_OS_WINDOWS && !GTEST_OS_FUCHSIA + +namespace internal { + +// Utilities needed for death tests. + +// Generates a textual description of a given exit code, in the format +// specified by wait(2). +static std::string ExitSummary(int exit_code) { + Message m; + +# if GTEST_OS_WINDOWS || GTEST_OS_FUCHSIA + + m << "Exited with exit status " << exit_code; + +# else + + if (WIFEXITED(exit_code)) { + m << "Exited with exit status " << WEXITSTATUS(exit_code); + } else if (WIFSIGNALED(exit_code)) { + m << "Terminated by signal " << WTERMSIG(exit_code); + } +# ifdef WCOREDUMP + if (WCOREDUMP(exit_code)) { + m << " (core dumped)"; + } +# endif +# endif // GTEST_OS_WINDOWS || GTEST_OS_FUCHSIA + + return m.GetString(); +} + +// Returns true if exit_status describes a process that was terminated +// by a signal, or exited normally with a nonzero exit code. +bool ExitedUnsuccessfully(int exit_status) { + return !ExitedWithCode(0)(exit_status); +} + +# if !GTEST_OS_WINDOWS && !GTEST_OS_FUCHSIA +// Generates a textual failure message when a death test finds more than +// one thread running, or cannot determine the number of threads, prior +// to executing the given statement. It is the responsibility of the +// caller not to pass a thread_count of 1. +static std::string DeathTestThreadWarning(size_t thread_count) { + Message msg; + msg << "Death tests use fork(), which is unsafe particularly" + << " in a threaded context. For this test, " << GTEST_NAME_ << " "; + if (thread_count == 0) { + msg << "couldn't detect the number of threads."; + } else { + msg << "detected " << thread_count << " threads."; + } + msg << " See " + "https://github.com/google/googletest/blob/master/googletest/docs/" + "advanced.md#death-tests-and-threads" + << " for more explanation and suggested solutions, especially if" + << " this is the last message you see before your test times out."; + return msg.GetString(); +} +# endif // !GTEST_OS_WINDOWS && !GTEST_OS_FUCHSIA + +// Flag characters for reporting a death test that did not die. +static const char kDeathTestLived = 'L'; +static const char kDeathTestReturned = 'R'; +static const char kDeathTestThrew = 'T'; +static const char kDeathTestInternalError = 'I'; + +#if GTEST_OS_FUCHSIA + +// File descriptor used for the pipe in the child process. +static const int kFuchsiaReadPipeFd = 3; + +#endif + +// An enumeration describing all of the possible ways that a death test can +// conclude. DIED means that the process died while executing the test +// code; LIVED means that process lived beyond the end of the test code; +// RETURNED means that the test statement attempted to execute a return +// statement, which is not allowed; THREW means that the test statement +// returned control by throwing an exception. IN_PROGRESS means the test +// has not yet concluded. +enum DeathTestOutcome { IN_PROGRESS, DIED, LIVED, RETURNED, THREW }; + +// Routine for aborting the program which is safe to call from an +// exec-style death test child process, in which case the error +// message is propagated back to the parent process. Otherwise, the +// message is simply printed to stderr. In either case, the program +// then exits with status 1. +static void DeathTestAbort(const std::string& message) { + // On a POSIX system, this function may be called from a threadsafe-style + // death test child process, which operates on a very small stack. Use + // the heap for any additional non-minuscule memory requirements. + const InternalRunDeathTestFlag* const flag = + GetUnitTestImpl()->internal_run_death_test_flag(); + if (flag != nullptr) { + FILE* parent = posix::FDOpen(flag->write_fd(), "w"); + fputc(kDeathTestInternalError, parent); + fprintf(parent, "%s", message.c_str()); + fflush(parent); + _exit(1); + } else { + fprintf(stderr, "%s", message.c_str()); + fflush(stderr); + posix::Abort(); + } +} + +// A replacement for CHECK that calls DeathTestAbort if the assertion +// fails. +# define GTEST_DEATH_TEST_CHECK_(expression) \ + do { \ + if (!::testing::internal::IsTrue(expression)) { \ + DeathTestAbort( \ + ::std::string("CHECK failed: File ") + __FILE__ + ", line " \ + + ::testing::internal::StreamableToString(__LINE__) + ": " \ + + #expression); \ + } \ + } while (::testing::internal::AlwaysFalse()) + +// This macro is similar to GTEST_DEATH_TEST_CHECK_, but it is meant for +// evaluating any system call that fulfills two conditions: it must return +// -1 on failure, and set errno to EINTR when it is interrupted and +// should be tried again. The macro expands to a loop that repeatedly +// evaluates the expression as long as it evaluates to -1 and sets +// errno to EINTR. If the expression evaluates to -1 but errno is +// something other than EINTR, DeathTestAbort is called. +# define GTEST_DEATH_TEST_CHECK_SYSCALL_(expression) \ + do { \ + int gtest_retval; \ + do { \ + gtest_retval = (expression); \ + } while (gtest_retval == -1 && errno == EINTR); \ + if (gtest_retval == -1) { \ + DeathTestAbort( \ + ::std::string("CHECK failed: File ") + __FILE__ + ", line " \ + + ::testing::internal::StreamableToString(__LINE__) + ": " \ + + #expression + " != -1"); \ + } \ + } while (::testing::internal::AlwaysFalse()) + +// Returns the message describing the last system error in errno. +std::string GetLastErrnoDescription() { + return errno == 0 ? "" : posix::StrError(errno); +} + +// This is called from a death test parent process to read a failure +// message from the death test child process and log it with the FATAL +// severity. On Windows, the message is read from a pipe handle. On other +// platforms, it is read from a file descriptor. +static void FailFromInternalError(int fd) { + Message error; + char buffer[256]; + int num_read; + + do { + while ((num_read = posix::Read(fd, buffer, 255)) > 0) { + buffer[num_read] = '\0'; + error << buffer; + } + } while (num_read == -1 && errno == EINTR); + + if (num_read == 0) { + GTEST_LOG_(FATAL) << error.GetString(); + } else { + const int last_error = errno; + GTEST_LOG_(FATAL) << "Error while reading death test internal: " + << GetLastErrnoDescription() << " [" << last_error << "]"; + } +} + +// Death test constructor. Increments the running death test count +// for the current test. +DeathTest::DeathTest() { + TestInfo* const info = GetUnitTestImpl()->current_test_info(); + if (info == nullptr) { + DeathTestAbort("Cannot run a death test outside of a TEST or " + "TEST_F construct"); + } +} + +// Creates and returns a death test by dispatching to the current +// death test factory. +bool DeathTest::Create(const char* statement, + Matcher matcher, const char* file, + int line, DeathTest** test) { + return GetUnitTestImpl()->death_test_factory()->Create( + statement, std::move(matcher), file, line, test); +} + +const char* DeathTest::LastMessage() { + return last_death_test_message_.c_str(); +} + +void DeathTest::set_last_death_test_message(const std::string& message) { + last_death_test_message_ = message; +} + +std::string DeathTest::last_death_test_message_; + +// Provides cross platform implementation for some death functionality. +class DeathTestImpl : public DeathTest { + protected: + DeathTestImpl(const char* a_statement, Matcher matcher) + : statement_(a_statement), + matcher_(std::move(matcher)), + spawned_(false), + status_(-1), + outcome_(IN_PROGRESS), + read_fd_(-1), + write_fd_(-1) {} + + // read_fd_ is expected to be closed and cleared by a derived class. + ~DeathTestImpl() override { GTEST_DEATH_TEST_CHECK_(read_fd_ == -1); } + + void Abort(AbortReason reason) override; + bool Passed(bool status_ok) override; + + const char* statement() const { return statement_; } + bool spawned() const { return spawned_; } + void set_spawned(bool is_spawned) { spawned_ = is_spawned; } + int status() const { return status_; } + void set_status(int a_status) { status_ = a_status; } + DeathTestOutcome outcome() const { return outcome_; } + void set_outcome(DeathTestOutcome an_outcome) { outcome_ = an_outcome; } + int read_fd() const { return read_fd_; } + void set_read_fd(int fd) { read_fd_ = fd; } + int write_fd() const { return write_fd_; } + void set_write_fd(int fd) { write_fd_ = fd; } + + // Called in the parent process only. Reads the result code of the death + // test child process via a pipe, interprets it to set the outcome_ + // member, and closes read_fd_. Outputs diagnostics and terminates in + // case of unexpected codes. + void ReadAndInterpretStatusByte(); + + // Returns stderr output from the child process. + virtual std::string GetErrorLogs(); + + private: + // The textual content of the code this object is testing. This class + // doesn't own this string and should not attempt to delete it. + const char* const statement_; + // A matcher that's expected to match the stderr output by the child process. + Matcher matcher_; + // True if the death test child process has been successfully spawned. + bool spawned_; + // The exit status of the child process. + int status_; + // How the death test concluded. + DeathTestOutcome outcome_; + // Descriptor to the read end of the pipe to the child process. It is + // always -1 in the child process. The child keeps its write end of the + // pipe in write_fd_. + int read_fd_; + // Descriptor to the child's write end of the pipe to the parent process. + // It is always -1 in the parent process. The parent keeps its end of the + // pipe in read_fd_. + int write_fd_; +}; + +// Called in the parent process only. Reads the result code of the death +// test child process via a pipe, interprets it to set the outcome_ +// member, and closes read_fd_. Outputs diagnostics and terminates in +// case of unexpected codes. +void DeathTestImpl::ReadAndInterpretStatusByte() { + char flag; + int bytes_read; + + // The read() here blocks until data is available (signifying the + // failure of the death test) or until the pipe is closed (signifying + // its success), so it's okay to call this in the parent before + // the child process has exited. + do { + bytes_read = posix::Read(read_fd(), &flag, 1); + } while (bytes_read == -1 && errno == EINTR); + + if (bytes_read == 0) { + set_outcome(DIED); + } else if (bytes_read == 1) { + switch (flag) { + case kDeathTestReturned: + set_outcome(RETURNED); + break; + case kDeathTestThrew: + set_outcome(THREW); + break; + case kDeathTestLived: + set_outcome(LIVED); + break; + case kDeathTestInternalError: + FailFromInternalError(read_fd()); // Does not return. + break; + default: + GTEST_LOG_(FATAL) << "Death test child process reported " + << "unexpected status byte (" + << static_cast(flag) << ")"; + } + } else { + GTEST_LOG_(FATAL) << "Read from death test child process failed: " + << GetLastErrnoDescription(); + } + GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Close(read_fd())); + set_read_fd(-1); +} + +std::string DeathTestImpl::GetErrorLogs() { + return GetCapturedStderr(); +} + +// Signals that the death test code which should have exited, didn't. +// Should be called only in a death test child process. +// Writes a status byte to the child's status file descriptor, then +// calls _exit(1). +void DeathTestImpl::Abort(AbortReason reason) { + // The parent process considers the death test to be a failure if + // it finds any data in our pipe. So, here we write a single flag byte + // to the pipe, then exit. + const char status_ch = + reason == TEST_DID_NOT_DIE ? kDeathTestLived : + reason == TEST_THREW_EXCEPTION ? kDeathTestThrew : kDeathTestReturned; + + GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Write(write_fd(), &status_ch, 1)); + // We are leaking the descriptor here because on some platforms (i.e., + // when built as Windows DLL), destructors of global objects will still + // run after calling _exit(). On such systems, write_fd_ will be + // indirectly closed from the destructor of UnitTestImpl, causing double + // close if it is also closed here. On debug configurations, double close + // may assert. As there are no in-process buffers to flush here, we are + // relying on the OS to close the descriptor after the process terminates + // when the destructors are not run. + _exit(1); // Exits w/o any normal exit hooks (we were supposed to crash) +} + +// Returns an indented copy of stderr output for a death test. +// This makes distinguishing death test output lines from regular log lines +// much easier. +static ::std::string FormatDeathTestOutput(const ::std::string& output) { + ::std::string ret; + for (size_t at = 0; ; ) { + const size_t line_end = output.find('\n', at); + ret += "[ DEATH ] "; + if (line_end == ::std::string::npos) { + ret += output.substr(at); + break; + } + ret += output.substr(at, line_end + 1 - at); + at = line_end + 1; + } + return ret; +} + +// Assesses the success or failure of a death test, using both private +// members which have previously been set, and one argument: +// +// Private data members: +// outcome: An enumeration describing how the death test +// concluded: DIED, LIVED, THREW, or RETURNED. The death test +// fails in the latter three cases. +// status: The exit status of the child process. On *nix, it is in the +// in the format specified by wait(2). On Windows, this is the +// value supplied to the ExitProcess() API or a numeric code +// of the exception that terminated the program. +// matcher_: A matcher that's expected to match the stderr output by the child +// process. +// +// Argument: +// status_ok: true if exit_status is acceptable in the context of +// this particular death test, which fails if it is false +// +// Returns true iff all of the above conditions are met. Otherwise, the +// first failing condition, in the order given above, is the one that is +// reported. Also sets the last death test message string. +bool DeathTestImpl::Passed(bool status_ok) { + if (!spawned()) + return false; + + const std::string error_message = GetErrorLogs(); + + bool success = false; + Message buffer; + + buffer << "Death test: " << statement() << "\n"; + switch (outcome()) { + case LIVED: + buffer << " Result: failed to die.\n" + << " Error msg:\n" << FormatDeathTestOutput(error_message); + break; + case THREW: + buffer << " Result: threw an exception.\n" + << " Error msg:\n" << FormatDeathTestOutput(error_message); + break; + case RETURNED: + buffer << " Result: illegal return in test statement.\n" + << " Error msg:\n" << FormatDeathTestOutput(error_message); + break; + case DIED: + if (status_ok) { + if (matcher_.Matches(error_message)) { + success = true; + } else { + std::ostringstream stream; + matcher_.DescribeTo(&stream); + buffer << " Result: died but not with expected error.\n" + << " Expected: " << stream.str() << "\n" + << "Actual msg:\n" + << FormatDeathTestOutput(error_message); + } + } else { + buffer << " Result: died but not with expected exit code:\n" + << " " << ExitSummary(status()) << "\n" + << "Actual msg:\n" << FormatDeathTestOutput(error_message); + } + break; + case IN_PROGRESS: + default: + GTEST_LOG_(FATAL) + << "DeathTest::Passed somehow called before conclusion of test"; + } + + DeathTest::set_last_death_test_message(buffer.GetString()); + return success; +} + +# if GTEST_OS_WINDOWS +// WindowsDeathTest implements death tests on Windows. Due to the +// specifics of starting new processes on Windows, death tests there are +// always threadsafe, and Google Test considers the +// --gtest_death_test_style=fast setting to be equivalent to +// --gtest_death_test_style=threadsafe there. +// +// A few implementation notes: Like the Linux version, the Windows +// implementation uses pipes for child-to-parent communication. But due to +// the specifics of pipes on Windows, some extra steps are required: +// +// 1. The parent creates a communication pipe and stores handles to both +// ends of it. +// 2. The parent starts the child and provides it with the information +// necessary to acquire the handle to the write end of the pipe. +// 3. The child acquires the write end of the pipe and signals the parent +// using a Windows event. +// 4. Now the parent can release the write end of the pipe on its side. If +// this is done before step 3, the object's reference count goes down to +// 0 and it is destroyed, preventing the child from acquiring it. The +// parent now has to release it, or read operations on the read end of +// the pipe will not return when the child terminates. +// 5. The parent reads child's output through the pipe (outcome code and +// any possible error messages) from the pipe, and its stderr and then +// determines whether to fail the test. +// +// Note: to distinguish Win32 API calls from the local method and function +// calls, the former are explicitly resolved in the global namespace. +// +class WindowsDeathTest : public DeathTestImpl { + public: + WindowsDeathTest(const char* a_statement, Matcher matcher, + const char* file, int line) + : DeathTestImpl(a_statement, std::move(matcher)), + file_(file), + line_(line) {} + + // All of these virtual functions are inherited from DeathTest. + virtual int Wait(); + virtual TestRole AssumeRole(); + + private: + // The name of the file in which the death test is located. + const char* const file_; + // The line number on which the death test is located. + const int line_; + // Handle to the write end of the pipe to the child process. + AutoHandle write_handle_; + // Child process handle. + AutoHandle child_handle_; + // Event the child process uses to signal the parent that it has + // acquired the handle to the write end of the pipe. After seeing this + // event the parent can release its own handles to make sure its + // ReadFile() calls return when the child terminates. + AutoHandle event_handle_; +}; + +// Waits for the child in a death test to exit, returning its exit +// status, or 0 if no child process exists. As a side effect, sets the +// outcome data member. +int WindowsDeathTest::Wait() { + if (!spawned()) + return 0; + + // Wait until the child either signals that it has acquired the write end + // of the pipe or it dies. + const HANDLE wait_handles[2] = { child_handle_.Get(), event_handle_.Get() }; + switch (::WaitForMultipleObjects(2, + wait_handles, + FALSE, // Waits for any of the handles. + INFINITE)) { + case WAIT_OBJECT_0: + case WAIT_OBJECT_0 + 1: + break; + default: + GTEST_DEATH_TEST_CHECK_(false); // Should not get here. + } + + // The child has acquired the write end of the pipe or exited. + // We release the handle on our side and continue. + write_handle_.Reset(); + event_handle_.Reset(); + + ReadAndInterpretStatusByte(); + + // Waits for the child process to exit if it haven't already. This + // returns immediately if the child has already exited, regardless of + // whether previous calls to WaitForMultipleObjects synchronized on this + // handle or not. + GTEST_DEATH_TEST_CHECK_( + WAIT_OBJECT_0 == ::WaitForSingleObject(child_handle_.Get(), + INFINITE)); + DWORD status_code; + GTEST_DEATH_TEST_CHECK_( + ::GetExitCodeProcess(child_handle_.Get(), &status_code) != FALSE); + child_handle_.Reset(); + set_status(static_cast(status_code)); + return status(); +} + +// The AssumeRole process for a Windows death test. It creates a child +// process with the same executable as the current process to run the +// death test. The child process is given the --gtest_filter and +// --gtest_internal_run_death_test flags such that it knows to run the +// current death test only. +DeathTest::TestRole WindowsDeathTest::AssumeRole() { + const UnitTestImpl* const impl = GetUnitTestImpl(); + const InternalRunDeathTestFlag* const flag = + impl->internal_run_death_test_flag(); + const TestInfo* const info = impl->current_test_info(); + const int death_test_index = info->result()->death_test_count(); + + if (flag != nullptr) { + // ParseInternalRunDeathTestFlag() has performed all the necessary + // processing. + set_write_fd(flag->write_fd()); + return EXECUTE_TEST; + } + + // WindowsDeathTest uses an anonymous pipe to communicate results of + // a death test. + SECURITY_ATTRIBUTES handles_are_inheritable = {sizeof(SECURITY_ATTRIBUTES), + nullptr, TRUE}; + HANDLE read_handle, write_handle; + GTEST_DEATH_TEST_CHECK_( + ::CreatePipe(&read_handle, &write_handle, &handles_are_inheritable, + 0) // Default buffer size. + != FALSE); + set_read_fd(::_open_osfhandle(reinterpret_cast(read_handle), + O_RDONLY)); + write_handle_.Reset(write_handle); + event_handle_.Reset(::CreateEvent( + &handles_are_inheritable, + TRUE, // The event will automatically reset to non-signaled state. + FALSE, // The initial state is non-signalled. + nullptr)); // The even is unnamed. + GTEST_DEATH_TEST_CHECK_(event_handle_.Get() != nullptr); + const std::string filter_flag = std::string("--") + GTEST_FLAG_PREFIX_ + + kFilterFlag + "=" + info->test_suite_name() + + "." + info->name(); + const std::string internal_flag = + std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + + "=" + file_ + "|" + StreamableToString(line_) + "|" + + StreamableToString(death_test_index) + "|" + + StreamableToString(static_cast(::GetCurrentProcessId())) + + // size_t has the same width as pointers on both 32-bit and 64-bit + // Windows platforms. + // See http://msdn.microsoft.com/en-us/library/tcxf1dw6.aspx. + "|" + StreamableToString(reinterpret_cast(write_handle)) + + "|" + StreamableToString(reinterpret_cast(event_handle_.Get())); + + char executable_path[_MAX_PATH + 1]; // NOLINT + GTEST_DEATH_TEST_CHECK_(_MAX_PATH + 1 != ::GetModuleFileNameA(nullptr, + executable_path, + _MAX_PATH)); + + std::string command_line = + std::string(::GetCommandLineA()) + " " + filter_flag + " \"" + + internal_flag + "\""; + + DeathTest::set_last_death_test_message(""); + + CaptureStderr(); + // Flush the log buffers since the log streams are shared with the child. + FlushInfoLog(); + + // The child process will share the standard handles with the parent. + STARTUPINFOA startup_info; + memset(&startup_info, 0, sizeof(STARTUPINFO)); + startup_info.dwFlags = STARTF_USESTDHANDLES; + startup_info.hStdInput = ::GetStdHandle(STD_INPUT_HANDLE); + startup_info.hStdOutput = ::GetStdHandle(STD_OUTPUT_HANDLE); + startup_info.hStdError = ::GetStdHandle(STD_ERROR_HANDLE); + + PROCESS_INFORMATION process_info; + GTEST_DEATH_TEST_CHECK_( + ::CreateProcessA( + executable_path, const_cast(command_line.c_str()), + nullptr, // Retuned process handle is not inheritable. + nullptr, // Retuned thread handle is not inheritable. + TRUE, // Child inherits all inheritable handles (for write_handle_). + 0x0, // Default creation flags. + nullptr, // Inherit the parent's environment. + UnitTest::GetInstance()->original_working_dir(), &startup_info, + &process_info) != FALSE); + child_handle_.Reset(process_info.hProcess); + ::CloseHandle(process_info.hThread); + set_spawned(true); + return OVERSEE_TEST; +} + +# elif GTEST_OS_FUCHSIA + +class FuchsiaDeathTest : public DeathTestImpl { + public: + FuchsiaDeathTest(const char* a_statement, Matcher matcher, + const char* file, int line) + : DeathTestImpl(a_statement, std::move(matcher)), + file_(file), + line_(line) {} + + // All of these virtual functions are inherited from DeathTest. + int Wait() override; + TestRole AssumeRole() override; + std::string GetErrorLogs() override; + + private: + // The name of the file in which the death test is located. + const char* const file_; + // The line number on which the death test is located. + const int line_; + // The stderr data captured by the child process. + std::string captured_stderr_; + + zx::process child_process_; + zx::port port_; + zx::socket stderr_socket_; +}; + +// Utility class for accumulating command-line arguments. +class Arguments { + public: + Arguments() { args_.push_back(nullptr); } + + ~Arguments() { + for (std::vector::iterator i = args_.begin(); i != args_.end(); + ++i) { + free(*i); + } + } + void AddArgument(const char* argument) { + args_.insert(args_.end() - 1, posix::StrDup(argument)); + } + + template + void AddArguments(const ::std::vector& arguments) { + for (typename ::std::vector::const_iterator i = arguments.begin(); + i != arguments.end(); + ++i) { + args_.insert(args_.end() - 1, posix::StrDup(i->c_str())); + } + } + char* const* Argv() { + return &args_[0]; + } + + int size() { + return args_.size() - 1; + } + + private: + std::vector args_; +}; + +// Waits for the child in a death test to exit, returning its exit +// status, or 0 if no child process exists. As a side effect, sets the +// outcome data member. +int FuchsiaDeathTest::Wait() { + const int kProcessKey = 0; + const int kSocketKey = 1; + + if (!spawned()) + return 0; + + // Register to wait for the child process to terminate. + zx_status_t status_zx; + status_zx = child_process_.wait_async( + port_, kProcessKey, ZX_PROCESS_TERMINATED, ZX_WAIT_ASYNC_ONCE); + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_OK); + // Register to wait for the socket to be readable or closed. + status_zx = stderr_socket_.wait_async( + port_, kSocketKey, ZX_SOCKET_READABLE | ZX_SOCKET_PEER_CLOSED, + ZX_WAIT_ASYNC_ONCE); + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_OK); + + bool process_terminated = false; + bool socket_closed = false; + do { + zx_port_packet_t packet = {}; + status_zx = port_.wait(zx::time::infinite(), &packet); + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_OK); + + if (packet.key == kProcessKey) { + if (ZX_PKT_IS_EXCEPTION(packet.type)) { + // Process encountered an exception. Kill it directly rather than + // letting other handlers process the event. We will get a second + // kProcessKey event when the process actually terminates. + status_zx = child_process_.kill(); + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_OK); + } else { + // Process terminated. + GTEST_DEATH_TEST_CHECK_(ZX_PKT_IS_SIGNAL_ONE(packet.type)); + GTEST_DEATH_TEST_CHECK_(packet.signal.observed & ZX_PROCESS_TERMINATED); + process_terminated = true; + } + } else if (packet.key == kSocketKey) { + GTEST_DEATH_TEST_CHECK_(ZX_PKT_IS_SIGNAL_ONE(packet.type)); + if (packet.signal.observed & ZX_SOCKET_READABLE) { + // Read data from the socket. + constexpr size_t kBufferSize = 1024; + do { + size_t old_length = captured_stderr_.length(); + size_t bytes_read = 0; + captured_stderr_.resize(old_length + kBufferSize); + status_zx = stderr_socket_.read( + 0, &captured_stderr_.front() + old_length, kBufferSize, + &bytes_read); + captured_stderr_.resize(old_length + bytes_read); + } while (status_zx == ZX_OK); + if (status_zx == ZX_ERR_PEER_CLOSED) { + socket_closed = true; + } else { + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_ERR_SHOULD_WAIT); + status_zx = stderr_socket_.wait_async( + port_, kSocketKey, ZX_SOCKET_READABLE | ZX_SOCKET_PEER_CLOSED, + ZX_WAIT_ASYNC_ONCE); + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_OK); + } + } else { + GTEST_DEATH_TEST_CHECK_(packet.signal.observed & ZX_SOCKET_PEER_CLOSED); + socket_closed = true; + } + } + } while (!process_terminated && !socket_closed); + + ReadAndInterpretStatusByte(); + + zx_info_process_t buffer; + status_zx = child_process_.get_info( + ZX_INFO_PROCESS, &buffer, sizeof(buffer), nullptr, nullptr); + GTEST_DEATH_TEST_CHECK_(status_zx == ZX_OK); + + GTEST_DEATH_TEST_CHECK_(buffer.exited); + set_status(buffer.return_code); + return status(); +} + +// The AssumeRole process for a Fuchsia death test. It creates a child +// process with the same executable as the current process to run the +// death test. The child process is given the --gtest_filter and +// --gtest_internal_run_death_test flags such that it knows to run the +// current death test only. +DeathTest::TestRole FuchsiaDeathTest::AssumeRole() { + const UnitTestImpl* const impl = GetUnitTestImpl(); + const InternalRunDeathTestFlag* const flag = + impl->internal_run_death_test_flag(); + const TestInfo* const info = impl->current_test_info(); + const int death_test_index = info->result()->death_test_count(); + + if (flag != nullptr) { + // ParseInternalRunDeathTestFlag() has performed all the necessary + // processing. + set_write_fd(kFuchsiaReadPipeFd); + return EXECUTE_TEST; + } + + // Flush the log buffers since the log streams are shared with the child. + FlushInfoLog(); + + // Build the child process command line. + const std::string filter_flag = std::string("--") + GTEST_FLAG_PREFIX_ + + kFilterFlag + "=" + info->test_suite_name() + + "." + info->name(); + const std::string internal_flag = + std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + "=" + + file_ + "|" + + StreamableToString(line_) + "|" + + StreamableToString(death_test_index); + Arguments args; + args.AddArguments(GetInjectableArgvs()); + args.AddArgument(filter_flag.c_str()); + args.AddArgument(internal_flag.c_str()); + + // Build the pipe for communication with the child. + zx_status_t status; + zx_handle_t child_pipe_handle; + int child_pipe_fd; + status = fdio_pipe_half2(&child_pipe_fd, &child_pipe_handle); + GTEST_DEATH_TEST_CHECK_(status == ZX_OK); + set_read_fd(child_pipe_fd); + + // Set the pipe handle for the child. + fdio_spawn_action_t spawn_actions[2] = {}; + fdio_spawn_action_t* add_handle_action = &spawn_actions[0]; + add_handle_action->action = FDIO_SPAWN_ACTION_ADD_HANDLE; + add_handle_action->h.id = PA_HND(PA_FD, kFuchsiaReadPipeFd); + add_handle_action->h.handle = child_pipe_handle; + + // Create a socket pair will be used to receive the child process' stderr. + zx::socket stderr_producer_socket; + status = + zx::socket::create(0, &stderr_producer_socket, &stderr_socket_); + GTEST_DEATH_TEST_CHECK_(status >= 0); + int stderr_producer_fd = -1; + status = + fdio_fd_create(stderr_producer_socket.release(), &stderr_producer_fd); + GTEST_DEATH_TEST_CHECK_(status >= 0); + + // Make the stderr socket nonblocking. + GTEST_DEATH_TEST_CHECK_(fcntl(stderr_producer_fd, F_SETFL, 0) == 0); + + fdio_spawn_action_t* add_stderr_action = &spawn_actions[1]; + add_stderr_action->action = FDIO_SPAWN_ACTION_CLONE_FD; + add_stderr_action->fd.local_fd = stderr_producer_fd; + add_stderr_action->fd.target_fd = STDERR_FILENO; + + // Create a child job. + zx_handle_t child_job = ZX_HANDLE_INVALID; + status = zx_job_create(zx_job_default(), 0, & child_job); + GTEST_DEATH_TEST_CHECK_(status == ZX_OK); + zx_policy_basic_t policy; + policy.condition = ZX_POL_NEW_ANY; + policy.policy = ZX_POL_ACTION_ALLOW; + status = zx_job_set_policy( + child_job, ZX_JOB_POL_RELATIVE, ZX_JOB_POL_BASIC, &policy, 1); + GTEST_DEATH_TEST_CHECK_(status == ZX_OK); + + // Create an exception port and attach it to the |child_job|, to allow + // us to suppress the system default exception handler from firing. + status = zx::port::create(0, &port_); + GTEST_DEATH_TEST_CHECK_(status == ZX_OK); + status = zx_task_bind_exception_port( + child_job, port_.get(), 0 /* key */, 0 /*options */); + GTEST_DEATH_TEST_CHECK_(status == ZX_OK); + + // Spawn the child process. + status = fdio_spawn_etc( + child_job, FDIO_SPAWN_CLONE_ALL, args.Argv()[0], args.Argv(), nullptr, + 2, spawn_actions, child_process_.reset_and_get_address(), nullptr); + GTEST_DEATH_TEST_CHECK_(status == ZX_OK); + + set_spawned(true); + return OVERSEE_TEST; +} + +std::string FuchsiaDeathTest::GetErrorLogs() { + return captured_stderr_; +} + +#else // We are neither on Windows, nor on Fuchsia. + +// ForkingDeathTest provides implementations for most of the abstract +// methods of the DeathTest interface. Only the AssumeRole method is +// left undefined. +class ForkingDeathTest : public DeathTestImpl { + public: + ForkingDeathTest(const char* statement, Matcher matcher); + + // All of these virtual functions are inherited from DeathTest. + int Wait() override; + + protected: + void set_child_pid(pid_t child_pid) { child_pid_ = child_pid; } + + private: + // PID of child process during death test; 0 in the child process itself. + pid_t child_pid_; +}; + +// Constructs a ForkingDeathTest. +ForkingDeathTest::ForkingDeathTest(const char* a_statement, + Matcher matcher) + : DeathTestImpl(a_statement, std::move(matcher)), child_pid_(-1) {} + +// Waits for the child in a death test to exit, returning its exit +// status, or 0 if no child process exists. As a side effect, sets the +// outcome data member. +int ForkingDeathTest::Wait() { + if (!spawned()) + return 0; + + ReadAndInterpretStatusByte(); + + int status_value; + GTEST_DEATH_TEST_CHECK_SYSCALL_(waitpid(child_pid_, &status_value, 0)); + set_status(status_value); + return status_value; +} + +// A concrete death test class that forks, then immediately runs the test +// in the child process. +class NoExecDeathTest : public ForkingDeathTest { + public: + NoExecDeathTest(const char* a_statement, Matcher matcher) + : ForkingDeathTest(a_statement, std::move(matcher)) {} + TestRole AssumeRole() override; +}; + +// The AssumeRole process for a fork-and-run death test. It implements a +// straightforward fork, with a simple pipe to transmit the status byte. +DeathTest::TestRole NoExecDeathTest::AssumeRole() { + const size_t thread_count = GetThreadCount(); + if (thread_count != 1) { + GTEST_LOG_(WARNING) << DeathTestThreadWarning(thread_count); + } + + int pipe_fd[2]; + GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1); + + DeathTest::set_last_death_test_message(""); + CaptureStderr(); + // When we fork the process below, the log file buffers are copied, but the + // file descriptors are shared. We flush all log files here so that closing + // the file descriptors in the child process doesn't throw off the + // synchronization between descriptors and buffers in the parent process. + // This is as close to the fork as possible to avoid a race condition in case + // there are multiple threads running before the death test, and another + // thread writes to the log file. + FlushInfoLog(); + + const pid_t child_pid = fork(); + GTEST_DEATH_TEST_CHECK_(child_pid != -1); + set_child_pid(child_pid); + if (child_pid == 0) { + GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[0])); + set_write_fd(pipe_fd[1]); + // Redirects all logging to stderr in the child process to prevent + // concurrent writes to the log files. We capture stderr in the parent + // process and append the child process' output to a log. + LogToStderr(); + // Event forwarding to the listeners of event listener API mush be shut + // down in death test subprocesses. + GetUnitTestImpl()->listeners()->SuppressEventForwarding(); + g_in_fast_death_test_child = true; + return EXECUTE_TEST; + } else { + GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1])); + set_read_fd(pipe_fd[0]); + set_spawned(true); + return OVERSEE_TEST; + } +} + +// A concrete death test class that forks and re-executes the main +// program from the beginning, with command-line flags set that cause +// only this specific death test to be run. +class ExecDeathTest : public ForkingDeathTest { + public: + ExecDeathTest(const char* a_statement, Matcher matcher, + const char* file, int line) + : ForkingDeathTest(a_statement, std::move(matcher)), + file_(file), + line_(line) {} + TestRole AssumeRole() override; + + private: + static ::std::vector GetArgvsForDeathTestChildProcess() { + ::std::vector args = GetInjectableArgvs(); +# if defined(GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_) + ::std::vector extra_args = + GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_(); + args.insert(args.end(), extra_args.begin(), extra_args.end()); +# endif // defined(GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_) + return args; + } + // The name of the file in which the death test is located. + const char* const file_; + // The line number on which the death test is located. + const int line_; +}; + +// Utility class for accumulating command-line arguments. +class Arguments { + public: + Arguments() { args_.push_back(nullptr); } + + ~Arguments() { + for (std::vector::iterator i = args_.begin(); i != args_.end(); + ++i) { + free(*i); + } + } + void AddArgument(const char* argument) { + args_.insert(args_.end() - 1, posix::StrDup(argument)); + } + + template + void AddArguments(const ::std::vector& arguments) { + for (typename ::std::vector::const_iterator i = arguments.begin(); + i != arguments.end(); + ++i) { + args_.insert(args_.end() - 1, posix::StrDup(i->c_str())); + } + } + char* const* Argv() { + return &args_[0]; + } + + private: + std::vector args_; +}; + +// A struct that encompasses the arguments to the child process of a +// threadsafe-style death test process. +struct ExecDeathTestArgs { + char* const* argv; // Command-line arguments for the child's call to exec + int close_fd; // File descriptor to close; the read end of a pipe +}; + +# if GTEST_OS_MAC +inline char** GetEnviron() { + // When Google Test is built as a framework on MacOS X, the environ variable + // is unavailable. Apple's documentation (man environ) recommends using + // _NSGetEnviron() instead. + return *_NSGetEnviron(); +} +# else +// Some POSIX platforms expect you to declare environ. extern "C" makes +// it reside in the global namespace. +extern "C" char** environ; +inline char** GetEnviron() { return environ; } +# endif // GTEST_OS_MAC + +# if !GTEST_OS_QNX +// The main function for a threadsafe-style death test child process. +// This function is called in a clone()-ed process and thus must avoid +// any potentially unsafe operations like malloc or libc functions. +static int ExecDeathTestChildMain(void* child_arg) { + ExecDeathTestArgs* const args = static_cast(child_arg); + GTEST_DEATH_TEST_CHECK_SYSCALL_(close(args->close_fd)); + + // We need to execute the test program in the same environment where + // it was originally invoked. Therefore we change to the original + // working directory first. + const char* const original_dir = + UnitTest::GetInstance()->original_working_dir(); + // We can safely call chdir() as it's a direct system call. + if (chdir(original_dir) != 0) { + DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " + + GetLastErrnoDescription()); + return EXIT_FAILURE; + } + + // We can safely call execve() as it's a direct system call. We + // cannot use execvp() as it's a libc function and thus potentially + // unsafe. Since execve() doesn't search the PATH, the user must + // invoke the test program via a valid path that contains at least + // one path separator. + execve(args->argv[0], args->argv, GetEnviron()); + DeathTestAbort(std::string("execve(") + args->argv[0] + ", ...) in " + + original_dir + " failed: " + + GetLastErrnoDescription()); + return EXIT_FAILURE; +} +# endif // !GTEST_OS_QNX + +# if GTEST_HAS_CLONE +// Two utility routines that together determine the direction the stack +// grows. +// This could be accomplished more elegantly by a single recursive +// function, but we want to guard against the unlikely possibility of +// a smart compiler optimizing the recursion away. +// +// GTEST_NO_INLINE_ is required to prevent GCC 4.6 from inlining +// StackLowerThanAddress into StackGrowsDown, which then doesn't give +// correct answer. +static void StackLowerThanAddress(const void* ptr, + bool* result) GTEST_NO_INLINE_; +// HWAddressSanitizer add a random tag to the MSB of the local variable address, +// making comparison result unpredictable. +GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +static void StackLowerThanAddress(const void* ptr, bool* result) { + int dummy; + *result = (&dummy < ptr); +} + +// Make sure AddressSanitizer does not tamper with the stack here. +GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +static bool StackGrowsDown() { + int dummy; + bool result; + StackLowerThanAddress(&dummy, &result); + return result; +} +# endif // GTEST_HAS_CLONE + +// Spawns a child process with the same executable as the current process in +// a thread-safe manner and instructs it to run the death test. The +// implementation uses fork(2) + exec. On systems where clone(2) is +// available, it is used instead, being slightly more thread-safe. On QNX, +// fork supports only single-threaded environments, so this function uses +// spawn(2) there instead. The function dies with an error message if +// anything goes wrong. +static pid_t ExecDeathTestSpawnChild(char* const* argv, int close_fd) { + ExecDeathTestArgs args = { argv, close_fd }; + pid_t child_pid = -1; + +# if GTEST_OS_QNX + // Obtains the current directory and sets it to be closed in the child + // process. + const int cwd_fd = open(".", O_RDONLY); + GTEST_DEATH_TEST_CHECK_(cwd_fd != -1); + GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(cwd_fd, F_SETFD, FD_CLOEXEC)); + // We need to execute the test program in the same environment where + // it was originally invoked. Therefore we change to the original + // working directory first. + const char* const original_dir = + UnitTest::GetInstance()->original_working_dir(); + // We can safely call chdir() as it's a direct system call. + if (chdir(original_dir) != 0) { + DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " + + GetLastErrnoDescription()); + return EXIT_FAILURE; + } + + int fd_flags; + // Set close_fd to be closed after spawn. + GTEST_DEATH_TEST_CHECK_SYSCALL_(fd_flags = fcntl(close_fd, F_GETFD)); + GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(close_fd, F_SETFD, + fd_flags | FD_CLOEXEC)); + struct inheritance inherit = {0}; + // spawn is a system call. + child_pid = + spawn(args.argv[0], 0, nullptr, &inherit, args.argv, GetEnviron()); + // Restores the current working directory. + GTEST_DEATH_TEST_CHECK_(fchdir(cwd_fd) != -1); + GTEST_DEATH_TEST_CHECK_SYSCALL_(close(cwd_fd)); + +# else // GTEST_OS_QNX +# if GTEST_OS_LINUX + // When a SIGPROF signal is received while fork() or clone() are executing, + // the process may hang. To avoid this, we ignore SIGPROF here and re-enable + // it after the call to fork()/clone() is complete. + struct sigaction saved_sigprof_action; + struct sigaction ignore_sigprof_action; + memset(&ignore_sigprof_action, 0, sizeof(ignore_sigprof_action)); + sigemptyset(&ignore_sigprof_action.sa_mask); + ignore_sigprof_action.sa_handler = SIG_IGN; + GTEST_DEATH_TEST_CHECK_SYSCALL_(sigaction( + SIGPROF, &ignore_sigprof_action, &saved_sigprof_action)); +# endif // GTEST_OS_LINUX + +# if GTEST_HAS_CLONE + const bool use_fork = GTEST_FLAG(death_test_use_fork); + + if (!use_fork) { + static const bool stack_grows_down = StackGrowsDown(); + const auto stack_size = static_cast(getpagesize()); + // MMAP_ANONYMOUS is not defined on Mac, so we use MAP_ANON instead. + void* const stack = mmap(nullptr, stack_size, PROT_READ | PROT_WRITE, + MAP_ANON | MAP_PRIVATE, -1, 0); + GTEST_DEATH_TEST_CHECK_(stack != MAP_FAILED); + + // Maximum stack alignment in bytes: For a downward-growing stack, this + // amount is subtracted from size of the stack space to get an address + // that is within the stack space and is aligned on all systems we care + // about. As far as I know there is no ABI with stack alignment greater + // than 64. We assume stack and stack_size already have alignment of + // kMaxStackAlignment. + const size_t kMaxStackAlignment = 64; + void* const stack_top = + static_cast(stack) + + (stack_grows_down ? stack_size - kMaxStackAlignment : 0); + GTEST_DEATH_TEST_CHECK_( + static_cast(stack_size) > kMaxStackAlignment && + reinterpret_cast(stack_top) % kMaxStackAlignment == 0); + + child_pid = clone(&ExecDeathTestChildMain, stack_top, SIGCHLD, &args); + + GTEST_DEATH_TEST_CHECK_(munmap(stack, stack_size) != -1); + } +# else + const bool use_fork = true; +# endif // GTEST_HAS_CLONE + + if (use_fork && (child_pid = fork()) == 0) { + ExecDeathTestChildMain(&args); + _exit(0); + } +# endif // GTEST_OS_QNX +# if GTEST_OS_LINUX + GTEST_DEATH_TEST_CHECK_SYSCALL_( + sigaction(SIGPROF, &saved_sigprof_action, nullptr)); +# endif // GTEST_OS_LINUX + + GTEST_DEATH_TEST_CHECK_(child_pid != -1); + return child_pid; +} + +// The AssumeRole process for a fork-and-exec death test. It re-executes the +// main program from the beginning, setting the --gtest_filter +// and --gtest_internal_run_death_test flags to cause only the current +// death test to be re-run. +DeathTest::TestRole ExecDeathTest::AssumeRole() { + const UnitTestImpl* const impl = GetUnitTestImpl(); + const InternalRunDeathTestFlag* const flag = + impl->internal_run_death_test_flag(); + const TestInfo* const info = impl->current_test_info(); + const int death_test_index = info->result()->death_test_count(); + + if (flag != nullptr) { + set_write_fd(flag->write_fd()); + return EXECUTE_TEST; + } + + int pipe_fd[2]; + GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1); + // Clear the close-on-exec flag on the write end of the pipe, lest + // it be closed when the child process does an exec: + GTEST_DEATH_TEST_CHECK_(fcntl(pipe_fd[1], F_SETFD, 0) != -1); + + const std::string filter_flag = std::string("--") + GTEST_FLAG_PREFIX_ + + kFilterFlag + "=" + info->test_suite_name() + + "." + info->name(); + const std::string internal_flag = + std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + "=" + + file_ + "|" + StreamableToString(line_) + "|" + + StreamableToString(death_test_index) + "|" + + StreamableToString(pipe_fd[1]); + Arguments args; + args.AddArguments(GetArgvsForDeathTestChildProcess()); + args.AddArgument(filter_flag.c_str()); + args.AddArgument(internal_flag.c_str()); + + DeathTest::set_last_death_test_message(""); + + CaptureStderr(); + // See the comment in NoExecDeathTest::AssumeRole for why the next line + // is necessary. + FlushInfoLog(); + + const pid_t child_pid = ExecDeathTestSpawnChild(args.Argv(), pipe_fd[0]); + GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1])); + set_child_pid(child_pid); + set_read_fd(pipe_fd[0]); + set_spawned(true); + return OVERSEE_TEST; +} + +# endif // !GTEST_OS_WINDOWS + +// Creates a concrete DeathTest-derived class that depends on the +// --gtest_death_test_style flag, and sets the pointer pointed to +// by the "test" argument to its address. If the test should be +// skipped, sets that pointer to NULL. Returns true, unless the +// flag is set to an invalid value. +bool DefaultDeathTestFactory::Create(const char* statement, + Matcher matcher, + const char* file, int line, + DeathTest** test) { + UnitTestImpl* const impl = GetUnitTestImpl(); + const InternalRunDeathTestFlag* const flag = + impl->internal_run_death_test_flag(); + const int death_test_index = impl->current_test_info() + ->increment_death_test_count(); + + if (flag != nullptr) { + if (death_test_index > flag->index()) { + DeathTest::set_last_death_test_message( + "Death test count (" + StreamableToString(death_test_index) + + ") somehow exceeded expected maximum (" + + StreamableToString(flag->index()) + ")"); + return false; + } + + if (!(flag->file() == file && flag->line() == line && + flag->index() == death_test_index)) { + *test = nullptr; + return true; + } + } + +# if GTEST_OS_WINDOWS + + if (GTEST_FLAG(death_test_style) == "threadsafe" || + GTEST_FLAG(death_test_style) == "fast") { + *test = new WindowsDeathTest(statement, std::move(matcher), file, line); + } + +# elif GTEST_OS_FUCHSIA + + if (GTEST_FLAG(death_test_style) == "threadsafe" || + GTEST_FLAG(death_test_style) == "fast") { + *test = new FuchsiaDeathTest(statement, std::move(matcher), file, line); + } + +# else + + if (GTEST_FLAG(death_test_style) == "threadsafe") { + *test = new ExecDeathTest(statement, std::move(matcher), file, line); + } else if (GTEST_FLAG(death_test_style) == "fast") { + *test = new NoExecDeathTest(statement, std::move(matcher)); + } + +# endif // GTEST_OS_WINDOWS + + else { // NOLINT - this is more readable than unbalanced brackets inside #if. + DeathTest::set_last_death_test_message( + "Unknown death test style \"" + GTEST_FLAG(death_test_style) + + "\" encountered"); + return false; + } + + return true; +} + +# if GTEST_OS_WINDOWS +// Recreates the pipe and event handles from the provided parameters, +// signals the event, and returns a file descriptor wrapped around the pipe +// handle. This function is called in the child process only. +static int GetStatusFileDescriptor(unsigned int parent_process_id, + size_t write_handle_as_size_t, + size_t event_handle_as_size_t) { + AutoHandle parent_process_handle(::OpenProcess(PROCESS_DUP_HANDLE, + FALSE, // Non-inheritable. + parent_process_id)); + if (parent_process_handle.Get() == INVALID_HANDLE_VALUE) { + DeathTestAbort("Unable to open parent process " + + StreamableToString(parent_process_id)); + } + + GTEST_CHECK_(sizeof(HANDLE) <= sizeof(size_t)); + + const HANDLE write_handle = + reinterpret_cast(write_handle_as_size_t); + HANDLE dup_write_handle; + + // The newly initialized handle is accessible only in the parent + // process. To obtain one accessible within the child, we need to use + // DuplicateHandle. + if (!::DuplicateHandle(parent_process_handle.Get(), write_handle, + ::GetCurrentProcess(), &dup_write_handle, + 0x0, // Requested privileges ignored since + // DUPLICATE_SAME_ACCESS is used. + FALSE, // Request non-inheritable handler. + DUPLICATE_SAME_ACCESS)) { + DeathTestAbort("Unable to duplicate the pipe handle " + + StreamableToString(write_handle_as_size_t) + + " from the parent process " + + StreamableToString(parent_process_id)); + } + + const HANDLE event_handle = reinterpret_cast(event_handle_as_size_t); + HANDLE dup_event_handle; + + if (!::DuplicateHandle(parent_process_handle.Get(), event_handle, + ::GetCurrentProcess(), &dup_event_handle, + 0x0, + FALSE, + DUPLICATE_SAME_ACCESS)) { + DeathTestAbort("Unable to duplicate the event handle " + + StreamableToString(event_handle_as_size_t) + + " from the parent process " + + StreamableToString(parent_process_id)); + } + + const int write_fd = + ::_open_osfhandle(reinterpret_cast(dup_write_handle), O_APPEND); + if (write_fd == -1) { + DeathTestAbort("Unable to convert pipe handle " + + StreamableToString(write_handle_as_size_t) + + " to a file descriptor"); + } + + // Signals the parent that the write end of the pipe has been acquired + // so the parent can release its own write end. + ::SetEvent(dup_event_handle); + + return write_fd; +} +# endif // GTEST_OS_WINDOWS + +// Returns a newly created InternalRunDeathTestFlag object with fields +// initialized from the GTEST_FLAG(internal_run_death_test) flag if +// the flag is specified; otherwise returns NULL. +InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag() { + if (GTEST_FLAG(internal_run_death_test) == "") return nullptr; + + // GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we + // can use it here. + int line = -1; + int index = -1; + ::std::vector< ::std::string> fields; + SplitString(GTEST_FLAG(internal_run_death_test).c_str(), '|', &fields); + int write_fd = -1; + +# if GTEST_OS_WINDOWS + + unsigned int parent_process_id = 0; + size_t write_handle_as_size_t = 0; + size_t event_handle_as_size_t = 0; + + if (fields.size() != 6 + || !ParseNaturalNumber(fields[1], &line) + || !ParseNaturalNumber(fields[2], &index) + || !ParseNaturalNumber(fields[3], &parent_process_id) + || !ParseNaturalNumber(fields[4], &write_handle_as_size_t) + || !ParseNaturalNumber(fields[5], &event_handle_as_size_t)) { + DeathTestAbort("Bad --gtest_internal_run_death_test flag: " + + GTEST_FLAG(internal_run_death_test)); + } + write_fd = GetStatusFileDescriptor(parent_process_id, + write_handle_as_size_t, + event_handle_as_size_t); + +# elif GTEST_OS_FUCHSIA + + if (fields.size() != 3 + || !ParseNaturalNumber(fields[1], &line) + || !ParseNaturalNumber(fields[2], &index)) { + DeathTestAbort("Bad --gtest_internal_run_death_test flag: " + + GTEST_FLAG(internal_run_death_test)); + } + +# else + + if (fields.size() != 4 + || !ParseNaturalNumber(fields[1], &line) + || !ParseNaturalNumber(fields[2], &index) + || !ParseNaturalNumber(fields[3], &write_fd)) { + DeathTestAbort("Bad --gtest_internal_run_death_test flag: " + + GTEST_FLAG(internal_run_death_test)); + } + +# endif // GTEST_OS_WINDOWS + + return new InternalRunDeathTestFlag(fields[0], line, index, write_fd); +} + +} // namespace internal + +#endif // GTEST_HAS_DEATH_TEST + +} // namespace testing diff --git a/deps/googletest/src/gtest-filepath.cc b/deps/googletest/src/gtest-filepath.cc new file mode 100644 index 000000000..322fbb1b4 --- /dev/null +++ b/deps/googletest/src/gtest-filepath.cc @@ -0,0 +1,379 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include "gtest/internal/gtest-filepath.h" + +#include +#include "gtest/internal/gtest-port.h" +#include "gtest/gtest-message.h" + +#if GTEST_OS_WINDOWS_MOBILE +# include +#elif GTEST_OS_WINDOWS +# include +# include +#else +# include +# include // Some Linux distributions define PATH_MAX here. +#endif // GTEST_OS_WINDOWS_MOBILE + +#include "gtest/internal/gtest-string.h" + +#if GTEST_OS_WINDOWS +# define GTEST_PATH_MAX_ _MAX_PATH +#elif defined(PATH_MAX) +# define GTEST_PATH_MAX_ PATH_MAX +#elif defined(_XOPEN_PATH_MAX) +# define GTEST_PATH_MAX_ _XOPEN_PATH_MAX +#else +# define GTEST_PATH_MAX_ _POSIX_PATH_MAX +#endif // GTEST_OS_WINDOWS + +namespace testing { +namespace internal { + +#if GTEST_OS_WINDOWS +// On Windows, '\\' is the standard path separator, but many tools and the +// Windows API also accept '/' as an alternate path separator. Unless otherwise +// noted, a file path can contain either kind of path separators, or a mixture +// of them. +const char kPathSeparator = '\\'; +const char kAlternatePathSeparator = '/'; +const char kAlternatePathSeparatorString[] = "/"; +# if GTEST_OS_WINDOWS_MOBILE +// Windows CE doesn't have a current directory. You should not use +// the current directory in tests on Windows CE, but this at least +// provides a reasonable fallback. +const char kCurrentDirectoryString[] = "\\"; +// Windows CE doesn't define INVALID_FILE_ATTRIBUTES +const DWORD kInvalidFileAttributes = 0xffffffff; +# else +const char kCurrentDirectoryString[] = ".\\"; +# endif // GTEST_OS_WINDOWS_MOBILE +#else +const char kPathSeparator = '/'; +const char kCurrentDirectoryString[] = "./"; +#endif // GTEST_OS_WINDOWS + +// Returns whether the given character is a valid path separator. +static bool IsPathSeparator(char c) { +#if GTEST_HAS_ALT_PATH_SEP_ + return (c == kPathSeparator) || (c == kAlternatePathSeparator); +#else + return c == kPathSeparator; +#endif +} + +// Returns the current working directory, or "" if unsuccessful. +FilePath FilePath::GetCurrentDir() { +#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE || \ + GTEST_OS_WINDOWS_RT || ARDUINO + // Windows CE and Arduino don't have a current directory, so we just return + // something reasonable. + return FilePath(kCurrentDirectoryString); +#elif GTEST_OS_WINDOWS + char cwd[GTEST_PATH_MAX_ + 1] = { '\0' }; + return FilePath(_getcwd(cwd, sizeof(cwd)) == nullptr ? "" : cwd); +#else + char cwd[GTEST_PATH_MAX_ + 1] = { '\0' }; + char* result = getcwd(cwd, sizeof(cwd)); +# if GTEST_OS_NACL + // getcwd will likely fail in NaCl due to the sandbox, so return something + // reasonable. The user may have provided a shim implementation for getcwd, + // however, so fallback only when failure is detected. + return FilePath(result == nullptr ? kCurrentDirectoryString : cwd); +# endif // GTEST_OS_NACL + return FilePath(result == nullptr ? "" : cwd); +#endif // GTEST_OS_WINDOWS_MOBILE +} + +// Returns a copy of the FilePath with the case-insensitive extension removed. +// Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns +// FilePath("dir/file"). If a case-insensitive extension is not +// found, returns a copy of the original FilePath. +FilePath FilePath::RemoveExtension(const char* extension) const { + const std::string dot_extension = std::string(".") + extension; + if (String::EndsWithCaseInsensitive(pathname_, dot_extension)) { + return FilePath(pathname_.substr( + 0, pathname_.length() - dot_extension.length())); + } + return *this; +} + +// Returns a pointer to the last occurrence of a valid path separator in +// the FilePath. On Windows, for example, both '/' and '\' are valid path +// separators. Returns NULL if no path separator was found. +const char* FilePath::FindLastPathSeparator() const { + const char* const last_sep = strrchr(c_str(), kPathSeparator); +#if GTEST_HAS_ALT_PATH_SEP_ + const char* const last_alt_sep = strrchr(c_str(), kAlternatePathSeparator); + // Comparing two pointers of which only one is NULL is undefined. + if (last_alt_sep != nullptr && + (last_sep == nullptr || last_alt_sep > last_sep)) { + return last_alt_sep; + } +#endif + return last_sep; +} + +// Returns a copy of the FilePath with the directory part removed. +// Example: FilePath("path/to/file").RemoveDirectoryName() returns +// FilePath("file"). If there is no directory part ("just_a_file"), it returns +// the FilePath unmodified. If there is no file part ("just_a_dir/") it +// returns an empty FilePath (""). +// On Windows platform, '\' is the path separator, otherwise it is '/'. +FilePath FilePath::RemoveDirectoryName() const { + const char* const last_sep = FindLastPathSeparator(); + return last_sep ? FilePath(last_sep + 1) : *this; +} + +// RemoveFileName returns the directory path with the filename removed. +// Example: FilePath("path/to/file").RemoveFileName() returns "path/to/". +// If the FilePath is "a_file" or "/a_file", RemoveFileName returns +// FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does +// not have a file, like "just/a/dir/", it returns the FilePath unmodified. +// On Windows platform, '\' is the path separator, otherwise it is '/'. +FilePath FilePath::RemoveFileName() const { + const char* const last_sep = FindLastPathSeparator(); + std::string dir; + if (last_sep) { + dir = std::string(c_str(), static_cast(last_sep + 1 - c_str())); + } else { + dir = kCurrentDirectoryString; + } + return FilePath(dir); +} + +// Helper functions for naming files in a directory for xml output. + +// Given directory = "dir", base_name = "test", number = 0, +// extension = "xml", returns "dir/test.xml". If number is greater +// than zero (e.g., 12), returns "dir/test_12.xml". +// On Windows platform, uses \ as the separator rather than /. +FilePath FilePath::MakeFileName(const FilePath& directory, + const FilePath& base_name, + int number, + const char* extension) { + std::string file; + if (number == 0) { + file = base_name.string() + "." + extension; + } else { + file = base_name.string() + "_" + StreamableToString(number) + + "." + extension; + } + return ConcatPaths(directory, FilePath(file)); +} + +// Given directory = "dir", relative_path = "test.xml", returns "dir/test.xml". +// On Windows, uses \ as the separator rather than /. +FilePath FilePath::ConcatPaths(const FilePath& directory, + const FilePath& relative_path) { + if (directory.IsEmpty()) + return relative_path; + const FilePath dir(directory.RemoveTrailingPathSeparator()); + return FilePath(dir.string() + kPathSeparator + relative_path.string()); +} + +// Returns true if pathname describes something findable in the file-system, +// either a file, directory, or whatever. +bool FilePath::FileOrDirectoryExists() const { +#if GTEST_OS_WINDOWS_MOBILE + LPCWSTR unicode = String::AnsiToUtf16(pathname_.c_str()); + const DWORD attributes = GetFileAttributes(unicode); + delete [] unicode; + return attributes != kInvalidFileAttributes; +#else + posix::StatStruct file_stat; + return posix::Stat(pathname_.c_str(), &file_stat) == 0; +#endif // GTEST_OS_WINDOWS_MOBILE +} + +// Returns true if pathname describes a directory in the file-system +// that exists. +bool FilePath::DirectoryExists() const { + bool result = false; +#if GTEST_OS_WINDOWS + // Don't strip off trailing separator if path is a root directory on + // Windows (like "C:\\"). + const FilePath& path(IsRootDirectory() ? *this : + RemoveTrailingPathSeparator()); +#else + const FilePath& path(*this); +#endif + +#if GTEST_OS_WINDOWS_MOBILE + LPCWSTR unicode = String::AnsiToUtf16(path.c_str()); + const DWORD attributes = GetFileAttributes(unicode); + delete [] unicode; + if ((attributes != kInvalidFileAttributes) && + (attributes & FILE_ATTRIBUTE_DIRECTORY)) { + result = true; + } +#else + posix::StatStruct file_stat; + result = posix::Stat(path.c_str(), &file_stat) == 0 && + posix::IsDir(file_stat); +#endif // GTEST_OS_WINDOWS_MOBILE + + return result; +} + +// Returns true if pathname describes a root directory. (Windows has one +// root directory per disk drive.) +bool FilePath::IsRootDirectory() const { +#if GTEST_OS_WINDOWS + return pathname_.length() == 3 && IsAbsolutePath(); +#else + return pathname_.length() == 1 && IsPathSeparator(pathname_.c_str()[0]); +#endif +} + +// Returns true if pathname describes an absolute path. +bool FilePath::IsAbsolutePath() const { + const char* const name = pathname_.c_str(); +#if GTEST_OS_WINDOWS + return pathname_.length() >= 3 && + ((name[0] >= 'a' && name[0] <= 'z') || + (name[0] >= 'A' && name[0] <= 'Z')) && + name[1] == ':' && + IsPathSeparator(name[2]); +#else + return IsPathSeparator(name[0]); +#endif +} + +// Returns a pathname for a file that does not currently exist. The pathname +// will be directory/base_name.extension or +// directory/base_name_.extension if directory/base_name.extension +// already exists. The number will be incremented until a pathname is found +// that does not already exist. +// Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'. +// There could be a race condition if two or more processes are calling this +// function at the same time -- they could both pick the same filename. +FilePath FilePath::GenerateUniqueFileName(const FilePath& directory, + const FilePath& base_name, + const char* extension) { + FilePath full_pathname; + int number = 0; + do { + full_pathname.Set(MakeFileName(directory, base_name, number++, extension)); + } while (full_pathname.FileOrDirectoryExists()); + return full_pathname; +} + +// Returns true if FilePath ends with a path separator, which indicates that +// it is intended to represent a directory. Returns false otherwise. +// This does NOT check that a directory (or file) actually exists. +bool FilePath::IsDirectory() const { + return !pathname_.empty() && + IsPathSeparator(pathname_.c_str()[pathname_.length() - 1]); +} + +// Create directories so that path exists. Returns true if successful or if +// the directories already exist; returns false if unable to create directories +// for any reason. +bool FilePath::CreateDirectoriesRecursively() const { + if (!this->IsDirectory()) { + return false; + } + + if (pathname_.length() == 0 || this->DirectoryExists()) { + return true; + } + + const FilePath parent(this->RemoveTrailingPathSeparator().RemoveFileName()); + return parent.CreateDirectoriesRecursively() && this->CreateFolder(); +} + +// Create the directory so that path exists. Returns true if successful or +// if the directory already exists; returns false if unable to create the +// directory for any reason, including if the parent directory does not +// exist. Not named "CreateDirectory" because that's a macro on Windows. +bool FilePath::CreateFolder() const { +#if GTEST_OS_WINDOWS_MOBILE + FilePath removed_sep(this->RemoveTrailingPathSeparator()); + LPCWSTR unicode = String::AnsiToUtf16(removed_sep.c_str()); + int result = CreateDirectory(unicode, nullptr) ? 0 : -1; + delete [] unicode; +#elif GTEST_OS_WINDOWS + int result = _mkdir(pathname_.c_str()); +#else + int result = mkdir(pathname_.c_str(), 0777); +#endif // GTEST_OS_WINDOWS_MOBILE + + if (result == -1) { + return this->DirectoryExists(); // An error is OK if the directory exists. + } + return true; // No error. +} + +// If input name has a trailing separator character, remove it and return the +// name, otherwise return the name string unmodified. +// On Windows platform, uses \ as the separator, other platforms use /. +FilePath FilePath::RemoveTrailingPathSeparator() const { + return IsDirectory() + ? FilePath(pathname_.substr(0, pathname_.length() - 1)) + : *this; +} + +// Removes any redundant separators that might be in the pathname. +// For example, "bar///foo" becomes "bar/foo". Does not eliminate other +// redundancies that might be in a pathname involving "." or "..". +void FilePath::Normalize() { + if (pathname_.c_str() == nullptr) { + pathname_ = ""; + return; + } + const char* src = pathname_.c_str(); + char* const dest = new char[pathname_.length() + 1]; + char* dest_ptr = dest; + memset(dest_ptr, 0, pathname_.length() + 1); + + while (*src != '\0') { + *dest_ptr = *src; + if (!IsPathSeparator(*src)) { + src++; + } else { +#if GTEST_HAS_ALT_PATH_SEP_ + if (*dest_ptr == kAlternatePathSeparator) { + *dest_ptr = kPathSeparator; + } +#endif + while (IsPathSeparator(*src)) + src++; + } + dest_ptr++; + } + *dest_ptr = '\0'; + pathname_ = dest; + delete[] dest; +} + +} // namespace internal +} // namespace testing diff --git a/deps/googletest/src/gtest-internal-inl.h b/deps/googletest/src/gtest-internal-inl.h new file mode 100644 index 000000000..53cd22b0f --- /dev/null +++ b/deps/googletest/src/gtest-internal-inl.h @@ -0,0 +1,1210 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// Utility functions and classes used by the Google C++ testing framework.// +// This file contains purely Google Test's internal implementation. Please +// DO NOT #INCLUDE IT IN A USER PROGRAM. + +#ifndef GTEST_SRC_GTEST_INTERNAL_INL_H_ +#define GTEST_SRC_GTEST_INTERNAL_INL_H_ + +#ifndef _WIN32_WCE +# include +#endif // !_WIN32_WCE +#include +#include // For strtoll/_strtoul64/malloc/free. +#include // For memmove. + +#include +#include +#include +#include + +#include "gtest/internal/gtest-port.h" + +#if GTEST_CAN_STREAM_RESULTS_ +# include // NOLINT +# include // NOLINT +#endif + +#if GTEST_OS_WINDOWS +# include // NOLINT +#endif // GTEST_OS_WINDOWS + +#include "gtest/gtest.h" +#include "gtest/gtest-spi.h" + +GTEST_DISABLE_MSC_WARNINGS_PUSH_(4251 \ +/* class A needs to have dll-interface to be used by clients of class B */) + +namespace testing { + +// Declares the flags. +// +// We don't want the users to modify this flag in the code, but want +// Google Test's own unit tests to be able to access it. Therefore we +// declare it here as opposed to in gtest.h. +GTEST_DECLARE_bool_(death_test_use_fork); + +namespace internal { + +// The value of GetTestTypeId() as seen from within the Google Test +// library. This is solely for testing GetTestTypeId(). +GTEST_API_ extern const TypeId kTestTypeIdInGoogleTest; + +// Names of the flags (needed for parsing Google Test flags). +const char kAlsoRunDisabledTestsFlag[] = "also_run_disabled_tests"; +const char kBreakOnFailureFlag[] = "break_on_failure"; +const char kCatchExceptionsFlag[] = "catch_exceptions"; +const char kColorFlag[] = "color"; +const char kFilterFlag[] = "filter"; +const char kListTestsFlag[] = "list_tests"; +const char kOutputFlag[] = "output"; +const char kPrintTimeFlag[] = "print_time"; +const char kPrintUTF8Flag[] = "print_utf8"; +const char kRandomSeedFlag[] = "random_seed"; +const char kRepeatFlag[] = "repeat"; +const char kShuffleFlag[] = "shuffle"; +const char kStackTraceDepthFlag[] = "stack_trace_depth"; +const char kStreamResultToFlag[] = "stream_result_to"; +const char kThrowOnFailureFlag[] = "throw_on_failure"; +const char kFlagfileFlag[] = "flagfile"; + +// A valid random seed must be in [1, kMaxRandomSeed]. +const int kMaxRandomSeed = 99999; + +// g_help_flag is true iff the --help flag or an equivalent form is +// specified on the command line. +GTEST_API_ extern bool g_help_flag; + +// Returns the current time in milliseconds. +GTEST_API_ TimeInMillis GetTimeInMillis(); + +// Returns true iff Google Test should use colors in the output. +GTEST_API_ bool ShouldUseColor(bool stdout_is_tty); + +// Formats the given time in milliseconds as seconds. +GTEST_API_ std::string FormatTimeInMillisAsSeconds(TimeInMillis ms); + +// Converts the given time in milliseconds to a date string in the ISO 8601 +// format, without the timezone information. N.B.: due to the use the +// non-reentrant localtime() function, this function is not thread safe. Do +// not use it in any code that can be called from multiple threads. +GTEST_API_ std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms); + +// Parses a string for an Int32 flag, in the form of "--flag=value". +// +// On success, stores the value of the flag in *value, and returns +// true. On failure, returns false without changing *value. +GTEST_API_ bool ParseInt32Flag( + const char* str, const char* flag, Int32* value); + +// Returns a random seed in range [1, kMaxRandomSeed] based on the +// given --gtest_random_seed flag value. +inline int GetRandomSeedFromFlag(Int32 random_seed_flag) { + const unsigned int raw_seed = (random_seed_flag == 0) ? + static_cast(GetTimeInMillis()) : + static_cast(random_seed_flag); + + // Normalizes the actual seed to range [1, kMaxRandomSeed] such that + // it's easy to type. + const int normalized_seed = + static_cast((raw_seed - 1U) % + static_cast(kMaxRandomSeed)) + 1; + return normalized_seed; +} + +// Returns the first valid random seed after 'seed'. The behavior is +// undefined if 'seed' is invalid. The seed after kMaxRandomSeed is +// considered to be 1. +inline int GetNextRandomSeed(int seed) { + GTEST_CHECK_(1 <= seed && seed <= kMaxRandomSeed) + << "Invalid random seed " << seed << " - must be in [1, " + << kMaxRandomSeed << "]."; + const int next_seed = seed + 1; + return (next_seed > kMaxRandomSeed) ? 1 : next_seed; +} + +// This class saves the values of all Google Test flags in its c'tor, and +// restores them in its d'tor. +class GTestFlagSaver { + public: + // The c'tor. + GTestFlagSaver() { + also_run_disabled_tests_ = GTEST_FLAG(also_run_disabled_tests); + break_on_failure_ = GTEST_FLAG(break_on_failure); + catch_exceptions_ = GTEST_FLAG(catch_exceptions); + color_ = GTEST_FLAG(color); + death_test_style_ = GTEST_FLAG(death_test_style); + death_test_use_fork_ = GTEST_FLAG(death_test_use_fork); + filter_ = GTEST_FLAG(filter); + internal_run_death_test_ = GTEST_FLAG(internal_run_death_test); + list_tests_ = GTEST_FLAG(list_tests); + output_ = GTEST_FLAG(output); + print_time_ = GTEST_FLAG(print_time); + print_utf8_ = GTEST_FLAG(print_utf8); + random_seed_ = GTEST_FLAG(random_seed); + repeat_ = GTEST_FLAG(repeat); + shuffle_ = GTEST_FLAG(shuffle); + stack_trace_depth_ = GTEST_FLAG(stack_trace_depth); + stream_result_to_ = GTEST_FLAG(stream_result_to); + throw_on_failure_ = GTEST_FLAG(throw_on_failure); + } + + // The d'tor is not virtual. DO NOT INHERIT FROM THIS CLASS. + ~GTestFlagSaver() { + GTEST_FLAG(also_run_disabled_tests) = also_run_disabled_tests_; + GTEST_FLAG(break_on_failure) = break_on_failure_; + GTEST_FLAG(catch_exceptions) = catch_exceptions_; + GTEST_FLAG(color) = color_; + GTEST_FLAG(death_test_style) = death_test_style_; + GTEST_FLAG(death_test_use_fork) = death_test_use_fork_; + GTEST_FLAG(filter) = filter_; + GTEST_FLAG(internal_run_death_test) = internal_run_death_test_; + GTEST_FLAG(list_tests) = list_tests_; + GTEST_FLAG(output) = output_; + GTEST_FLAG(print_time) = print_time_; + GTEST_FLAG(print_utf8) = print_utf8_; + GTEST_FLAG(random_seed) = random_seed_; + GTEST_FLAG(repeat) = repeat_; + GTEST_FLAG(shuffle) = shuffle_; + GTEST_FLAG(stack_trace_depth) = stack_trace_depth_; + GTEST_FLAG(stream_result_to) = stream_result_to_; + GTEST_FLAG(throw_on_failure) = throw_on_failure_; + } + + private: + // Fields for saving the original values of flags. + bool also_run_disabled_tests_; + bool break_on_failure_; + bool catch_exceptions_; + std::string color_; + std::string death_test_style_; + bool death_test_use_fork_; + std::string filter_; + std::string internal_run_death_test_; + bool list_tests_; + std::string output_; + bool print_time_; + bool print_utf8_; + internal::Int32 random_seed_; + internal::Int32 repeat_; + bool shuffle_; + internal::Int32 stack_trace_depth_; + std::string stream_result_to_; + bool throw_on_failure_; +} GTEST_ATTRIBUTE_UNUSED_; + +// Converts a Unicode code point to a narrow string in UTF-8 encoding. +// code_point parameter is of type UInt32 because wchar_t may not be +// wide enough to contain a code point. +// If the code_point is not a valid Unicode code point +// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted +// to "(Invalid Unicode 0xXXXXXXXX)". +GTEST_API_ std::string CodePointToUtf8(UInt32 code_point); + +// Converts a wide string to a narrow string in UTF-8 encoding. +// The wide string is assumed to have the following encoding: +// UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin) +// UTF-32 if sizeof(wchar_t) == 4 (on Linux) +// Parameter str points to a null-terminated wide string. +// Parameter num_chars may additionally limit the number +// of wchar_t characters processed. -1 is used when the entire string +// should be processed. +// If the string contains code points that are not valid Unicode code points +// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output +// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding +// and contains invalid UTF-16 surrogate pairs, values in those pairs +// will be encoded as individual Unicode characters from Basic Normal Plane. +GTEST_API_ std::string WideStringToUtf8(const wchar_t* str, int num_chars); + +// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file +// if the variable is present. If a file already exists at this location, this +// function will write over it. If the variable is present, but the file cannot +// be created, prints an error and exits. +void WriteToShardStatusFileIfNeeded(); + +// Checks whether sharding is enabled by examining the relevant +// environment variable values. If the variables are present, +// but inconsistent (e.g., shard_index >= total_shards), prints +// an error and exits. If in_subprocess_for_death_test, sharding is +// disabled because it must only be applied to the original test +// process. Otherwise, we could filter out death tests we intended to execute. +GTEST_API_ bool ShouldShard(const char* total_shards_str, + const char* shard_index_str, + bool in_subprocess_for_death_test); + +// Parses the environment variable var as an Int32. If it is unset, +// returns default_val. If it is not an Int32, prints an error and +// and aborts. +GTEST_API_ Int32 Int32FromEnvOrDie(const char* env_var, Int32 default_val); + +// Given the total number of shards, the shard index, and the test id, +// returns true iff the test should be run on this shard. The test id is +// some arbitrary but unique non-negative integer assigned to each test +// method. Assumes that 0 <= shard_index < total_shards. +GTEST_API_ bool ShouldRunTestOnShard( + int total_shards, int shard_index, int test_id); + +// STL container utilities. + +// Returns the number of elements in the given container that satisfy +// the given predicate. +template +inline int CountIf(const Container& c, Predicate predicate) { + // Implemented as an explicit loop since std::count_if() in libCstd on + // Solaris has a non-standard signature. + int count = 0; + for (typename Container::const_iterator it = c.begin(); it != c.end(); ++it) { + if (predicate(*it)) + ++count; + } + return count; +} + +// Applies a function/functor to each element in the container. +template +void ForEach(const Container& c, Functor functor) { + std::for_each(c.begin(), c.end(), functor); +} + +// Returns the i-th element of the vector, or default_value if i is not +// in range [0, v.size()). +template +inline E GetElementOr(const std::vector& v, int i, E default_value) { + return (i < 0 || i >= static_cast(v.size())) ? default_value + : v[static_cast(i)]; +} + +// Performs an in-place shuffle of a range of the vector's elements. +// 'begin' and 'end' are element indices as an STL-style range; +// i.e. [begin, end) are shuffled, where 'end' == size() means to +// shuffle to the end of the vector. +template +void ShuffleRange(internal::Random* random, int begin, int end, + std::vector* v) { + const int size = static_cast(v->size()); + GTEST_CHECK_(0 <= begin && begin <= size) + << "Invalid shuffle range start " << begin << ": must be in range [0, " + << size << "]."; + GTEST_CHECK_(begin <= end && end <= size) + << "Invalid shuffle range finish " << end << ": must be in range [" + << begin << ", " << size << "]."; + + // Fisher-Yates shuffle, from + // http://en.wikipedia.org/wiki/Fisher-Yates_shuffle + for (int range_width = end - begin; range_width >= 2; range_width--) { + const int last_in_range = begin + range_width - 1; + const int selected = + begin + + static_cast(random->Generate(static_cast(range_width))); + std::swap((*v)[static_cast(selected)], + (*v)[static_cast(last_in_range)]); + } +} + +// Performs an in-place shuffle of the vector's elements. +template +inline void Shuffle(internal::Random* random, std::vector* v) { + ShuffleRange(random, 0, static_cast(v->size()), v); +} + +// A function for deleting an object. Handy for being used as a +// functor. +template +static void Delete(T* x) { + delete x; +} + +// A predicate that checks the key of a TestProperty against a known key. +// +// TestPropertyKeyIs is copyable. +class TestPropertyKeyIs { + public: + // Constructor. + // + // TestPropertyKeyIs has NO default constructor. + explicit TestPropertyKeyIs(const std::string& key) : key_(key) {} + + // Returns true iff the test name of test property matches on key_. + bool operator()(const TestProperty& test_property) const { + return test_property.key() == key_; + } + + private: + std::string key_; +}; + +// Class UnitTestOptions. +// +// This class contains functions for processing options the user +// specifies when running the tests. It has only static members. +// +// In most cases, the user can specify an option using either an +// environment variable or a command line flag. E.g. you can set the +// test filter using either GTEST_FILTER or --gtest_filter. If both +// the variable and the flag are present, the latter overrides the +// former. +class GTEST_API_ UnitTestOptions { + public: + // Functions for processing the gtest_output flag. + + // Returns the output format, or "" for normal printed output. + static std::string GetOutputFormat(); + + // Returns the absolute path of the requested output file, or the + // default (test_detail.xml in the original working directory) if + // none was explicitly specified. + static std::string GetAbsolutePathToOutputFile(); + + // Functions for processing the gtest_filter flag. + + // Returns true iff the wildcard pattern matches the string. The + // first ':' or '\0' character in pattern marks the end of it. + // + // This recursive algorithm isn't very efficient, but is clear and + // works well enough for matching test names, which are short. + static bool PatternMatchesString(const char *pattern, const char *str); + + // Returns true iff the user-specified filter matches the test suite + // name and the test name. + static bool FilterMatchesTest(const std::string& test_suite_name, + const std::string& test_name); + +#if GTEST_OS_WINDOWS + // Function for supporting the gtest_catch_exception flag. + + // Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the + // given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise. + // This function is useful as an __except condition. + static int GTestShouldProcessSEH(DWORD exception_code); +#endif // GTEST_OS_WINDOWS + + // Returns true if "name" matches the ':' separated list of glob-style + // filters in "filter". + static bool MatchesFilter(const std::string& name, const char* filter); +}; + +// Returns the current application's name, removing directory path if that +// is present. Used by UnitTestOptions::GetOutputFile. +GTEST_API_ FilePath GetCurrentExecutableName(); + +// The role interface for getting the OS stack trace as a string. +class OsStackTraceGetterInterface { + public: + OsStackTraceGetterInterface() {} + virtual ~OsStackTraceGetterInterface() {} + + // Returns the current OS stack trace as an std::string. Parameters: + // + // max_depth - the maximum number of stack frames to be included + // in the trace. + // skip_count - the number of top frames to be skipped; doesn't count + // against max_depth. + virtual std::string CurrentStackTrace(int max_depth, int skip_count) = 0; + + // UponLeavingGTest() should be called immediately before Google Test calls + // user code. It saves some information about the current stack that + // CurrentStackTrace() will use to find and hide Google Test stack frames. + virtual void UponLeavingGTest() = 0; + + // This string is inserted in place of stack frames that are part of + // Google Test's implementation. + static const char* const kElidedFramesMarker; + + private: + GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetterInterface); +}; + +// A working implementation of the OsStackTraceGetterInterface interface. +class OsStackTraceGetter : public OsStackTraceGetterInterface { + public: + OsStackTraceGetter() {} + + std::string CurrentStackTrace(int max_depth, int skip_count) override; + void UponLeavingGTest() override; + + private: +#if GTEST_HAS_ABSL + Mutex mutex_; // Protects all internal state. + + // We save the stack frame below the frame that calls user code. + // We do this because the address of the frame immediately below + // the user code changes between the call to UponLeavingGTest() + // and any calls to the stack trace code from within the user code. + void* caller_frame_ = nullptr; +#endif // GTEST_HAS_ABSL + + GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetter); +}; + +// Information about a Google Test trace point. +struct TraceInfo { + const char* file; + int line; + std::string message; +}; + +// This is the default global test part result reporter used in UnitTestImpl. +// This class should only be used by UnitTestImpl. +class DefaultGlobalTestPartResultReporter + : public TestPartResultReporterInterface { + public: + explicit DefaultGlobalTestPartResultReporter(UnitTestImpl* unit_test); + // Implements the TestPartResultReporterInterface. Reports the test part + // result in the current test. + void ReportTestPartResult(const TestPartResult& result) override; + + private: + UnitTestImpl* const unit_test_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultGlobalTestPartResultReporter); +}; + +// This is the default per thread test part result reporter used in +// UnitTestImpl. This class should only be used by UnitTestImpl. +class DefaultPerThreadTestPartResultReporter + : public TestPartResultReporterInterface { + public: + explicit DefaultPerThreadTestPartResultReporter(UnitTestImpl* unit_test); + // Implements the TestPartResultReporterInterface. The implementation just + // delegates to the current global test part result reporter of *unit_test_. + void ReportTestPartResult(const TestPartResult& result) override; + + private: + UnitTestImpl* const unit_test_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultPerThreadTestPartResultReporter); +}; + +// The private implementation of the UnitTest class. We don't protect +// the methods under a mutex, as this class is not accessible by a +// user and the UnitTest class that delegates work to this class does +// proper locking. +class GTEST_API_ UnitTestImpl { + public: + explicit UnitTestImpl(UnitTest* parent); + virtual ~UnitTestImpl(); + + // There are two different ways to register your own TestPartResultReporter. + // You can register your own repoter to listen either only for test results + // from the current thread or for results from all threads. + // By default, each per-thread test result repoter just passes a new + // TestPartResult to the global test result reporter, which registers the + // test part result for the currently running test. + + // Returns the global test part result reporter. + TestPartResultReporterInterface* GetGlobalTestPartResultReporter(); + + // Sets the global test part result reporter. + void SetGlobalTestPartResultReporter( + TestPartResultReporterInterface* reporter); + + // Returns the test part result reporter for the current thread. + TestPartResultReporterInterface* GetTestPartResultReporterForCurrentThread(); + + // Sets the test part result reporter for the current thread. + void SetTestPartResultReporterForCurrentThread( + TestPartResultReporterInterface* reporter); + + // Gets the number of successful test suites. + int successful_test_suite_count() const; + + // Gets the number of failed test suites. + int failed_test_suite_count() const; + + // Gets the number of all test suites. + int total_test_suite_count() const; + + // Gets the number of all test suites that contain at least one test + // that should run. + int test_suite_to_run_count() const; + + // Gets the number of successful tests. + int successful_test_count() const; + + // Gets the number of skipped tests. + int skipped_test_count() const; + + // Gets the number of failed tests. + int failed_test_count() const; + + // Gets the number of disabled tests that will be reported in the XML report. + int reportable_disabled_test_count() const; + + // Gets the number of disabled tests. + int disabled_test_count() const; + + // Gets the number of tests to be printed in the XML report. + int reportable_test_count() const; + + // Gets the number of all tests. + int total_test_count() const; + + // Gets the number of tests that should run. + int test_to_run_count() const; + + // Gets the time of the test program start, in ms from the start of the + // UNIX epoch. + TimeInMillis start_timestamp() const { return start_timestamp_; } + + // Gets the elapsed time, in milliseconds. + TimeInMillis elapsed_time() const { return elapsed_time_; } + + // Returns true iff the unit test passed (i.e. all test suites passed). + bool Passed() const { return !Failed(); } + + // Returns true iff the unit test failed (i.e. some test suite failed + // or something outside of all tests failed). + bool Failed() const { + return failed_test_suite_count() > 0 || ad_hoc_test_result()->Failed(); + } + + // Gets the i-th test suite among all the test suites. i can range from 0 to + // total_test_suite_count() - 1. If i is not in that range, returns NULL. + const TestSuite* GetTestSuite(int i) const { + const int index = GetElementOr(test_suite_indices_, i, -1); + return index < 0 ? nullptr : test_suites_[static_cast(i)]; + } + + // Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + const TestCase* GetTestCase(int i) const { return GetTestSuite(i); } +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Gets the i-th test suite among all the test suites. i can range from 0 to + // total_test_suite_count() - 1. If i is not in that range, returns NULL. + TestSuite* GetMutableSuiteCase(int i) { + const int index = GetElementOr(test_suite_indices_, i, -1); + return index < 0 ? nullptr : test_suites_[static_cast(index)]; + } + + // Provides access to the event listener list. + TestEventListeners* listeners() { return &listeners_; } + + // Returns the TestResult for the test that's currently running, or + // the TestResult for the ad hoc test if no test is running. + TestResult* current_test_result(); + + // Returns the TestResult for the ad hoc test. + const TestResult* ad_hoc_test_result() const { return &ad_hoc_test_result_; } + + // Sets the OS stack trace getter. + // + // Does nothing if the input and the current OS stack trace getter + // are the same; otherwise, deletes the old getter and makes the + // input the current getter. + void set_os_stack_trace_getter(OsStackTraceGetterInterface* getter); + + // Returns the current OS stack trace getter if it is not NULL; + // otherwise, creates an OsStackTraceGetter, makes it the current + // getter, and returns it. + OsStackTraceGetterInterface* os_stack_trace_getter(); + + // Returns the current OS stack trace as an std::string. + // + // The maximum number of stack frames to be included is specified by + // the gtest_stack_trace_depth flag. The skip_count parameter + // specifies the number of top frames to be skipped, which doesn't + // count against the number of frames to be included. + // + // For example, if Foo() calls Bar(), which in turn calls + // CurrentOsStackTraceExceptTop(1), Foo() will be included in the + // trace but Bar() and CurrentOsStackTraceExceptTop() won't. + std::string CurrentOsStackTraceExceptTop(int skip_count) GTEST_NO_INLINE_; + + // Finds and returns a TestSuite with the given name. If one doesn't + // exist, creates one and returns it. + // + // Arguments: + // + // test_suite_name: name of the test suite + // type_param: the name of the test's type parameter, or NULL if + // this is not a typed or a type-parameterized test. + // set_up_tc: pointer to the function that sets up the test suite + // tear_down_tc: pointer to the function that tears down the test suite + TestSuite* GetTestSuite(const char* test_suite_name, const char* type_param, + internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc); + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + TestCase* GetTestCase(const char* test_case_name, const char* type_param, + internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc) { + return GetTestSuite(test_case_name, type_param, set_up_tc, tear_down_tc); + } +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + + // Adds a TestInfo to the unit test. + // + // Arguments: + // + // set_up_tc: pointer to the function that sets up the test suite + // tear_down_tc: pointer to the function that tears down the test suite + // test_info: the TestInfo object + void AddTestInfo(internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc, + TestInfo* test_info) { + // In order to support thread-safe death tests, we need to + // remember the original working directory when the test program + // was first invoked. We cannot do this in RUN_ALL_TESTS(), as + // the user may have changed the current directory before calling + // RUN_ALL_TESTS(). Therefore we capture the current directory in + // AddTestInfo(), which is called to register a TEST or TEST_F + // before main() is reached. + if (original_working_dir_.IsEmpty()) { + original_working_dir_.Set(FilePath::GetCurrentDir()); + GTEST_CHECK_(!original_working_dir_.IsEmpty()) + << "Failed to get the current working directory."; + } + + GetTestSuite(test_info->test_suite_name(), test_info->type_param(), + set_up_tc, tear_down_tc) + ->AddTestInfo(test_info); + } + + // Returns ParameterizedTestSuiteRegistry object used to keep track of + // value-parameterized tests and instantiate and register them. + internal::ParameterizedTestSuiteRegistry& parameterized_test_registry() { + return parameterized_test_registry_; + } + + // Sets the TestSuite object for the test that's currently running. + void set_current_test_suite(TestSuite* a_current_test_suite) { + current_test_suite_ = a_current_test_suite; + } + + // Sets the TestInfo object for the test that's currently running. If + // current_test_info is NULL, the assertion results will be stored in + // ad_hoc_test_result_. + void set_current_test_info(TestInfo* a_current_test_info) { + current_test_info_ = a_current_test_info; + } + + // Registers all parameterized tests defined using TEST_P and + // INSTANTIATE_TEST_SUITE_P, creating regular tests for each test/parameter + // combination. This method can be called more then once; it has guards + // protecting from registering the tests more then once. If + // value-parameterized tests are disabled, RegisterParameterizedTests is + // present but does nothing. + void RegisterParameterizedTests(); + + // Runs all tests in this UnitTest object, prints the result, and + // returns true if all tests are successful. If any exception is + // thrown during a test, this test is considered to be failed, but + // the rest of the tests will still be run. + bool RunAllTests(); + + // Clears the results of all tests, except the ad hoc tests. + void ClearNonAdHocTestResult() { + ForEach(test_suites_, TestSuite::ClearTestSuiteResult); + } + + // Clears the results of ad-hoc test assertions. + void ClearAdHocTestResult() { + ad_hoc_test_result_.Clear(); + } + + // Adds a TestProperty to the current TestResult object when invoked in a + // context of a test or a test suite, or to the global property set. If the + // result already contains a property with the same key, the value will be + // updated. + void RecordProperty(const TestProperty& test_property); + + enum ReactionToSharding { + HONOR_SHARDING_PROTOCOL, + IGNORE_SHARDING_PROTOCOL + }; + + // Matches the full name of each test against the user-specified + // filter to decide whether the test should run, then records the + // result in each TestSuite and TestInfo object. + // If shard_tests == HONOR_SHARDING_PROTOCOL, further filters tests + // based on sharding variables in the environment. + // Returns the number of tests that should run. + int FilterTests(ReactionToSharding shard_tests); + + // Prints the names of the tests matching the user-specified filter flag. + void ListTestsMatchingFilter(); + + const TestSuite* current_test_suite() const { return current_test_suite_; } + TestInfo* current_test_info() { return current_test_info_; } + const TestInfo* current_test_info() const { return current_test_info_; } + + // Returns the vector of environments that need to be set-up/torn-down + // before/after the tests are run. + std::vector& environments() { return environments_; } + + // Getters for the per-thread Google Test trace stack. + std::vector& gtest_trace_stack() { + return *(gtest_trace_stack_.pointer()); + } + const std::vector& gtest_trace_stack() const { + return gtest_trace_stack_.get(); + } + +#if GTEST_HAS_DEATH_TEST + void InitDeathTestSubprocessControlInfo() { + internal_run_death_test_flag_.reset(ParseInternalRunDeathTestFlag()); + } + // Returns a pointer to the parsed --gtest_internal_run_death_test + // flag, or NULL if that flag was not specified. + // This information is useful only in a death test child process. + // Must not be called before a call to InitGoogleTest. + const InternalRunDeathTestFlag* internal_run_death_test_flag() const { + return internal_run_death_test_flag_.get(); + } + + // Returns a pointer to the current death test factory. + internal::DeathTestFactory* death_test_factory() { + return death_test_factory_.get(); + } + + void SuppressTestEventsIfInSubprocess(); + + friend class ReplaceDeathTestFactory; +#endif // GTEST_HAS_DEATH_TEST + + // Initializes the event listener performing XML output as specified by + // UnitTestOptions. Must not be called before InitGoogleTest. + void ConfigureXmlOutput(); + +#if GTEST_CAN_STREAM_RESULTS_ + // Initializes the event listener for streaming test results to a socket. + // Must not be called before InitGoogleTest. + void ConfigureStreamingOutput(); +#endif + + // Performs initialization dependent upon flag values obtained in + // ParseGoogleTestFlagsOnly. Is called from InitGoogleTest after the call to + // ParseGoogleTestFlagsOnly. In case a user neglects to call InitGoogleTest + // this function is also called from RunAllTests. Since this function can be + // called more than once, it has to be idempotent. + void PostFlagParsingInit(); + + // Gets the random seed used at the start of the current test iteration. + int random_seed() const { return random_seed_; } + + // Gets the random number generator. + internal::Random* random() { return &random_; } + + // Shuffles all test suites, and the tests within each test suite, + // making sure that death tests are still run first. + void ShuffleTests(); + + // Restores the test suites and tests to their order before the first shuffle. + void UnshuffleTests(); + + // Returns the value of GTEST_FLAG(catch_exceptions) at the moment + // UnitTest::Run() starts. + bool catch_exceptions() const { return catch_exceptions_; } + + private: + friend class ::testing::UnitTest; + + // Used by UnitTest::Run() to capture the state of + // GTEST_FLAG(catch_exceptions) at the moment it starts. + void set_catch_exceptions(bool value) { catch_exceptions_ = value; } + + // The UnitTest object that owns this implementation object. + UnitTest* const parent_; + + // The working directory when the first TEST() or TEST_F() was + // executed. + internal::FilePath original_working_dir_; + + // The default test part result reporters. + DefaultGlobalTestPartResultReporter default_global_test_part_result_reporter_; + DefaultPerThreadTestPartResultReporter + default_per_thread_test_part_result_reporter_; + + // Points to (but doesn't own) the global test part result reporter. + TestPartResultReporterInterface* global_test_part_result_repoter_; + + // Protects read and write access to global_test_part_result_reporter_. + internal::Mutex global_test_part_result_reporter_mutex_; + + // Points to (but doesn't own) the per-thread test part result reporter. + internal::ThreadLocal + per_thread_test_part_result_reporter_; + + // The vector of environments that need to be set-up/torn-down + // before/after the tests are run. + std::vector environments_; + + // The vector of TestSuites in their original order. It owns the + // elements in the vector. + std::vector test_suites_; + + // Provides a level of indirection for the test suite list to allow + // easy shuffling and restoring the test suite order. The i-th + // element of this vector is the index of the i-th test suite in the + // shuffled order. + std::vector test_suite_indices_; + + // ParameterizedTestRegistry object used to register value-parameterized + // tests. + internal::ParameterizedTestSuiteRegistry parameterized_test_registry_; + + // Indicates whether RegisterParameterizedTests() has been called already. + bool parameterized_tests_registered_; + + // Index of the last death test suite registered. Initially -1. + int last_death_test_suite_; + + // This points to the TestSuite for the currently running test. It + // changes as Google Test goes through one test suite after another. + // When no test is running, this is set to NULL and Google Test + // stores assertion results in ad_hoc_test_result_. Initially NULL. + TestSuite* current_test_suite_; + + // This points to the TestInfo for the currently running test. It + // changes as Google Test goes through one test after another. When + // no test is running, this is set to NULL and Google Test stores + // assertion results in ad_hoc_test_result_. Initially NULL. + TestInfo* current_test_info_; + + // Normally, a user only writes assertions inside a TEST or TEST_F, + // or inside a function called by a TEST or TEST_F. Since Google + // Test keeps track of which test is current running, it can + // associate such an assertion with the test it belongs to. + // + // If an assertion is encountered when no TEST or TEST_F is running, + // Google Test attributes the assertion result to an imaginary "ad hoc" + // test, and records the result in ad_hoc_test_result_. + TestResult ad_hoc_test_result_; + + // The list of event listeners that can be used to track events inside + // Google Test. + TestEventListeners listeners_; + + // The OS stack trace getter. Will be deleted when the UnitTest + // object is destructed. By default, an OsStackTraceGetter is used, + // but the user can set this field to use a custom getter if that is + // desired. + OsStackTraceGetterInterface* os_stack_trace_getter_; + + // True iff PostFlagParsingInit() has been called. + bool post_flag_parse_init_performed_; + + // The random number seed used at the beginning of the test run. + int random_seed_; + + // Our random number generator. + internal::Random random_; + + // The time of the test program start, in ms from the start of the + // UNIX epoch. + TimeInMillis start_timestamp_; + + // How long the test took to run, in milliseconds. + TimeInMillis elapsed_time_; + +#if GTEST_HAS_DEATH_TEST + // The decomposed components of the gtest_internal_run_death_test flag, + // parsed when RUN_ALL_TESTS is called. + std::unique_ptr internal_run_death_test_flag_; + std::unique_ptr death_test_factory_; +#endif // GTEST_HAS_DEATH_TEST + + // A per-thread stack of traces created by the SCOPED_TRACE() macro. + internal::ThreadLocal > gtest_trace_stack_; + + // The value of GTEST_FLAG(catch_exceptions) at the moment RunAllTests() + // starts. + bool catch_exceptions_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTestImpl); +}; // class UnitTestImpl + +// Convenience function for accessing the global UnitTest +// implementation object. +inline UnitTestImpl* GetUnitTestImpl() { + return UnitTest::GetInstance()->impl(); +} + +#if GTEST_USES_SIMPLE_RE + +// Internal helper functions for implementing the simple regular +// expression matcher. +GTEST_API_ bool IsInSet(char ch, const char* str); +GTEST_API_ bool IsAsciiDigit(char ch); +GTEST_API_ bool IsAsciiPunct(char ch); +GTEST_API_ bool IsRepeat(char ch); +GTEST_API_ bool IsAsciiWhiteSpace(char ch); +GTEST_API_ bool IsAsciiWordChar(char ch); +GTEST_API_ bool IsValidEscape(char ch); +GTEST_API_ bool AtomMatchesChar(bool escaped, char pattern, char ch); +GTEST_API_ bool ValidateRegex(const char* regex); +GTEST_API_ bool MatchRegexAtHead(const char* regex, const char* str); +GTEST_API_ bool MatchRepetitionAndRegexAtHead( + bool escaped, char ch, char repeat, const char* regex, const char* str); +GTEST_API_ bool MatchRegexAnywhere(const char* regex, const char* str); + +#endif // GTEST_USES_SIMPLE_RE + +// Parses the command line for Google Test flags, without initializing +// other parts of Google Test. +GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, char** argv); +GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv); + +#if GTEST_HAS_DEATH_TEST + +// Returns the message describing the last system error, regardless of the +// platform. +GTEST_API_ std::string GetLastErrnoDescription(); + +// Attempts to parse a string into a positive integer pointed to by the +// number parameter. Returns true if that is possible. +// GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we can use +// it here. +template +bool ParseNaturalNumber(const ::std::string& str, Integer* number) { + // Fail fast if the given string does not begin with a digit; + // this bypasses strtoXXX's "optional leading whitespace and plus + // or minus sign" semantics, which are undesirable here. + if (str.empty() || !IsDigit(str[0])) { + return false; + } + errno = 0; + + char* end; + // BiggestConvertible is the largest integer type that system-provided + // string-to-number conversion routines can return. + +# if GTEST_OS_WINDOWS && !defined(__GNUC__) + + // MSVC and C++ Builder define __int64 instead of the standard long long. + typedef unsigned __int64 BiggestConvertible; + const BiggestConvertible parsed = _strtoui64(str.c_str(), &end, 10); + +# else + + typedef unsigned long long BiggestConvertible; // NOLINT + const BiggestConvertible parsed = strtoull(str.c_str(), &end, 10); + +# endif // GTEST_OS_WINDOWS && !defined(__GNUC__) + + const bool parse_success = *end == '\0' && errno == 0; + + GTEST_CHECK_(sizeof(Integer) <= sizeof(parsed)); + + const Integer result = static_cast(parsed); + if (parse_success && static_cast(result) == parsed) { + *number = result; + return true; + } + return false; +} +#endif // GTEST_HAS_DEATH_TEST + +// TestResult contains some private methods that should be hidden from +// Google Test user but are required for testing. This class allow our tests +// to access them. +// +// This class is supplied only for the purpose of testing Google Test's own +// constructs. Do not use it in user tests, either directly or indirectly. +class TestResultAccessor { + public: + static void RecordProperty(TestResult* test_result, + const std::string& xml_element, + const TestProperty& property) { + test_result->RecordProperty(xml_element, property); + } + + static void ClearTestPartResults(TestResult* test_result) { + test_result->ClearTestPartResults(); + } + + static const std::vector& test_part_results( + const TestResult& test_result) { + return test_result.test_part_results(); + } +}; + +#if GTEST_CAN_STREAM_RESULTS_ + +// Streams test results to the given port on the given host machine. +class StreamingListener : public EmptyTestEventListener { + public: + // Abstract base class for writing strings to a socket. + class AbstractSocketWriter { + public: + virtual ~AbstractSocketWriter() {} + + // Sends a string to the socket. + virtual void Send(const std::string& message) = 0; + + // Closes the socket. + virtual void CloseConnection() {} + + // Sends a string and a newline to the socket. + void SendLn(const std::string& message) { Send(message + "\n"); } + }; + + // Concrete class for actually writing strings to a socket. + class SocketWriter : public AbstractSocketWriter { + public: + SocketWriter(const std::string& host, const std::string& port) + : sockfd_(-1), host_name_(host), port_num_(port) { + MakeConnection(); + } + + ~SocketWriter() override { + if (sockfd_ != -1) + CloseConnection(); + } + + // Sends a string to the socket. + void Send(const std::string& message) override { + GTEST_CHECK_(sockfd_ != -1) + << "Send() can be called only when there is a connection."; + + const auto len = static_cast(message.length()); + if (write(sockfd_, message.c_str(), len) != static_cast(len)) { + GTEST_LOG_(WARNING) + << "stream_result_to: failed to stream to " + << host_name_ << ":" << port_num_; + } + } + + private: + // Creates a client socket and connects to the server. + void MakeConnection(); + + // Closes the socket. + void CloseConnection() override { + GTEST_CHECK_(sockfd_ != -1) + << "CloseConnection() can be called only when there is a connection."; + + close(sockfd_); + sockfd_ = -1; + } + + int sockfd_; // socket file descriptor + const std::string host_name_; + const std::string port_num_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(SocketWriter); + }; // class SocketWriter + + // Escapes '=', '&', '%', and '\n' characters in str as "%xx". + static std::string UrlEncode(const char* str); + + StreamingListener(const std::string& host, const std::string& port) + : socket_writer_(new SocketWriter(host, port)) { + Start(); + } + + explicit StreamingListener(AbstractSocketWriter* socket_writer) + : socket_writer_(socket_writer) { Start(); } + + void OnTestProgramStart(const UnitTest& /* unit_test */) override { + SendLn("event=TestProgramStart"); + } + + void OnTestProgramEnd(const UnitTest& unit_test) override { + // Note that Google Test current only report elapsed time for each + // test iteration, not for the entire test program. + SendLn("event=TestProgramEnd&passed=" + FormatBool(unit_test.Passed())); + + // Notify the streaming server to stop. + socket_writer_->CloseConnection(); + } + + void OnTestIterationStart(const UnitTest& /* unit_test */, + int iteration) override { + SendLn("event=TestIterationStart&iteration=" + + StreamableToString(iteration)); + } + + void OnTestIterationEnd(const UnitTest& unit_test, + int /* iteration */) override { + SendLn("event=TestIterationEnd&passed=" + + FormatBool(unit_test.Passed()) + "&elapsed_time=" + + StreamableToString(unit_test.elapsed_time()) + "ms"); + } + + // Note that "event=TestCaseStart" is a wire format and has to remain + // "case" for compatibilty + void OnTestCaseStart(const TestCase& test_case) override { + SendLn(std::string("event=TestCaseStart&name=") + test_case.name()); + } + + // Note that "event=TestCaseEnd" is a wire format and has to remain + // "case" for compatibilty + void OnTestCaseEnd(const TestCase& test_case) override { + SendLn("event=TestCaseEnd&passed=" + FormatBool(test_case.Passed()) + + "&elapsed_time=" + StreamableToString(test_case.elapsed_time()) + + "ms"); + } + + void OnTestStart(const TestInfo& test_info) override { + SendLn(std::string("event=TestStart&name=") + test_info.name()); + } + + void OnTestEnd(const TestInfo& test_info) override { + SendLn("event=TestEnd&passed=" + + FormatBool((test_info.result())->Passed()) + + "&elapsed_time=" + + StreamableToString((test_info.result())->elapsed_time()) + "ms"); + } + + void OnTestPartResult(const TestPartResult& test_part_result) override { + const char* file_name = test_part_result.file_name(); + if (file_name == nullptr) file_name = ""; + SendLn("event=TestPartResult&file=" + UrlEncode(file_name) + + "&line=" + StreamableToString(test_part_result.line_number()) + + "&message=" + UrlEncode(test_part_result.message())); + } + + private: + // Sends the given message and a newline to the socket. + void SendLn(const std::string& message) { socket_writer_->SendLn(message); } + + // Called at the start of streaming to notify the receiver what + // protocol we are using. + void Start() { SendLn("gtest_streaming_protocol_version=1.0"); } + + std::string FormatBool(bool value) { return value ? "1" : "0"; } + + const std::unique_ptr socket_writer_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(StreamingListener); +}; // class StreamingListener + +#endif // GTEST_CAN_STREAM_RESULTS_ + +} // namespace internal +} // namespace testing + +GTEST_DISABLE_MSC_WARNINGS_POP_() // 4251 + +#endif // GTEST_SRC_GTEST_INTERNAL_INL_H_ diff --git a/deps/googletest/src/gtest-matchers.cc b/deps/googletest/src/gtest-matchers.cc new file mode 100644 index 000000000..7d2fb6851 --- /dev/null +++ b/deps/googletest/src/gtest-matchers.cc @@ -0,0 +1,97 @@ +// Copyright 2007, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// The Google C++ Testing and Mocking Framework (Google Test) +// +// This file implements just enough of the matcher interface to allow +// EXPECT_DEATH and friends to accept a matcher argument. + +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-port.h" +#include "gtest/gtest-matchers.h" + +#include + +namespace testing { + +// Constructs a matcher that matches a const std::string& whose value is +// equal to s. +Matcher::Matcher(const std::string& s) { *this = Eq(s); } + +// Constructs a matcher that matches a const std::string& whose value is +// equal to s. +Matcher::Matcher(const char* s) { + *this = Eq(std::string(s)); +} + +// Constructs a matcher that matches a std::string whose value is equal to +// s. +Matcher::Matcher(const std::string& s) { *this = Eq(s); } + +// Constructs a matcher that matches a std::string whose value is equal to +// s. +Matcher::Matcher(const char* s) { *this = Eq(std::string(s)); } + +#if GTEST_HAS_ABSL +// Constructs a matcher that matches a const absl::string_view& whose value is +// equal to s. +Matcher::Matcher(const std::string& s) { + *this = Eq(s); +} + +// Constructs a matcher that matches a const absl::string_view& whose value is +// equal to s. +Matcher::Matcher(const char* s) { + *this = Eq(std::string(s)); +} + +// Constructs a matcher that matches a const absl::string_view& whose value is +// equal to s. +Matcher::Matcher(absl::string_view s) { + *this = Eq(std::string(s)); +} + +// Constructs a matcher that matches a absl::string_view whose value is equal to +// s. +Matcher::Matcher(const std::string& s) { *this = Eq(s); } + +// Constructs a matcher that matches a absl::string_view whose value is equal to +// s. +Matcher::Matcher(const char* s) { + *this = Eq(std::string(s)); +} + +// Constructs a matcher that matches a absl::string_view whose value is equal to +// s. +Matcher::Matcher(absl::string_view s) { + *this = Eq(std::string(s)); +} +#endif // GTEST_HAS_ABSL + +} // namespace testing diff --git a/deps/googletest/src/gtest-port.cc b/deps/googletest/src/gtest-port.cc new file mode 100644 index 000000000..2cba2693e --- /dev/null +++ b/deps/googletest/src/gtest-port.cc @@ -0,0 +1,1404 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +#include "gtest/internal/gtest-port.h" + +#include +#include +#include +#include +#include +#include + +#if GTEST_OS_WINDOWS +# include +# include +# include +# include // Used in ThreadLocal. +# ifdef _MSC_VER +# include +# endif // _MSC_VER +#else +# include +#endif // GTEST_OS_WINDOWS + +#if GTEST_OS_MAC +# include +# include +# include +#endif // GTEST_OS_MAC + +#if GTEST_OS_DRAGONFLY || GTEST_OS_FREEBSD || GTEST_OS_GNU_KFREEBSD || \ + GTEST_OS_NETBSD || GTEST_OS_OPENBSD +# include +# if GTEST_OS_DRAGONFLY || GTEST_OS_FREEBSD || GTEST_OS_GNU_KFREEBSD +# include +# endif +#endif + +#if GTEST_OS_QNX +# include +# include +# include +#endif // GTEST_OS_QNX + +#if GTEST_OS_AIX +# include +# include +#endif // GTEST_OS_AIX + +#if GTEST_OS_FUCHSIA +# include +# include +#endif // GTEST_OS_FUCHSIA + +#include "gtest/gtest-spi.h" +#include "gtest/gtest-message.h" +#include "gtest/internal/gtest-internal.h" +#include "gtest/internal/gtest-string.h" +#include "src/gtest-internal-inl.h" + +namespace testing { +namespace internal { + +#if defined(_MSC_VER) || defined(__BORLANDC__) +// MSVC and C++Builder do not provide a definition of STDERR_FILENO. +const int kStdOutFileno = 1; +const int kStdErrFileno = 2; +#else +const int kStdOutFileno = STDOUT_FILENO; +const int kStdErrFileno = STDERR_FILENO; +#endif // _MSC_VER + +#if GTEST_OS_LINUX + +namespace { +template +T ReadProcFileField(const std::string& filename, int field) { + std::string dummy; + std::ifstream file(filename.c_str()); + while (field-- > 0) { + file >> dummy; + } + T output = 0; + file >> output; + return output; +} +} // namespace + +// Returns the number of active threads, or 0 when there is an error. +size_t GetThreadCount() { + const std::string filename = + (Message() << "/proc/" << getpid() << "/stat").GetString(); + return ReadProcFileField(filename, 19); +} + +#elif GTEST_OS_MAC + +size_t GetThreadCount() { + const task_t task = mach_task_self(); + mach_msg_type_number_t thread_count; + thread_act_array_t thread_list; + const kern_return_t status = task_threads(task, &thread_list, &thread_count); + if (status == KERN_SUCCESS) { + // task_threads allocates resources in thread_list and we need to free them + // to avoid leaks. + vm_deallocate(task, + reinterpret_cast(thread_list), + sizeof(thread_t) * thread_count); + return static_cast(thread_count); + } else { + return 0; + } +} + +#elif GTEST_OS_DRAGONFLY || GTEST_OS_FREEBSD || GTEST_OS_GNU_KFREEBSD || \ + GTEST_OS_NETBSD + +#if GTEST_OS_NETBSD +#undef KERN_PROC +#define KERN_PROC KERN_PROC2 +#define kinfo_proc kinfo_proc2 +#endif + +#if GTEST_OS_DRAGONFLY +#define KP_NLWP(kp) (kp.kp_nthreads) +#elif GTEST_OS_FREEBSD || GTEST_OS_GNU_KFREEBSD +#define KP_NLWP(kp) (kp.ki_numthreads) +#elif GTEST_OS_NETBSD +#define KP_NLWP(kp) (kp.p_nlwps) +#endif + +// Returns the number of threads running in the process, or 0 to indicate that +// we cannot detect it. +size_t GetThreadCount() { + int mib[] = { + CTL_KERN, + KERN_PROC, + KERN_PROC_PID, + getpid(), +#if GTEST_OS_NETBSD + sizeof(struct kinfo_proc), + 1, +#endif + }; + u_int miblen = sizeof(mib) / sizeof(mib[0]); + struct kinfo_proc info; + size_t size = sizeof(info); + if (sysctl(mib, miblen, &info, &size, NULL, 0)) { + return 0; + } + return static_cast(KP_NLWP(info)); +} +#elif GTEST_OS_OPENBSD + +// Returns the number of threads running in the process, or 0 to indicate that +// we cannot detect it. +size_t GetThreadCount() { + int mib[] = { + CTL_KERN, + KERN_PROC, + KERN_PROC_PID | KERN_PROC_SHOW_THREADS, + getpid(), + sizeof(struct kinfo_proc), + 0, + }; + u_int miblen = sizeof(mib) / sizeof(mib[0]); + + // get number of structs + size_t size; + if (sysctl(mib, miblen, NULL, &size, NULL, 0)) { + return 0; + } + mib[5] = size / mib[4]; + + // populate array of structs + struct kinfo_proc info[mib[5]]; + if (sysctl(mib, miblen, &info, &size, NULL, 0)) { + return 0; + } + + // exclude empty members + int nthreads = 0; + for (int i = 0; i < size / mib[4]; i++) { + if (info[i].p_tid != -1) + nthreads++; + } + return nthreads; +} + +#elif GTEST_OS_QNX + +// Returns the number of threads running in the process, or 0 to indicate that +// we cannot detect it. +size_t GetThreadCount() { + const int fd = open("/proc/self/as", O_RDONLY); + if (fd < 0) { + return 0; + } + procfs_info process_info; + const int status = + devctl(fd, DCMD_PROC_INFO, &process_info, sizeof(process_info), nullptr); + close(fd); + if (status == EOK) { + return static_cast(process_info.num_threads); + } else { + return 0; + } +} + +#elif GTEST_OS_AIX + +size_t GetThreadCount() { + struct procentry64 entry; + pid_t pid = getpid(); + int status = getprocs64(&entry, sizeof(entry), nullptr, 0, &pid, 1); + if (status == 1) { + return entry.pi_thcount; + } else { + return 0; + } +} + +#elif GTEST_OS_FUCHSIA + +size_t GetThreadCount() { + int dummy_buffer; + size_t avail; + zx_status_t status = zx_object_get_info( + zx_process_self(), + ZX_INFO_PROCESS_THREADS, + &dummy_buffer, + 0, + nullptr, + &avail); + if (status == ZX_OK) { + return avail; + } else { + return 0; + } +} + +#else + +size_t GetThreadCount() { + // There's no portable way to detect the number of threads, so we just + // return 0 to indicate that we cannot detect it. + return 0; +} + +#endif // GTEST_OS_LINUX + +#if GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS + +void SleepMilliseconds(int n) { + ::Sleep(n); +} + +AutoHandle::AutoHandle() + : handle_(INVALID_HANDLE_VALUE) {} + +AutoHandle::AutoHandle(Handle handle) + : handle_(handle) {} + +AutoHandle::~AutoHandle() { + Reset(); +} + +AutoHandle::Handle AutoHandle::Get() const { + return handle_; +} + +void AutoHandle::Reset() { + Reset(INVALID_HANDLE_VALUE); +} + +void AutoHandle::Reset(HANDLE handle) { + // Resetting with the same handle we already own is invalid. + if (handle_ != handle) { + if (IsCloseable()) { + ::CloseHandle(handle_); + } + handle_ = handle; + } else { + GTEST_CHECK_(!IsCloseable()) + << "Resetting a valid handle to itself is likely a programmer error " + "and thus not allowed."; + } +} + +bool AutoHandle::IsCloseable() const { + // Different Windows APIs may use either of these values to represent an + // invalid handle. + return handle_ != nullptr && handle_ != INVALID_HANDLE_VALUE; +} + +Notification::Notification() + : event_(::CreateEvent(nullptr, // Default security attributes. + TRUE, // Do not reset automatically. + FALSE, // Initially unset. + nullptr)) { // Anonymous event. + GTEST_CHECK_(event_.Get() != nullptr); +} + +void Notification::Notify() { + GTEST_CHECK_(::SetEvent(event_.Get()) != FALSE); +} + +void Notification::WaitForNotification() { + GTEST_CHECK_( + ::WaitForSingleObject(event_.Get(), INFINITE) == WAIT_OBJECT_0); +} + +Mutex::Mutex() + : owner_thread_id_(0), + type_(kDynamic), + critical_section_init_phase_(0), + critical_section_(new CRITICAL_SECTION) { + ::InitializeCriticalSection(critical_section_); +} + +Mutex::~Mutex() { + // Static mutexes are leaked intentionally. It is not thread-safe to try + // to clean them up. + if (type_ == kDynamic) { + ::DeleteCriticalSection(critical_section_); + delete critical_section_; + critical_section_ = nullptr; + } +} + +void Mutex::Lock() { + ThreadSafeLazyInit(); + ::EnterCriticalSection(critical_section_); + owner_thread_id_ = ::GetCurrentThreadId(); +} + +void Mutex::Unlock() { + ThreadSafeLazyInit(); + // We don't protect writing to owner_thread_id_ here, as it's the + // caller's responsibility to ensure that the current thread holds the + // mutex when this is called. + owner_thread_id_ = 0; + ::LeaveCriticalSection(critical_section_); +} + +// Does nothing if the current thread holds the mutex. Otherwise, crashes +// with high probability. +void Mutex::AssertHeld() { + ThreadSafeLazyInit(); + GTEST_CHECK_(owner_thread_id_ == ::GetCurrentThreadId()) + << "The current thread is not holding the mutex @" << this; +} + +namespace { + +#ifdef _MSC_VER +// Use the RAII idiom to flag mem allocs that are intentionally never +// deallocated. The motivation is to silence the false positive mem leaks +// that are reported by the debug version of MS's CRT which can only detect +// if an alloc is missing a matching deallocation. +// Example: +// MemoryIsNotDeallocated memory_is_not_deallocated; +// critical_section_ = new CRITICAL_SECTION; +// +class MemoryIsNotDeallocated +{ + public: + MemoryIsNotDeallocated() : old_crtdbg_flag_(0) { + old_crtdbg_flag_ = _CrtSetDbgFlag(_CRTDBG_REPORT_FLAG); + // Set heap allocation block type to _IGNORE_BLOCK so that MS debug CRT + // doesn't report mem leak if there's no matching deallocation. + _CrtSetDbgFlag(old_crtdbg_flag_ & ~_CRTDBG_ALLOC_MEM_DF); + } + + ~MemoryIsNotDeallocated() { + // Restore the original _CRTDBG_ALLOC_MEM_DF flag + _CrtSetDbgFlag(old_crtdbg_flag_); + } + + private: + int old_crtdbg_flag_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(MemoryIsNotDeallocated); +}; +#endif // _MSC_VER + +} // namespace + +// Initializes owner_thread_id_ and critical_section_ in static mutexes. +void Mutex::ThreadSafeLazyInit() { + // Dynamic mutexes are initialized in the constructor. + if (type_ == kStatic) { + switch ( + ::InterlockedCompareExchange(&critical_section_init_phase_, 1L, 0L)) { + case 0: + // If critical_section_init_phase_ was 0 before the exchange, we + // are the first to test it and need to perform the initialization. + owner_thread_id_ = 0; + { + // Use RAII to flag that following mem alloc is never deallocated. +#ifdef _MSC_VER + MemoryIsNotDeallocated memory_is_not_deallocated; +#endif // _MSC_VER + critical_section_ = new CRITICAL_SECTION; + } + ::InitializeCriticalSection(critical_section_); + // Updates the critical_section_init_phase_ to 2 to signal + // initialization complete. + GTEST_CHECK_(::InterlockedCompareExchange( + &critical_section_init_phase_, 2L, 1L) == + 1L); + break; + case 1: + // Somebody else is already initializing the mutex; spin until they + // are done. + while (::InterlockedCompareExchange(&critical_section_init_phase_, + 2L, + 2L) != 2L) { + // Possibly yields the rest of the thread's time slice to other + // threads. + ::Sleep(0); + } + break; + + case 2: + break; // The mutex is already initialized and ready for use. + + default: + GTEST_CHECK_(false) + << "Unexpected value of critical_section_init_phase_ " + << "while initializing a static mutex."; + } + } +} + +namespace { + +class ThreadWithParamSupport : public ThreadWithParamBase { + public: + static HANDLE CreateThread(Runnable* runnable, + Notification* thread_can_start) { + ThreadMainParam* param = new ThreadMainParam(runnable, thread_can_start); + DWORD thread_id; + HANDLE thread_handle = ::CreateThread( + nullptr, // Default security. + 0, // Default stack size. + &ThreadWithParamSupport::ThreadMain, + param, // Parameter to ThreadMainStatic + 0x0, // Default creation flags. + &thread_id); // Need a valid pointer for the call to work under Win98. + GTEST_CHECK_(thread_handle != nullptr) + << "CreateThread failed with error " << ::GetLastError() << "."; + if (thread_handle == nullptr) { + delete param; + } + return thread_handle; + } + + private: + struct ThreadMainParam { + ThreadMainParam(Runnable* runnable, Notification* thread_can_start) + : runnable_(runnable), + thread_can_start_(thread_can_start) { + } + std::unique_ptr runnable_; + // Does not own. + Notification* thread_can_start_; + }; + + static DWORD WINAPI ThreadMain(void* ptr) { + // Transfers ownership. + std::unique_ptr param(static_cast(ptr)); + if (param->thread_can_start_ != nullptr) + param->thread_can_start_->WaitForNotification(); + param->runnable_->Run(); + return 0; + } + + // Prohibit instantiation. + ThreadWithParamSupport(); + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParamSupport); +}; + +} // namespace + +ThreadWithParamBase::ThreadWithParamBase(Runnable *runnable, + Notification* thread_can_start) + : thread_(ThreadWithParamSupport::CreateThread(runnable, + thread_can_start)) { +} + +ThreadWithParamBase::~ThreadWithParamBase() { + Join(); +} + +void ThreadWithParamBase::Join() { + GTEST_CHECK_(::WaitForSingleObject(thread_.Get(), INFINITE) == WAIT_OBJECT_0) + << "Failed to join the thread with error " << ::GetLastError() << "."; +} + +// Maps a thread to a set of ThreadIdToThreadLocals that have values +// instantiated on that thread and notifies them when the thread exits. A +// ThreadLocal instance is expected to persist until all threads it has +// values on have terminated. +class ThreadLocalRegistryImpl { + public: + // Registers thread_local_instance as having value on the current thread. + // Returns a value that can be used to identify the thread from other threads. + static ThreadLocalValueHolderBase* GetValueOnCurrentThread( + const ThreadLocalBase* thread_local_instance) { + DWORD current_thread = ::GetCurrentThreadId(); + MutexLock lock(&mutex_); + ThreadIdToThreadLocals* const thread_to_thread_locals = + GetThreadLocalsMapLocked(); + ThreadIdToThreadLocals::iterator thread_local_pos = + thread_to_thread_locals->find(current_thread); + if (thread_local_pos == thread_to_thread_locals->end()) { + thread_local_pos = thread_to_thread_locals->insert( + std::make_pair(current_thread, ThreadLocalValues())).first; + StartWatcherThreadFor(current_thread); + } + ThreadLocalValues& thread_local_values = thread_local_pos->second; + ThreadLocalValues::iterator value_pos = + thread_local_values.find(thread_local_instance); + if (value_pos == thread_local_values.end()) { + value_pos = + thread_local_values + .insert(std::make_pair( + thread_local_instance, + std::shared_ptr( + thread_local_instance->NewValueForCurrentThread()))) + .first; + } + return value_pos->second.get(); + } + + static void OnThreadLocalDestroyed( + const ThreadLocalBase* thread_local_instance) { + std::vector > value_holders; + // Clean up the ThreadLocalValues data structure while holding the lock, but + // defer the destruction of the ThreadLocalValueHolderBases. + { + MutexLock lock(&mutex_); + ThreadIdToThreadLocals* const thread_to_thread_locals = + GetThreadLocalsMapLocked(); + for (ThreadIdToThreadLocals::iterator it = + thread_to_thread_locals->begin(); + it != thread_to_thread_locals->end(); + ++it) { + ThreadLocalValues& thread_local_values = it->second; + ThreadLocalValues::iterator value_pos = + thread_local_values.find(thread_local_instance); + if (value_pos != thread_local_values.end()) { + value_holders.push_back(value_pos->second); + thread_local_values.erase(value_pos); + // This 'if' can only be successful at most once, so theoretically we + // could break out of the loop here, but we don't bother doing so. + } + } + } + // Outside the lock, let the destructor for 'value_holders' deallocate the + // ThreadLocalValueHolderBases. + } + + static void OnThreadExit(DWORD thread_id) { + GTEST_CHECK_(thread_id != 0) << ::GetLastError(); + std::vector > value_holders; + // Clean up the ThreadIdToThreadLocals data structure while holding the + // lock, but defer the destruction of the ThreadLocalValueHolderBases. + { + MutexLock lock(&mutex_); + ThreadIdToThreadLocals* const thread_to_thread_locals = + GetThreadLocalsMapLocked(); + ThreadIdToThreadLocals::iterator thread_local_pos = + thread_to_thread_locals->find(thread_id); + if (thread_local_pos != thread_to_thread_locals->end()) { + ThreadLocalValues& thread_local_values = thread_local_pos->second; + for (ThreadLocalValues::iterator value_pos = + thread_local_values.begin(); + value_pos != thread_local_values.end(); + ++value_pos) { + value_holders.push_back(value_pos->second); + } + thread_to_thread_locals->erase(thread_local_pos); + } + } + // Outside the lock, let the destructor for 'value_holders' deallocate the + // ThreadLocalValueHolderBases. + } + + private: + // In a particular thread, maps a ThreadLocal object to its value. + typedef std::map > + ThreadLocalValues; + // Stores all ThreadIdToThreadLocals having values in a thread, indexed by + // thread's ID. + typedef std::map ThreadIdToThreadLocals; + + // Holds the thread id and thread handle that we pass from + // StartWatcherThreadFor to WatcherThreadFunc. + typedef std::pair ThreadIdAndHandle; + + static void StartWatcherThreadFor(DWORD thread_id) { + // The returned handle will be kept in thread_map and closed by + // watcher_thread in WatcherThreadFunc. + HANDLE thread = ::OpenThread(SYNCHRONIZE | THREAD_QUERY_INFORMATION, + FALSE, + thread_id); + GTEST_CHECK_(thread != nullptr); + // We need to pass a valid thread ID pointer into CreateThread for it + // to work correctly under Win98. + DWORD watcher_thread_id; + HANDLE watcher_thread = ::CreateThread( + nullptr, // Default security. + 0, // Default stack size + &ThreadLocalRegistryImpl::WatcherThreadFunc, + reinterpret_cast(new ThreadIdAndHandle(thread_id, thread)), + CREATE_SUSPENDED, &watcher_thread_id); + GTEST_CHECK_(watcher_thread != nullptr); + // Give the watcher thread the same priority as ours to avoid being + // blocked by it. + ::SetThreadPriority(watcher_thread, + ::GetThreadPriority(::GetCurrentThread())); + ::ResumeThread(watcher_thread); + ::CloseHandle(watcher_thread); + } + + // Monitors exit from a given thread and notifies those + // ThreadIdToThreadLocals about thread termination. + static DWORD WINAPI WatcherThreadFunc(LPVOID param) { + const ThreadIdAndHandle* tah = + reinterpret_cast(param); + GTEST_CHECK_( + ::WaitForSingleObject(tah->second, INFINITE) == WAIT_OBJECT_0); + OnThreadExit(tah->first); + ::CloseHandle(tah->second); + delete tah; + return 0; + } + + // Returns map of thread local instances. + static ThreadIdToThreadLocals* GetThreadLocalsMapLocked() { + mutex_.AssertHeld(); +#ifdef _MSC_VER + MemoryIsNotDeallocated memory_is_not_deallocated; +#endif // _MSC_VER + static ThreadIdToThreadLocals* map = new ThreadIdToThreadLocals(); + return map; + } + + // Protects access to GetThreadLocalsMapLocked() and its return value. + static Mutex mutex_; + // Protects access to GetThreadMapLocked() and its return value. + static Mutex thread_map_mutex_; +}; + +Mutex ThreadLocalRegistryImpl::mutex_(Mutex::kStaticMutex); +Mutex ThreadLocalRegistryImpl::thread_map_mutex_(Mutex::kStaticMutex); + +ThreadLocalValueHolderBase* ThreadLocalRegistry::GetValueOnCurrentThread( + const ThreadLocalBase* thread_local_instance) { + return ThreadLocalRegistryImpl::GetValueOnCurrentThread( + thread_local_instance); +} + +void ThreadLocalRegistry::OnThreadLocalDestroyed( + const ThreadLocalBase* thread_local_instance) { + ThreadLocalRegistryImpl::OnThreadLocalDestroyed(thread_local_instance); +} + +#endif // GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS + +#if GTEST_USES_POSIX_RE + +// Implements RE. Currently only needed for death tests. + +RE::~RE() { + if (is_valid_) { + // regfree'ing an invalid regex might crash because the content + // of the regex is undefined. Since the regex's are essentially + // the same, one cannot be valid (or invalid) without the other + // being so too. + regfree(&partial_regex_); + regfree(&full_regex_); + } + free(const_cast(pattern_)); +} + +// Returns true iff regular expression re matches the entire str. +bool RE::FullMatch(const char* str, const RE& re) { + if (!re.is_valid_) return false; + + regmatch_t match; + return regexec(&re.full_regex_, str, 1, &match, 0) == 0; +} + +// Returns true iff regular expression re matches a substring of str +// (including str itself). +bool RE::PartialMatch(const char* str, const RE& re) { + if (!re.is_valid_) return false; + + regmatch_t match; + return regexec(&re.partial_regex_, str, 1, &match, 0) == 0; +} + +// Initializes an RE from its string representation. +void RE::Init(const char* regex) { + pattern_ = posix::StrDup(regex); + + // Reserves enough bytes to hold the regular expression used for a + // full match. + const size_t full_regex_len = strlen(regex) + 10; + char* const full_pattern = new char[full_regex_len]; + + snprintf(full_pattern, full_regex_len, "^(%s)$", regex); + is_valid_ = regcomp(&full_regex_, full_pattern, REG_EXTENDED) == 0; + // We want to call regcomp(&partial_regex_, ...) even if the + // previous expression returns false. Otherwise partial_regex_ may + // not be properly initialized can may cause trouble when it's + // freed. + // + // Some implementation of POSIX regex (e.g. on at least some + // versions of Cygwin) doesn't accept the empty string as a valid + // regex. We change it to an equivalent form "()" to be safe. + if (is_valid_) { + const char* const partial_regex = (*regex == '\0') ? "()" : regex; + is_valid_ = regcomp(&partial_regex_, partial_regex, REG_EXTENDED) == 0; + } + EXPECT_TRUE(is_valid_) + << "Regular expression \"" << regex + << "\" is not a valid POSIX Extended regular expression."; + + delete[] full_pattern; +} + +#elif GTEST_USES_SIMPLE_RE + +// Returns true iff ch appears anywhere in str (excluding the +// terminating '\0' character). +bool IsInSet(char ch, const char* str) { + return ch != '\0' && strchr(str, ch) != nullptr; +} + +// Returns true iff ch belongs to the given classification. Unlike +// similar functions in , these aren't affected by the +// current locale. +bool IsAsciiDigit(char ch) { return '0' <= ch && ch <= '9'; } +bool IsAsciiPunct(char ch) { + return IsInSet(ch, "^-!\"#$%&'()*+,./:;<=>?@[\\]_`{|}~"); +} +bool IsRepeat(char ch) { return IsInSet(ch, "?*+"); } +bool IsAsciiWhiteSpace(char ch) { return IsInSet(ch, " \f\n\r\t\v"); } +bool IsAsciiWordChar(char ch) { + return ('a' <= ch && ch <= 'z') || ('A' <= ch && ch <= 'Z') || + ('0' <= ch && ch <= '9') || ch == '_'; +} + +// Returns true iff "\\c" is a supported escape sequence. +bool IsValidEscape(char c) { + return (IsAsciiPunct(c) || IsInSet(c, "dDfnrsStvwW")); +} + +// Returns true iff the given atom (specified by escaped and pattern) +// matches ch. The result is undefined if the atom is invalid. +bool AtomMatchesChar(bool escaped, char pattern_char, char ch) { + if (escaped) { // "\\p" where p is pattern_char. + switch (pattern_char) { + case 'd': return IsAsciiDigit(ch); + case 'D': return !IsAsciiDigit(ch); + case 'f': return ch == '\f'; + case 'n': return ch == '\n'; + case 'r': return ch == '\r'; + case 's': return IsAsciiWhiteSpace(ch); + case 'S': return !IsAsciiWhiteSpace(ch); + case 't': return ch == '\t'; + case 'v': return ch == '\v'; + case 'w': return IsAsciiWordChar(ch); + case 'W': return !IsAsciiWordChar(ch); + } + return IsAsciiPunct(pattern_char) && pattern_char == ch; + } + + return (pattern_char == '.' && ch != '\n') || pattern_char == ch; +} + +// Helper function used by ValidateRegex() to format error messages. +static std::string FormatRegexSyntaxError(const char* regex, int index) { + return (Message() << "Syntax error at index " << index + << " in simple regular expression \"" << regex << "\": ").GetString(); +} + +// Generates non-fatal failures and returns false if regex is invalid; +// otherwise returns true. +bool ValidateRegex(const char* regex) { + if (regex == nullptr) { + ADD_FAILURE() << "NULL is not a valid simple regular expression."; + return false; + } + + bool is_valid = true; + + // True iff ?, *, or + can follow the previous atom. + bool prev_repeatable = false; + for (int i = 0; regex[i]; i++) { + if (regex[i] == '\\') { // An escape sequence + i++; + if (regex[i] == '\0') { + ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1) + << "'\\' cannot appear at the end."; + return false; + } + + if (!IsValidEscape(regex[i])) { + ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1) + << "invalid escape sequence \"\\" << regex[i] << "\"."; + is_valid = false; + } + prev_repeatable = true; + } else { // Not an escape sequence. + const char ch = regex[i]; + + if (ch == '^' && i > 0) { + ADD_FAILURE() << FormatRegexSyntaxError(regex, i) + << "'^' can only appear at the beginning."; + is_valid = false; + } else if (ch == '$' && regex[i + 1] != '\0') { + ADD_FAILURE() << FormatRegexSyntaxError(regex, i) + << "'$' can only appear at the end."; + is_valid = false; + } else if (IsInSet(ch, "()[]{}|")) { + ADD_FAILURE() << FormatRegexSyntaxError(regex, i) + << "'" << ch << "' is unsupported."; + is_valid = false; + } else if (IsRepeat(ch) && !prev_repeatable) { + ADD_FAILURE() << FormatRegexSyntaxError(regex, i) + << "'" << ch << "' can only follow a repeatable token."; + is_valid = false; + } + + prev_repeatable = !IsInSet(ch, "^$?*+"); + } + } + + return is_valid; +} + +// Matches a repeated regex atom followed by a valid simple regular +// expression. The regex atom is defined as c if escaped is false, +// or \c otherwise. repeat is the repetition meta character (?, *, +// or +). The behavior is undefined if str contains too many +// characters to be indexable by size_t, in which case the test will +// probably time out anyway. We are fine with this limitation as +// std::string has it too. +bool MatchRepetitionAndRegexAtHead( + bool escaped, char c, char repeat, const char* regex, + const char* str) { + const size_t min_count = (repeat == '+') ? 1 : 0; + const size_t max_count = (repeat == '?') ? 1 : + static_cast(-1) - 1; + // We cannot call numeric_limits::max() as it conflicts with the + // max() macro on Windows. + + for (size_t i = 0; i <= max_count; ++i) { + // We know that the atom matches each of the first i characters in str. + if (i >= min_count && MatchRegexAtHead(regex, str + i)) { + // We have enough matches at the head, and the tail matches too. + // Since we only care about *whether* the pattern matches str + // (as opposed to *how* it matches), there is no need to find a + // greedy match. + return true; + } + if (str[i] == '\0' || !AtomMatchesChar(escaped, c, str[i])) + return false; + } + return false; +} + +// Returns true iff regex matches a prefix of str. regex must be a +// valid simple regular expression and not start with "^", or the +// result is undefined. +bool MatchRegexAtHead(const char* regex, const char* str) { + if (*regex == '\0') // An empty regex matches a prefix of anything. + return true; + + // "$" only matches the end of a string. Note that regex being + // valid guarantees that there's nothing after "$" in it. + if (*regex == '$') + return *str == '\0'; + + // Is the first thing in regex an escape sequence? + const bool escaped = *regex == '\\'; + if (escaped) + ++regex; + if (IsRepeat(regex[1])) { + // MatchRepetitionAndRegexAtHead() calls MatchRegexAtHead(), so + // here's an indirect recursion. It terminates as the regex gets + // shorter in each recursion. + return MatchRepetitionAndRegexAtHead( + escaped, regex[0], regex[1], regex + 2, str); + } else { + // regex isn't empty, isn't "$", and doesn't start with a + // repetition. We match the first atom of regex with the first + // character of str and recurse. + return (*str != '\0') && AtomMatchesChar(escaped, *regex, *str) && + MatchRegexAtHead(regex + 1, str + 1); + } +} + +// Returns true iff regex matches any substring of str. regex must be +// a valid simple regular expression, or the result is undefined. +// +// The algorithm is recursive, but the recursion depth doesn't exceed +// the regex length, so we won't need to worry about running out of +// stack space normally. In rare cases the time complexity can be +// exponential with respect to the regex length + the string length, +// but usually it's must faster (often close to linear). +bool MatchRegexAnywhere(const char* regex, const char* str) { + if (regex == nullptr || str == nullptr) return false; + + if (*regex == '^') + return MatchRegexAtHead(regex + 1, str); + + // A successful match can be anywhere in str. + do { + if (MatchRegexAtHead(regex, str)) + return true; + } while (*str++ != '\0'); + return false; +} + +// Implements the RE class. + +RE::~RE() { + free(const_cast(pattern_)); + free(const_cast(full_pattern_)); +} + +// Returns true iff regular expression re matches the entire str. +bool RE::FullMatch(const char* str, const RE& re) { + return re.is_valid_ && MatchRegexAnywhere(re.full_pattern_, str); +} + +// Returns true iff regular expression re matches a substring of str +// (including str itself). +bool RE::PartialMatch(const char* str, const RE& re) { + return re.is_valid_ && MatchRegexAnywhere(re.pattern_, str); +} + +// Initializes an RE from its string representation. +void RE::Init(const char* regex) { + pattern_ = full_pattern_ = nullptr; + if (regex != nullptr) { + pattern_ = posix::StrDup(regex); + } + + is_valid_ = ValidateRegex(regex); + if (!is_valid_) { + // No need to calculate the full pattern when the regex is invalid. + return; + } + + const size_t len = strlen(regex); + // Reserves enough bytes to hold the regular expression used for a + // full match: we need space to prepend a '^', append a '$', and + // terminate the string with '\0'. + char* buffer = static_cast(malloc(len + 3)); + full_pattern_ = buffer; + + if (*regex != '^') + *buffer++ = '^'; // Makes sure full_pattern_ starts with '^'. + + // We don't use snprintf or strncpy, as they trigger a warning when + // compiled with VC++ 8.0. + memcpy(buffer, regex, len); + buffer += len; + + if (len == 0 || regex[len - 1] != '$') + *buffer++ = '$'; // Makes sure full_pattern_ ends with '$'. + + *buffer = '\0'; +} + +#endif // GTEST_USES_POSIX_RE + +const char kUnknownFile[] = "unknown file"; + +// Formats a source file path and a line number as they would appear +// in an error message from the compiler used to compile this code. +GTEST_API_ ::std::string FormatFileLocation(const char* file, int line) { + const std::string file_name(file == nullptr ? kUnknownFile : file); + + if (line < 0) { + return file_name + ":"; + } +#ifdef _MSC_VER + return file_name + "(" + StreamableToString(line) + "):"; +#else + return file_name + ":" + StreamableToString(line) + ":"; +#endif // _MSC_VER +} + +// Formats a file location for compiler-independent XML output. +// Although this function is not platform dependent, we put it next to +// FormatFileLocation in order to contrast the two functions. +// Note that FormatCompilerIndependentFileLocation() does NOT append colon +// to the file location it produces, unlike FormatFileLocation(). +GTEST_API_ ::std::string FormatCompilerIndependentFileLocation( + const char* file, int line) { + const std::string file_name(file == nullptr ? kUnknownFile : file); + + if (line < 0) + return file_name; + else + return file_name + ":" + StreamableToString(line); +} + +GTestLog::GTestLog(GTestLogSeverity severity, const char* file, int line) + : severity_(severity) { + const char* const marker = + severity == GTEST_INFO ? "[ INFO ]" : + severity == GTEST_WARNING ? "[WARNING]" : + severity == GTEST_ERROR ? "[ ERROR ]" : "[ FATAL ]"; + GetStream() << ::std::endl << marker << " " + << FormatFileLocation(file, line).c_str() << ": "; +} + +// Flushes the buffers and, if severity is GTEST_FATAL, aborts the program. +GTestLog::~GTestLog() { + GetStream() << ::std::endl; + if (severity_ == GTEST_FATAL) { + fflush(stderr); + posix::Abort(); + } +} + +// Disable Microsoft deprecation warnings for POSIX functions called from +// this class (creat, dup, dup2, and close) +GTEST_DISABLE_MSC_DEPRECATED_PUSH_() + +#if GTEST_HAS_STREAM_REDIRECTION + +// Object that captures an output stream (stdout/stderr). +class CapturedStream { + public: + // The ctor redirects the stream to a temporary file. +# if GTEST_OS_WINDOWS + explicit CapturedStream(int fd) : fd_(fd), uncaptured_fd_(_dup(fd)) { + char temp_dir_path[MAX_PATH + 1] = { '\0' }; // NOLINT + char temp_file_path[MAX_PATH + 1] = { '\0' }; // NOLINT + + ::GetTempPathA(sizeof(temp_dir_path), temp_dir_path); + const UINT success = ::GetTempFileNameA(temp_dir_path, + "gtest_redir", + 0, // Generate unique file name. + temp_file_path); + GTEST_CHECK_(success != 0) + << "Unable to create a temporary file in " << temp_dir_path; + const int captured_fd = _creat(temp_file_path, _S_IREAD | _S_IWRITE); + GTEST_CHECK_(captured_fd != -1) << "Unable to open temporary file " + << temp_file_path; + filename_ = temp_file_path; +# else + // There's no guarantee that a test has write access to the current + // directory, so we create the temporary file in the /tmp directory + // instead. We use /tmp on most systems, and /sdcard on Android. + // That's because Android doesn't have /tmp. + explicit CapturedStream(int fd) : fd_(fd), uncaptured_fd_(dup(fd)) { +# if GTEST_OS_LINUX_ANDROID + // Note: Android applications are expected to call the framework's + // Context.getExternalStorageDirectory() method through JNI to get + // the location of the world-writable SD Card directory. However, + // this requires a Context handle, which cannot be retrieved + // globally from native code. Doing so also precludes running the + // code as part of a regular standalone executable, which doesn't + // run in a Dalvik process (e.g. when running it through 'adb shell'). + // + // The location /sdcard is directly accessible from native code + // and is the only location (unofficially) supported by the Android + // team. It's generally a symlink to the real SD Card mount point + // which can be /mnt/sdcard, /mnt/sdcard0, /system/media/sdcard, or + // other OEM-customized locations. Never rely on these, and always + // use /sdcard. + char name_template[] = "/sdcard/gtest_captured_stream.XXXXXX"; +# else + char name_template[] = "/tmp/captured_stream.XXXXXX"; +# endif // GTEST_OS_LINUX_ANDROID + const int captured_fd = mkstemp(name_template); + filename_ = name_template; +# endif // GTEST_OS_WINDOWS + fflush(nullptr); +#if GTEST_OS_WINDOWS + _dup2(captured_fd, fd_); + _close(captured_fd); +#else + dup2(captured_fd, fd_); + close(captured_fd); +#endif + } + + ~CapturedStream() { + remove(filename_.c_str()); + } + + std::string GetCapturedString() { + if (uncaptured_fd_ != -1) { + // Restores the original stream. + fflush(nullptr); +#if GTEST_OS_WINDOWS + _dup2(uncaptured_fd_, fd_); + _close(uncaptured_fd_); +#else + dup2(uncaptured_fd_, fd_); + close(uncaptured_fd_); +#endif + uncaptured_fd_ = -1; + } + + FILE* const file = posix::FOpen(filename_.c_str(), "r"); + const std::string content = ReadEntireFile(file); + posix::FClose(file); + return content; + } + + private: + const int fd_; // A stream to capture. + int uncaptured_fd_; + // Name of the temporary file holding the stderr output. + ::std::string filename_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(CapturedStream); +}; + +GTEST_DISABLE_MSC_DEPRECATED_POP_() + +static CapturedStream* g_captured_stderr = nullptr; +static CapturedStream* g_captured_stdout = nullptr; + +// Starts capturing an output stream (stdout/stderr). +static void CaptureStream(int fd, const char* stream_name, + CapturedStream** stream) { + if (*stream != nullptr) { + GTEST_LOG_(FATAL) << "Only one " << stream_name + << " capturer can exist at a time."; + } + *stream = new CapturedStream(fd); +} + +// Stops capturing the output stream and returns the captured string. +static std::string GetCapturedStream(CapturedStream** captured_stream) { + const std::string content = (*captured_stream)->GetCapturedString(); + + delete *captured_stream; + *captured_stream = nullptr; + + return content; +} + +// Starts capturing stdout. +void CaptureStdout() { + CaptureStream(kStdOutFileno, "stdout", &g_captured_stdout); +} + +// Starts capturing stderr. +void CaptureStderr() { + CaptureStream(kStdErrFileno, "stderr", &g_captured_stderr); +} + +// Stops capturing stdout and returns the captured string. +std::string GetCapturedStdout() { + return GetCapturedStream(&g_captured_stdout); +} + +// Stops capturing stderr and returns the captured string. +std::string GetCapturedStderr() { + return GetCapturedStream(&g_captured_stderr); +} + +#endif // GTEST_HAS_STREAM_REDIRECTION + + + + + +size_t GetFileSize(FILE* file) { + fseek(file, 0, SEEK_END); + return static_cast(ftell(file)); +} + +std::string ReadEntireFile(FILE* file) { + const size_t file_size = GetFileSize(file); + char* const buffer = new char[file_size]; + + size_t bytes_last_read = 0; // # of bytes read in the last fread() + size_t bytes_read = 0; // # of bytes read so far + + fseek(file, 0, SEEK_SET); + + // Keeps reading the file until we cannot read further or the + // pre-determined file size is reached. + do { + bytes_last_read = fread(buffer+bytes_read, 1, file_size-bytes_read, file); + bytes_read += bytes_last_read; + } while (bytes_last_read > 0 && bytes_read < file_size); + + const std::string content(buffer, bytes_read); + delete[] buffer; + + return content; +} + +#if GTEST_HAS_DEATH_TEST +static const std::vector* g_injected_test_argvs = + nullptr; // Owned. + +std::vector GetInjectableArgvs() { + if (g_injected_test_argvs != nullptr) { + return *g_injected_test_argvs; + } + return GetArgvs(); +} + +void SetInjectableArgvs(const std::vector* new_argvs) { + if (g_injected_test_argvs != new_argvs) delete g_injected_test_argvs; + g_injected_test_argvs = new_argvs; +} + +void SetInjectableArgvs(const std::vector& new_argvs) { + SetInjectableArgvs( + new std::vector(new_argvs.begin(), new_argvs.end())); +} + +void ClearInjectableArgvs() { + delete g_injected_test_argvs; + g_injected_test_argvs = nullptr; +} +#endif // GTEST_HAS_DEATH_TEST + +#if GTEST_OS_WINDOWS_MOBILE +namespace posix { +void Abort() { + DebugBreak(); + TerminateProcess(GetCurrentProcess(), 1); +} +} // namespace posix +#endif // GTEST_OS_WINDOWS_MOBILE + +// Returns the name of the environment variable corresponding to the +// given flag. For example, FlagToEnvVar("foo") will return +// "GTEST_FOO" in the open-source version. +static std::string FlagToEnvVar(const char* flag) { + const std::string full_flag = + (Message() << GTEST_FLAG_PREFIX_ << flag).GetString(); + + Message env_var; + for (size_t i = 0; i != full_flag.length(); i++) { + env_var << ToUpper(full_flag.c_str()[i]); + } + + return env_var.GetString(); +} + +// Parses 'str' for a 32-bit signed integer. If successful, writes +// the result to *value and returns true; otherwise leaves *value +// unchanged and returns false. +bool ParseInt32(const Message& src_text, const char* str, Int32* value) { + // Parses the environment variable as a decimal integer. + char* end = nullptr; + const long long_value = strtol(str, &end, 10); // NOLINT + + // Has strtol() consumed all characters in the string? + if (*end != '\0') { + // No - an invalid character was encountered. + Message msg; + msg << "WARNING: " << src_text + << " is expected to be a 32-bit integer, but actually" + << " has value \"" << str << "\".\n"; + printf("%s", msg.GetString().c_str()); + fflush(stdout); + return false; + } + + // Is the parsed value in the range of an Int32? + const Int32 result = static_cast(long_value); + if (long_value == LONG_MAX || long_value == LONG_MIN || + // The parsed value overflows as a long. (strtol() returns + // LONG_MAX or LONG_MIN when the input overflows.) + result != long_value + // The parsed value overflows as an Int32. + ) { + Message msg; + msg << "WARNING: " << src_text + << " is expected to be a 32-bit integer, but actually" + << " has value " << str << ", which overflows.\n"; + printf("%s", msg.GetString().c_str()); + fflush(stdout); + return false; + } + + *value = result; + return true; +} + +// Reads and returns the Boolean environment variable corresponding to +// the given flag; if it's not set, returns default_value. +// +// The value is considered true iff it's not "0". +bool BoolFromGTestEnv(const char* flag, bool default_value) { +#if defined(GTEST_GET_BOOL_FROM_ENV_) + return GTEST_GET_BOOL_FROM_ENV_(flag, default_value); +#else + const std::string env_var = FlagToEnvVar(flag); + const char* const string_value = posix::GetEnv(env_var.c_str()); + return string_value == nullptr ? default_value + : strcmp(string_value, "0") != 0; +#endif // defined(GTEST_GET_BOOL_FROM_ENV_) +} + +// Reads and returns a 32-bit integer stored in the environment +// variable corresponding to the given flag; if it isn't set or +// doesn't represent a valid 32-bit integer, returns default_value. +Int32 Int32FromGTestEnv(const char* flag, Int32 default_value) { +#if defined(GTEST_GET_INT32_FROM_ENV_) + return GTEST_GET_INT32_FROM_ENV_(flag, default_value); +#else + const std::string env_var = FlagToEnvVar(flag); + const char* const string_value = posix::GetEnv(env_var.c_str()); + if (string_value == nullptr) { + // The environment variable is not set. + return default_value; + } + + Int32 result = default_value; + if (!ParseInt32(Message() << "Environment variable " << env_var, + string_value, &result)) { + printf("The default value %s is used.\n", + (Message() << default_value).GetString().c_str()); + fflush(stdout); + return default_value; + } + + return result; +#endif // defined(GTEST_GET_INT32_FROM_ENV_) +} + +// As a special case for the 'output' flag, if GTEST_OUTPUT is not +// set, we look for XML_OUTPUT_FILE, which is set by the Bazel build +// system. The value of XML_OUTPUT_FILE is a filename without the +// "xml:" prefix of GTEST_OUTPUT. +// Note that this is meant to be called at the call site so it does +// not check that the flag is 'output' +// In essence this checks an env variable called XML_OUTPUT_FILE +// and if it is set we prepend "xml:" to its value, if it not set we return "" +std::string OutputFlagAlsoCheckEnvVar(){ + std::string default_value_for_output_flag = ""; + const char* xml_output_file_env = posix::GetEnv("XML_OUTPUT_FILE"); + if (nullptr != xml_output_file_env) { + default_value_for_output_flag = std::string("xml:") + xml_output_file_env; + } + return default_value_for_output_flag; +} + +// Reads and returns the string environment variable corresponding to +// the given flag; if it's not set, returns default_value. +const char* StringFromGTestEnv(const char* flag, const char* default_value) { +#if defined(GTEST_GET_STRING_FROM_ENV_) + return GTEST_GET_STRING_FROM_ENV_(flag, default_value); +#else + const std::string env_var = FlagToEnvVar(flag); + const char* const value = posix::GetEnv(env_var.c_str()); + return value == nullptr ? default_value : value; +#endif // defined(GTEST_GET_STRING_FROM_ENV_) +} + +} // namespace internal +} // namespace testing diff --git a/deps/googletest/src/gtest-printers.cc b/deps/googletest/src/gtest-printers.cc new file mode 100644 index 000000000..40a8817e1 --- /dev/null +++ b/deps/googletest/src/gtest-printers.cc @@ -0,0 +1,441 @@ +// Copyright 2007, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +// Google Test - The Google C++ Testing and Mocking Framework +// +// This file implements a universal value printer that can print a +// value of any type T: +// +// void ::testing::internal::UniversalPrinter::Print(value, ostream_ptr); +// +// It uses the << operator when possible, and prints the bytes in the +// object otherwise. A user can override its behavior for a class +// type Foo by defining either operator<<(::std::ostream&, const Foo&) +// or void PrintTo(const Foo&, ::std::ostream*) in the namespace that +// defines Foo. + +#include "gtest/gtest-printers.h" +#include +#include +#include +#include // NOLINT +#include +#include "gtest/internal/gtest-port.h" +#include "src/gtest-internal-inl.h" + +namespace testing { + +namespace { + +using ::std::ostream; + +// Prints a segment of bytes in the given object. +GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ +GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ +void PrintByteSegmentInObjectTo(const unsigned char* obj_bytes, size_t start, + size_t count, ostream* os) { + char text[5] = ""; + for (size_t i = 0; i != count; i++) { + const size_t j = start + i; + if (i != 0) { + // Organizes the bytes into groups of 2 for easy parsing by + // human. + if ((j % 2) == 0) + *os << ' '; + else + *os << '-'; + } + GTEST_SNPRINTF_(text, sizeof(text), "%02X", obj_bytes[j]); + *os << text; + } +} + +// Prints the bytes in the given value to the given ostream. +void PrintBytesInObjectToImpl(const unsigned char* obj_bytes, size_t count, + ostream* os) { + // Tells the user how big the object is. + *os << count << "-byte object <"; + + const size_t kThreshold = 132; + const size_t kChunkSize = 64; + // If the object size is bigger than kThreshold, we'll have to omit + // some details by printing only the first and the last kChunkSize + // bytes. + if (count < kThreshold) { + PrintByteSegmentInObjectTo(obj_bytes, 0, count, os); + } else { + PrintByteSegmentInObjectTo(obj_bytes, 0, kChunkSize, os); + *os << " ... "; + // Rounds up to 2-byte boundary. + const size_t resume_pos = (count - kChunkSize + 1)/2*2; + PrintByteSegmentInObjectTo(obj_bytes, resume_pos, count - resume_pos, os); + } + *os << ">"; +} + +} // namespace + +namespace internal2 { + +// Delegates to PrintBytesInObjectToImpl() to print the bytes in the +// given object. The delegation simplifies the implementation, which +// uses the << operator and thus is easier done outside of the +// ::testing::internal namespace, which contains a << operator that +// sometimes conflicts with the one in STL. +void PrintBytesInObjectTo(const unsigned char* obj_bytes, size_t count, + ostream* os) { + PrintBytesInObjectToImpl(obj_bytes, count, os); +} + +} // namespace internal2 + +namespace internal { + +// Depending on the value of a char (or wchar_t), we print it in one +// of three formats: +// - as is if it's a printable ASCII (e.g. 'a', '2', ' '), +// - as a hexadecimal escape sequence (e.g. '\x7F'), or +// - as a special escape sequence (e.g. '\r', '\n'). +enum CharFormat { + kAsIs, + kHexEscape, + kSpecialEscape +}; + +// Returns true if c is a printable ASCII character. We test the +// value of c directly instead of calling isprint(), which is buggy on +// Windows Mobile. +inline bool IsPrintableAscii(wchar_t c) { + return 0x20 <= c && c <= 0x7E; +} + +// Prints a wide or narrow char c as a character literal without the +// quotes, escaping it when necessary; returns how c was formatted. +// The template argument UnsignedChar is the unsigned version of Char, +// which is the type of c. +template +static CharFormat PrintAsCharLiteralTo(Char c, ostream* os) { + switch (static_cast(c)) { + case L'\0': + *os << "\\0"; + break; + case L'\'': + *os << "\\'"; + break; + case L'\\': + *os << "\\\\"; + break; + case L'\a': + *os << "\\a"; + break; + case L'\b': + *os << "\\b"; + break; + case L'\f': + *os << "\\f"; + break; + case L'\n': + *os << "\\n"; + break; + case L'\r': + *os << "\\r"; + break; + case L'\t': + *os << "\\t"; + break; + case L'\v': + *os << "\\v"; + break; + default: + if (IsPrintableAscii(c)) { + *os << static_cast(c); + return kAsIs; + } else { + ostream::fmtflags flags = os->flags(); + *os << "\\x" << std::hex << std::uppercase + << static_cast(static_cast(c)); + os->flags(flags); + return kHexEscape; + } + } + return kSpecialEscape; +} + +// Prints a wchar_t c as if it's part of a string literal, escaping it when +// necessary; returns how c was formatted. +static CharFormat PrintAsStringLiteralTo(wchar_t c, ostream* os) { + switch (c) { + case L'\'': + *os << "'"; + return kAsIs; + case L'"': + *os << "\\\""; + return kSpecialEscape; + default: + return PrintAsCharLiteralTo(c, os); + } +} + +// Prints a char c as if it's part of a string literal, escaping it when +// necessary; returns how c was formatted. +static CharFormat PrintAsStringLiteralTo(char c, ostream* os) { + return PrintAsStringLiteralTo( + static_cast(static_cast(c)), os); +} + +// Prints a wide or narrow character c and its code. '\0' is printed +// as "'\\0'", other unprintable characters are also properly escaped +// using the standard C++ escape sequence. The template argument +// UnsignedChar is the unsigned version of Char, which is the type of c. +template +void PrintCharAndCodeTo(Char c, ostream* os) { + // First, print c as a literal in the most readable form we can find. + *os << ((sizeof(c) > 1) ? "L'" : "'"); + const CharFormat format = PrintAsCharLiteralTo(c, os); + *os << "'"; + + // To aid user debugging, we also print c's code in decimal, unless + // it's 0 (in which case c was printed as '\\0', making the code + // obvious). + if (c == 0) + return; + *os << " (" << static_cast(c); + + // For more convenience, we print c's code again in hexadecimal, + // unless c was already printed in the form '\x##' or the code is in + // [1, 9]. + if (format == kHexEscape || (1 <= c && c <= 9)) { + // Do nothing. + } else { + *os << ", 0x" << String::FormatHexInt(static_cast(c)); + } + *os << ")"; +} + +void PrintTo(unsigned char c, ::std::ostream* os) { + PrintCharAndCodeTo(c, os); +} +void PrintTo(signed char c, ::std::ostream* os) { + PrintCharAndCodeTo(c, os); +} + +// Prints a wchar_t as a symbol if it is printable or as its internal +// code otherwise and also as its code. L'\0' is printed as "L'\\0'". +void PrintTo(wchar_t wc, ostream* os) { + PrintCharAndCodeTo(wc, os); +} + +// Prints the given array of characters to the ostream. CharType must be either +// char or wchar_t. +// The array starts at begin, the length is len, it may include '\0' characters +// and may not be NUL-terminated. +template +GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ +GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ +static CharFormat PrintCharsAsStringTo( + const CharType* begin, size_t len, ostream* os) { + const char* const kQuoteBegin = sizeof(CharType) == 1 ? "\"" : "L\""; + *os << kQuoteBegin; + bool is_previous_hex = false; + CharFormat print_format = kAsIs; + for (size_t index = 0; index < len; ++index) { + const CharType cur = begin[index]; + if (is_previous_hex && IsXDigit(cur)) { + // Previous character is of '\x..' form and this character can be + // interpreted as another hexadecimal digit in its number. Break string to + // disambiguate. + *os << "\" " << kQuoteBegin; + } + is_previous_hex = PrintAsStringLiteralTo(cur, os) == kHexEscape; + // Remember if any characters required hex escaping. + if (is_previous_hex) { + print_format = kHexEscape; + } + } + *os << "\""; + return print_format; +} + +// Prints a (const) char/wchar_t array of 'len' elements, starting at address +// 'begin'. CharType must be either char or wchar_t. +template +GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ +GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_HWADDRESS_ +GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ +static void UniversalPrintCharArray( + const CharType* begin, size_t len, ostream* os) { + // The code + // const char kFoo[] = "foo"; + // generates an array of 4, not 3, elements, with the last one being '\0'. + // + // Therefore when printing a char array, we don't print the last element if + // it's '\0', such that the output matches the string literal as it's + // written in the source code. + if (len > 0 && begin[len - 1] == '\0') { + PrintCharsAsStringTo(begin, len - 1, os); + return; + } + + // If, however, the last element in the array is not '\0', e.g. + // const char kFoo[] = { 'f', 'o', 'o' }; + // we must print the entire array. We also print a message to indicate + // that the array is not NUL-terminated. + PrintCharsAsStringTo(begin, len, os); + *os << " (no terminating NUL)"; +} + +// Prints a (const) char array of 'len' elements, starting at address 'begin'. +void UniversalPrintArray(const char* begin, size_t len, ostream* os) { + UniversalPrintCharArray(begin, len, os); +} + +// Prints a (const) wchar_t array of 'len' elements, starting at address +// 'begin'. +void UniversalPrintArray(const wchar_t* begin, size_t len, ostream* os) { + UniversalPrintCharArray(begin, len, os); +} + +// Prints the given C string to the ostream. +void PrintTo(const char* s, ostream* os) { + if (s == nullptr) { + *os << "NULL"; + } else { + *os << ImplicitCast_(s) << " pointing to "; + PrintCharsAsStringTo(s, strlen(s), os); + } +} + +// MSVC compiler can be configured to define whar_t as a typedef +// of unsigned short. Defining an overload for const wchar_t* in that case +// would cause pointers to unsigned shorts be printed as wide strings, +// possibly accessing more memory than intended and causing invalid +// memory accesses. MSVC defines _NATIVE_WCHAR_T_DEFINED symbol when +// wchar_t is implemented as a native type. +#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED) +// Prints the given wide C string to the ostream. +void PrintTo(const wchar_t* s, ostream* os) { + if (s == nullptr) { + *os << "NULL"; + } else { + *os << ImplicitCast_(s) << " pointing to "; + PrintCharsAsStringTo(s, wcslen(s), os); + } +} +#endif // wchar_t is native + +namespace { + +bool ContainsUnprintableControlCodes(const char* str, size_t length) { + const unsigned char *s = reinterpret_cast(str); + + for (size_t i = 0; i < length; i++) { + unsigned char ch = *s++; + if (std::iscntrl(ch)) { + switch (ch) { + case '\t': + case '\n': + case '\r': + break; + default: + return true; + } + } + } + return false; +} + +bool IsUTF8TrailByte(unsigned char t) { return 0x80 <= t && t<= 0xbf; } + +bool IsValidUTF8(const char* str, size_t length) { + const unsigned char *s = reinterpret_cast(str); + + for (size_t i = 0; i < length;) { + unsigned char lead = s[i++]; + + if (lead <= 0x7f) { + continue; // single-byte character (ASCII) 0..7F + } + if (lead < 0xc2) { + return false; // trail byte or non-shortest form + } else if (lead <= 0xdf && (i + 1) <= length && IsUTF8TrailByte(s[i])) { + ++i; // 2-byte character + } else if (0xe0 <= lead && lead <= 0xef && (i + 2) <= length && + IsUTF8TrailByte(s[i]) && + IsUTF8TrailByte(s[i + 1]) && + // check for non-shortest form and surrogate + (lead != 0xe0 || s[i] >= 0xa0) && + (lead != 0xed || s[i] < 0xa0)) { + i += 2; // 3-byte character + } else if (0xf0 <= lead && lead <= 0xf4 && (i + 3) <= length && + IsUTF8TrailByte(s[i]) && + IsUTF8TrailByte(s[i + 1]) && + IsUTF8TrailByte(s[i + 2]) && + // check for non-shortest form + (lead != 0xf0 || s[i] >= 0x90) && + (lead != 0xf4 || s[i] < 0x90)) { + i += 3; // 4-byte character + } else { + return false; + } + } + return true; +} + +void ConditionalPrintAsText(const char* str, size_t length, ostream* os) { + if (!ContainsUnprintableControlCodes(str, length) && + IsValidUTF8(str, length)) { + *os << "\n As Text: \"" << str << "\""; + } +} + +} // anonymous namespace + +void PrintStringTo(const ::std::string& s, ostream* os) { + if (PrintCharsAsStringTo(s.data(), s.size(), os) == kHexEscape) { + if (GTEST_FLAG(print_utf8)) { + ConditionalPrintAsText(s.data(), s.size(), os); + } + } +} + +#if GTEST_HAS_STD_WSTRING +void PrintWideStringTo(const ::std::wstring& s, ostream* os) { + PrintCharsAsStringTo(s.data(), s.size(), os); +} +#endif // GTEST_HAS_STD_WSTRING + +} // namespace internal + +} // namespace testing diff --git a/deps/googletest/src/gtest-test-part.cc b/deps/googletest/src/gtest-test-part.cc new file mode 100644 index 000000000..178317a6b --- /dev/null +++ b/deps/googletest/src/gtest-test-part.cc @@ -0,0 +1,104 @@ +// Copyright 2008, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// The Google C++ Testing and Mocking Framework (Google Test) + +#include "gtest/gtest-test-part.h" +#include "src/gtest-internal-inl.h" + +namespace testing { + +using internal::GetUnitTestImpl; + +// Gets the summary of the failure message by omitting the stack trace +// in it. +std::string TestPartResult::ExtractSummary(const char* message) { + const char* const stack_trace = strstr(message, internal::kStackTraceMarker); + return stack_trace == nullptr ? message : std::string(message, stack_trace); +} + +// Prints a TestPartResult object. +std::ostream& operator<<(std::ostream& os, const TestPartResult& result) { + return os << result.file_name() << ":" << result.line_number() << ": " + << (result.type() == TestPartResult::kSuccess + ? "Success" + : result.type() == TestPartResult::kSkip + ? "Skipped" + : result.type() == TestPartResult::kFatalFailure + ? "Fatal failure" + : "Non-fatal failure") + << ":\n" + << result.message() << std::endl; +} + +// Appends a TestPartResult to the array. +void TestPartResultArray::Append(const TestPartResult& result) { + array_.push_back(result); +} + +// Returns the TestPartResult at the given index (0-based). +const TestPartResult& TestPartResultArray::GetTestPartResult(int index) const { + if (index < 0 || index >= size()) { + printf("\nInvalid index (%d) into TestPartResultArray.\n", index); + internal::posix::Abort(); + } + + return array_[static_cast(index)]; +} + +// Returns the number of TestPartResult objects in the array. +int TestPartResultArray::size() const { + return static_cast(array_.size()); +} + +namespace internal { + +HasNewFatalFailureHelper::HasNewFatalFailureHelper() + : has_new_fatal_failure_(false), + original_reporter_(GetUnitTestImpl()-> + GetTestPartResultReporterForCurrentThread()) { + GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(this); +} + +HasNewFatalFailureHelper::~HasNewFatalFailureHelper() { + GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread( + original_reporter_); +} + +void HasNewFatalFailureHelper::ReportTestPartResult( + const TestPartResult& result) { + if (result.fatally_failed()) + has_new_fatal_failure_ = true; + original_reporter_->ReportTestPartResult(result); +} + +} // namespace internal + +} // namespace testing diff --git a/deps/googletest/src/gtest-typed-test.cc b/deps/googletest/src/gtest-typed-test.cc new file mode 100644 index 000000000..8677caf73 --- /dev/null +++ b/deps/googletest/src/gtest-typed-test.cc @@ -0,0 +1,118 @@ +// Copyright 2008 Google Inc. +// All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +#include "gtest/gtest-typed-test.h" + +#include "gtest/gtest.h" + +namespace testing { +namespace internal { + +#if GTEST_HAS_TYPED_TEST_P + +// Skips to the first non-space char in str. Returns an empty string if str +// contains only whitespace characters. +static const char* SkipSpaces(const char* str) { + while (IsSpace(*str)) + str++; + return str; +} + +static std::vector SplitIntoTestNames(const char* src) { + std::vector name_vec; + src = SkipSpaces(src); + for (; src != nullptr; src = SkipComma(src)) { + name_vec.push_back(StripTrailingSpaces(GetPrefixUntilComma(src))); + } + return name_vec; +} + +// Verifies that registered_tests match the test names in +// registered_tests_; returns registered_tests if successful, or +// aborts the program otherwise. +const char* TypedTestSuitePState::VerifyRegisteredTestNames( + const char* file, int line, const char* registered_tests) { + typedef RegisteredTestsMap::const_iterator RegisteredTestIter; + registered_ = true; + + std::vector name_vec = SplitIntoTestNames(registered_tests); + + Message errors; + + std::set tests; + for (std::vector::const_iterator name_it = name_vec.begin(); + name_it != name_vec.end(); ++name_it) { + const std::string& name = *name_it; + if (tests.count(name) != 0) { + errors << "Test " << name << " is listed more than once.\n"; + continue; + } + + bool found = false; + for (RegisteredTestIter it = registered_tests_.begin(); + it != registered_tests_.end(); + ++it) { + if (name == it->first) { + found = true; + break; + } + } + + if (found) { + tests.insert(name); + } else { + errors << "No test named " << name + << " can be found in this test suite.\n"; + } + } + + for (RegisteredTestIter it = registered_tests_.begin(); + it != registered_tests_.end(); + ++it) { + if (tests.count(it->first) == 0) { + errors << "You forgot to list test " << it->first << ".\n"; + } + } + + const std::string& errors_str = errors.GetString(); + if (errors_str != "") { + fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(), + errors_str.c_str()); + fflush(stderr); + posix::Abort(); + } + + return registered_tests; +} + +#endif // GTEST_HAS_TYPED_TEST_P + +} // namespace internal +} // namespace testing diff --git a/deps/googletest/src/gtest.cc b/deps/googletest/src/gtest.cc new file mode 100644 index 000000000..271a51a9e --- /dev/null +++ b/deps/googletest/src/gtest.cc @@ -0,0 +1,6124 @@ +// Copyright 2005, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +// +// The Google C++ Testing and Mocking Framework (Google Test) + +#include "gtest/gtest.h" +#include "gtest/internal/custom/gtest.h" +#include "gtest/gtest-spi.h" + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include // NOLINT +#include +#include + +#if GTEST_OS_LINUX + +# define GTEST_HAS_GETTIMEOFDAY_ 1 + +# include // NOLINT +# include // NOLINT +# include // NOLINT +// Declares vsnprintf(). This header is not available on Windows. +# include // NOLINT +# include // NOLINT +# include // NOLINT +# include // NOLINT +# include + +#elif GTEST_OS_ZOS +# define GTEST_HAS_GETTIMEOFDAY_ 1 +# include // NOLINT + +// On z/OS we additionally need strings.h for strcasecmp. +# include // NOLINT + +#elif GTEST_OS_WINDOWS_MOBILE // We are on Windows CE. + +# include // NOLINT +# undef min + +#elif GTEST_OS_WINDOWS // We are on Windows proper. + +# include // NOLINT +# include // NOLINT +# include // NOLINT +# include // NOLINT + +# if GTEST_OS_WINDOWS_MINGW +// MinGW has gettimeofday() but not _ftime64(). +# define GTEST_HAS_GETTIMEOFDAY_ 1 +# include // NOLINT +# endif // GTEST_OS_WINDOWS_MINGW + +// cpplint thinks that the header is already included, so we want to +// silence it. +# include // NOLINT +# undef min + +#else + +// Assume other platforms have gettimeofday(). +# define GTEST_HAS_GETTIMEOFDAY_ 1 + +// cpplint thinks that the header is already included, so we want to +// silence it. +# include // NOLINT +# include // NOLINT + +#endif // GTEST_OS_LINUX + +#if GTEST_HAS_EXCEPTIONS +# include +#endif + +#if GTEST_CAN_STREAM_RESULTS_ +# include // NOLINT +# include // NOLINT +# include // NOLINT +# include // NOLINT +#endif + +#include "src/gtest-internal-inl.h" + +#if GTEST_OS_WINDOWS +# define vsnprintf _vsnprintf +#endif // GTEST_OS_WINDOWS + +#if GTEST_OS_MAC +#ifndef GTEST_OS_IOS +#include +#endif +#endif + +#if GTEST_HAS_ABSL +#include "absl/debugging/failure_signal_handler.h" +#include "absl/debugging/stacktrace.h" +#include "absl/debugging/symbolize.h" +#include "absl/strings/str_cat.h" +#endif // GTEST_HAS_ABSL + +namespace testing { + +using internal::CountIf; +using internal::ForEach; +using internal::GetElementOr; +using internal::Shuffle; + +// Constants. + +// A test whose test suite name or test name matches this filter is +// disabled and not run. +static const char kDisableTestFilter[] = "DISABLED_*:*/DISABLED_*"; + +// A test suite whose name matches this filter is considered a death +// test suite and will be run before test suites whose name doesn't +// match this filter. +static const char kDeathTestSuiteFilter[] = "*DeathTest:*DeathTest/*"; + +// A test filter that matches everything. +static const char kUniversalFilter[] = "*"; + +// The default output format. +static const char kDefaultOutputFormat[] = "xml"; +// The default output file. +static const char kDefaultOutputFile[] = "test_detail"; + +// The environment variable name for the test shard index. +static const char kTestShardIndex[] = "GTEST_SHARD_INDEX"; +// The environment variable name for the total number of test shards. +static const char kTestTotalShards[] = "GTEST_TOTAL_SHARDS"; +// The environment variable name for the test shard status file. +static const char kTestShardStatusFile[] = "GTEST_SHARD_STATUS_FILE"; + +namespace internal { + +// The text used in failure messages to indicate the start of the +// stack trace. +const char kStackTraceMarker[] = "\nStack trace:\n"; + +// g_help_flag is true iff the --help flag or an equivalent form is +// specified on the command line. +bool g_help_flag = false; + +// Utilty function to Open File for Writing +static FILE* OpenFileForWriting(const std::string& output_file) { + FILE* fileout = nullptr; + FilePath output_file_path(output_file); + FilePath output_dir(output_file_path.RemoveFileName()); + + if (output_dir.CreateDirectoriesRecursively()) { + fileout = posix::FOpen(output_file.c_str(), "w"); + } + if (fileout == nullptr) { + GTEST_LOG_(FATAL) << "Unable to open file \"" << output_file << "\""; + } + return fileout; +} + +} // namespace internal + +// Bazel passes in the argument to '--test_filter' via the TESTBRIDGE_TEST_ONLY +// environment variable. +static const char* GetDefaultFilter() { + const char* const testbridge_test_only = + internal::posix::GetEnv("TESTBRIDGE_TEST_ONLY"); + if (testbridge_test_only != nullptr) { + return testbridge_test_only; + } + return kUniversalFilter; +} + +GTEST_DEFINE_bool_( + also_run_disabled_tests, + internal::BoolFromGTestEnv("also_run_disabled_tests", false), + "Run disabled tests too, in addition to the tests normally being run."); + +GTEST_DEFINE_bool_( + break_on_failure, + internal::BoolFromGTestEnv("break_on_failure", false), + "True iff a failed assertion should be a debugger break-point."); + +GTEST_DEFINE_bool_( + catch_exceptions, + internal::BoolFromGTestEnv("catch_exceptions", true), + "True iff " GTEST_NAME_ + " should catch exceptions and treat them as test failures."); + +GTEST_DEFINE_string_( + color, + internal::StringFromGTestEnv("color", "auto"), + "Whether to use colors in the output. Valid values: yes, no, " + "and auto. 'auto' means to use colors if the output is " + "being sent to a terminal and the TERM environment variable " + "is set to a terminal type that supports colors."); + +GTEST_DEFINE_string_( + filter, + internal::StringFromGTestEnv("filter", GetDefaultFilter()), + "A colon-separated list of glob (not regex) patterns " + "for filtering the tests to run, optionally followed by a " + "'-' and a : separated list of negative patterns (tests to " + "exclude). A test is run if it matches one of the positive " + "patterns and does not match any of the negative patterns."); + +GTEST_DEFINE_bool_( + install_failure_signal_handler, + internal::BoolFromGTestEnv("install_failure_signal_handler", false), + "If true and supported on the current platform, " GTEST_NAME_ " should " + "install a signal handler that dumps debugging information when fatal " + "signals are raised."); + +GTEST_DEFINE_bool_(list_tests, false, + "List all tests without running them."); + +// The net priority order after flag processing is thus: +// --gtest_output command line flag +// GTEST_OUTPUT environment variable +// XML_OUTPUT_FILE environment variable +// '' +GTEST_DEFINE_string_( + output, + internal::StringFromGTestEnv("output", + internal::OutputFlagAlsoCheckEnvVar().c_str()), + "A format (defaults to \"xml\" but can be specified to be \"json\"), " + "optionally followed by a colon and an output file name or directory. " + "A directory is indicated by a trailing pathname separator. " + "Examples: \"xml:filename.xml\", \"xml::directoryname/\". " + "If a directory is specified, output files will be created " + "within that directory, with file-names based on the test " + "executable's name and, if necessary, made unique by adding " + "digits."); + +GTEST_DEFINE_bool_( + print_time, + internal::BoolFromGTestEnv("print_time", true), + "True iff " GTEST_NAME_ + " should display elapsed time in text output."); + +GTEST_DEFINE_bool_( + print_utf8, + internal::BoolFromGTestEnv("print_utf8", true), + "True iff " GTEST_NAME_ + " prints UTF8 characters as text."); + +GTEST_DEFINE_int32_( + random_seed, + internal::Int32FromGTestEnv("random_seed", 0), + "Random number seed to use when shuffling test orders. Must be in range " + "[1, 99999], or 0 to use a seed based on the current time."); + +GTEST_DEFINE_int32_( + repeat, + internal::Int32FromGTestEnv("repeat", 1), + "How many times to repeat each test. Specify a negative number " + "for repeating forever. Useful for shaking out flaky tests."); + +GTEST_DEFINE_bool_( + show_internal_stack_frames, false, + "True iff " GTEST_NAME_ " should include internal stack frames when " + "printing test failure stack traces."); + +GTEST_DEFINE_bool_( + shuffle, + internal::BoolFromGTestEnv("shuffle", false), + "True iff " GTEST_NAME_ + " should randomize tests' order on every run."); + +GTEST_DEFINE_int32_( + stack_trace_depth, + internal::Int32FromGTestEnv("stack_trace_depth", kMaxStackTraceDepth), + "The maximum number of stack frames to print when an " + "assertion fails. The valid range is 0 through 100, inclusive."); + +GTEST_DEFINE_string_( + stream_result_to, + internal::StringFromGTestEnv("stream_result_to", ""), + "This flag specifies the host name and the port number on which to stream " + "test results. Example: \"localhost:555\". The flag is effective only on " + "Linux."); + +GTEST_DEFINE_bool_( + throw_on_failure, + internal::BoolFromGTestEnv("throw_on_failure", false), + "When this flag is specified, a failed assertion will throw an exception " + "if exceptions are enabled or exit the program with a non-zero code " + "otherwise. For use with an external test framework."); + +#if GTEST_USE_OWN_FLAGFILE_FLAG_ +GTEST_DEFINE_string_( + flagfile, + internal::StringFromGTestEnv("flagfile", ""), + "This flag specifies the flagfile to read command-line flags from."); +#endif // GTEST_USE_OWN_FLAGFILE_FLAG_ + +namespace internal { + +// Generates a random number from [0, range), using a Linear +// Congruential Generator (LCG). Crashes if 'range' is 0 or greater +// than kMaxRange. +UInt32 Random::Generate(UInt32 range) { + // These constants are the same as are used in glibc's rand(3). + // Use wider types than necessary to prevent unsigned overflow diagnostics. + state_ = static_cast(1103515245ULL*state_ + 12345U) % kMaxRange; + + GTEST_CHECK_(range > 0) + << "Cannot generate a number in the range [0, 0)."; + GTEST_CHECK_(range <= kMaxRange) + << "Generation of a number in [0, " << range << ") was requested, " + << "but this can only generate numbers in [0, " << kMaxRange << ")."; + + // Converting via modulus introduces a bit of downward bias, but + // it's simple, and a linear congruential generator isn't too good + // to begin with. + return state_ % range; +} + +// GTestIsInitialized() returns true iff the user has initialized +// Google Test. Useful for catching the user mistake of not initializing +// Google Test before calling RUN_ALL_TESTS(). +static bool GTestIsInitialized() { return GetArgvs().size() > 0; } + +// Iterates over a vector of TestSuites, keeping a running sum of the +// results of calling a given int-returning method on each. +// Returns the sum. +static int SumOverTestSuiteList(const std::vector& case_list, + int (TestSuite::*method)() const) { + int sum = 0; + for (size_t i = 0; i < case_list.size(); i++) { + sum += (case_list[i]->*method)(); + } + return sum; +} + +// Returns true iff the test suite passed. +static bool TestSuitePassed(const TestSuite* test_suite) { + return test_suite->should_run() && test_suite->Passed(); +} + +// Returns true iff the test suite failed. +static bool TestSuiteFailed(const TestSuite* test_suite) { + return test_suite->should_run() && test_suite->Failed(); +} + +// Returns true iff test_suite contains at least one test that should +// run. +static bool ShouldRunTestSuite(const TestSuite* test_suite) { + return test_suite->should_run(); +} + +// AssertHelper constructor. +AssertHelper::AssertHelper(TestPartResult::Type type, + const char* file, + int line, + const char* message) + : data_(new AssertHelperData(type, file, line, message)) { +} + +AssertHelper::~AssertHelper() { + delete data_; +} + +// Message assignment, for assertion streaming support. +void AssertHelper::operator=(const Message& message) const { + UnitTest::GetInstance()-> + AddTestPartResult(data_->type, data_->file, data_->line, + AppendUserMessage(data_->message, message), + UnitTest::GetInstance()->impl() + ->CurrentOsStackTraceExceptTop(1) + // Skips the stack frame for this function itself. + ); // NOLINT +} + +// A copy of all command line arguments. Set by InitGoogleTest(). +static ::std::vector g_argvs; + +::std::vector GetArgvs() { +#if defined(GTEST_CUSTOM_GET_ARGVS_) + // GTEST_CUSTOM_GET_ARGVS_() may return a container of std::string or + // ::string. This code converts it to the appropriate type. + const auto& custom = GTEST_CUSTOM_GET_ARGVS_(); + return ::std::vector(custom.begin(), custom.end()); +#else // defined(GTEST_CUSTOM_GET_ARGVS_) + return g_argvs; +#endif // defined(GTEST_CUSTOM_GET_ARGVS_) +} + +// Returns the current application's name, removing directory path if that +// is present. +FilePath GetCurrentExecutableName() { + FilePath result; + +#if GTEST_OS_WINDOWS || GTEST_OS_OS2 + result.Set(FilePath(GetArgvs()[0]).RemoveExtension("exe")); +#else + result.Set(FilePath(GetArgvs()[0])); +#endif // GTEST_OS_WINDOWS + + return result.RemoveDirectoryName(); +} + +// Functions for processing the gtest_output flag. + +// Returns the output format, or "" for normal printed output. +std::string UnitTestOptions::GetOutputFormat() { + const char* const gtest_output_flag = GTEST_FLAG(output).c_str(); + const char* const colon = strchr(gtest_output_flag, ':'); + return (colon == nullptr) + ? std::string(gtest_output_flag) + : std::string(gtest_output_flag, + static_cast(colon - gtest_output_flag)); +} + +// Returns the name of the requested output file, or the default if none +// was explicitly specified. +std::string UnitTestOptions::GetAbsolutePathToOutputFile() { + const char* const gtest_output_flag = GTEST_FLAG(output).c_str(); + + std::string format = GetOutputFormat(); + if (format.empty()) + format = std::string(kDefaultOutputFormat); + + const char* const colon = strchr(gtest_output_flag, ':'); + if (colon == nullptr) + return internal::FilePath::MakeFileName( + internal::FilePath( + UnitTest::GetInstance()->original_working_dir()), + internal::FilePath(kDefaultOutputFile), 0, + format.c_str()).string(); + + internal::FilePath output_name(colon + 1); + if (!output_name.IsAbsolutePath()) + output_name = internal::FilePath::ConcatPaths( + internal::FilePath(UnitTest::GetInstance()->original_working_dir()), + internal::FilePath(colon + 1)); + + if (!output_name.IsDirectory()) + return output_name.string(); + + internal::FilePath result(internal::FilePath::GenerateUniqueFileName( + output_name, internal::GetCurrentExecutableName(), + GetOutputFormat().c_str())); + return result.string(); +} + +// Returns true iff the wildcard pattern matches the string. The +// first ':' or '\0' character in pattern marks the end of it. +// +// This recursive algorithm isn't very efficient, but is clear and +// works well enough for matching test names, which are short. +bool UnitTestOptions::PatternMatchesString(const char *pattern, + const char *str) { + switch (*pattern) { + case '\0': + case ':': // Either ':' or '\0' marks the end of the pattern. + return *str == '\0'; + case '?': // Matches any single character. + return *str != '\0' && PatternMatchesString(pattern + 1, str + 1); + case '*': // Matches any string (possibly empty) of characters. + return (*str != '\0' && PatternMatchesString(pattern, str + 1)) || + PatternMatchesString(pattern + 1, str); + default: // Non-special character. Matches itself. + return *pattern == *str && + PatternMatchesString(pattern + 1, str + 1); + } +} + +bool UnitTestOptions::MatchesFilter( + const std::string& name, const char* filter) { + const char *cur_pattern = filter; + for (;;) { + if (PatternMatchesString(cur_pattern, name.c_str())) { + return true; + } + + // Finds the next pattern in the filter. + cur_pattern = strchr(cur_pattern, ':'); + + // Returns if no more pattern can be found. + if (cur_pattern == nullptr) { + return false; + } + + // Skips the pattern separater (the ':' character). + cur_pattern++; + } +} + +// Returns true iff the user-specified filter matches the test suite +// name and the test name. +bool UnitTestOptions::FilterMatchesTest(const std::string& test_suite_name, + const std::string& test_name) { + const std::string& full_name = test_suite_name + "." + test_name.c_str(); + + // Split --gtest_filter at '-', if there is one, to separate into + // positive filter and negative filter portions + const char* const p = GTEST_FLAG(filter).c_str(); + const char* const dash = strchr(p, '-'); + std::string positive; + std::string negative; + if (dash == nullptr) { + positive = GTEST_FLAG(filter).c_str(); // Whole string is a positive filter + negative = ""; + } else { + positive = std::string(p, dash); // Everything up to the dash + negative = std::string(dash + 1); // Everything after the dash + if (positive.empty()) { + // Treat '-test1' as the same as '*-test1' + positive = kUniversalFilter; + } + } + + // A filter is a colon-separated list of patterns. It matches a + // test if any pattern in it matches the test. + return (MatchesFilter(full_name, positive.c_str()) && + !MatchesFilter(full_name, negative.c_str())); +} + +#if GTEST_HAS_SEH +// Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the +// given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise. +// This function is useful as an __except condition. +int UnitTestOptions::GTestShouldProcessSEH(DWORD exception_code) { + // Google Test should handle a SEH exception if: + // 1. the user wants it to, AND + // 2. this is not a breakpoint exception, AND + // 3. this is not a C++ exception (VC++ implements them via SEH, + // apparently). + // + // SEH exception code for C++ exceptions. + // (see http://support.microsoft.com/kb/185294 for more information). + const DWORD kCxxExceptionCode = 0xe06d7363; + + bool should_handle = true; + + if (!GTEST_FLAG(catch_exceptions)) + should_handle = false; + else if (exception_code == EXCEPTION_BREAKPOINT) + should_handle = false; + else if (exception_code == kCxxExceptionCode) + should_handle = false; + + return should_handle ? EXCEPTION_EXECUTE_HANDLER : EXCEPTION_CONTINUE_SEARCH; +} +#endif // GTEST_HAS_SEH + +} // namespace internal + +// The c'tor sets this object as the test part result reporter used by +// Google Test. The 'result' parameter specifies where to report the +// results. Intercepts only failures from the current thread. +ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter( + TestPartResultArray* result) + : intercept_mode_(INTERCEPT_ONLY_CURRENT_THREAD), + result_(result) { + Init(); +} + +// The c'tor sets this object as the test part result reporter used by +// Google Test. The 'result' parameter specifies where to report the +// results. +ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter( + InterceptMode intercept_mode, TestPartResultArray* result) + : intercept_mode_(intercept_mode), + result_(result) { + Init(); +} + +void ScopedFakeTestPartResultReporter::Init() { + internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); + if (intercept_mode_ == INTERCEPT_ALL_THREADS) { + old_reporter_ = impl->GetGlobalTestPartResultReporter(); + impl->SetGlobalTestPartResultReporter(this); + } else { + old_reporter_ = impl->GetTestPartResultReporterForCurrentThread(); + impl->SetTestPartResultReporterForCurrentThread(this); + } +} + +// The d'tor restores the test part result reporter used by Google Test +// before. +ScopedFakeTestPartResultReporter::~ScopedFakeTestPartResultReporter() { + internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); + if (intercept_mode_ == INTERCEPT_ALL_THREADS) { + impl->SetGlobalTestPartResultReporter(old_reporter_); + } else { + impl->SetTestPartResultReporterForCurrentThread(old_reporter_); + } +} + +// Increments the test part result count and remembers the result. +// This method is from the TestPartResultReporterInterface interface. +void ScopedFakeTestPartResultReporter::ReportTestPartResult( + const TestPartResult& result) { + result_->Append(result); +} + +namespace internal { + +// Returns the type ID of ::testing::Test. We should always call this +// instead of GetTypeId< ::testing::Test>() to get the type ID of +// testing::Test. This is to work around a suspected linker bug when +// using Google Test as a framework on Mac OS X. The bug causes +// GetTypeId< ::testing::Test>() to return different values depending +// on whether the call is from the Google Test framework itself or +// from user test code. GetTestTypeId() is guaranteed to always +// return the same value, as it always calls GetTypeId<>() from the +// gtest.cc, which is within the Google Test framework. +TypeId GetTestTypeId() { + return GetTypeId(); +} + +// The value of GetTestTypeId() as seen from within the Google Test +// library. This is solely for testing GetTestTypeId(). +extern const TypeId kTestTypeIdInGoogleTest = GetTestTypeId(); + +// This predicate-formatter checks that 'results' contains a test part +// failure of the given type and that the failure message contains the +// given substring. +static AssertionResult HasOneFailure(const char* /* results_expr */, + const char* /* type_expr */, + const char* /* substr_expr */, + const TestPartResultArray& results, + TestPartResult::Type type, + const std::string& substr) { + const std::string expected(type == TestPartResult::kFatalFailure ? + "1 fatal failure" : + "1 non-fatal failure"); + Message msg; + if (results.size() != 1) { + msg << "Expected: " << expected << "\n" + << " Actual: " << results.size() << " failures"; + for (int i = 0; i < results.size(); i++) { + msg << "\n" << results.GetTestPartResult(i); + } + return AssertionFailure() << msg; + } + + const TestPartResult& r = results.GetTestPartResult(0); + if (r.type() != type) { + return AssertionFailure() << "Expected: " << expected << "\n" + << " Actual:\n" + << r; + } + + if (strstr(r.message(), substr.c_str()) == nullptr) { + return AssertionFailure() << "Expected: " << expected << " containing \"" + << substr << "\"\n" + << " Actual:\n" + << r; + } + + return AssertionSuccess(); +} + +// The constructor of SingleFailureChecker remembers where to look up +// test part results, what type of failure we expect, and what +// substring the failure message should contain. +SingleFailureChecker::SingleFailureChecker(const TestPartResultArray* results, + TestPartResult::Type type, + const std::string& substr) + : results_(results), type_(type), substr_(substr) {} + +// The destructor of SingleFailureChecker verifies that the given +// TestPartResultArray contains exactly one failure that has the given +// type and contains the given substring. If that's not the case, a +// non-fatal failure will be generated. +SingleFailureChecker::~SingleFailureChecker() { + EXPECT_PRED_FORMAT3(HasOneFailure, *results_, type_, substr_); +} + +DefaultGlobalTestPartResultReporter::DefaultGlobalTestPartResultReporter( + UnitTestImpl* unit_test) : unit_test_(unit_test) {} + +void DefaultGlobalTestPartResultReporter::ReportTestPartResult( + const TestPartResult& result) { + unit_test_->current_test_result()->AddTestPartResult(result); + unit_test_->listeners()->repeater()->OnTestPartResult(result); +} + +DefaultPerThreadTestPartResultReporter::DefaultPerThreadTestPartResultReporter( + UnitTestImpl* unit_test) : unit_test_(unit_test) {} + +void DefaultPerThreadTestPartResultReporter::ReportTestPartResult( + const TestPartResult& result) { + unit_test_->GetGlobalTestPartResultReporter()->ReportTestPartResult(result); +} + +// Returns the global test part result reporter. +TestPartResultReporterInterface* +UnitTestImpl::GetGlobalTestPartResultReporter() { + internal::MutexLock lock(&global_test_part_result_reporter_mutex_); + return global_test_part_result_repoter_; +} + +// Sets the global test part result reporter. +void UnitTestImpl::SetGlobalTestPartResultReporter( + TestPartResultReporterInterface* reporter) { + internal::MutexLock lock(&global_test_part_result_reporter_mutex_); + global_test_part_result_repoter_ = reporter; +} + +// Returns the test part result reporter for the current thread. +TestPartResultReporterInterface* +UnitTestImpl::GetTestPartResultReporterForCurrentThread() { + return per_thread_test_part_result_reporter_.get(); +} + +// Sets the test part result reporter for the current thread. +void UnitTestImpl::SetTestPartResultReporterForCurrentThread( + TestPartResultReporterInterface* reporter) { + per_thread_test_part_result_reporter_.set(reporter); +} + +// Gets the number of successful test suites. +int UnitTestImpl::successful_test_suite_count() const { + return CountIf(test_suites_, TestSuitePassed); +} + +// Gets the number of failed test suites. +int UnitTestImpl::failed_test_suite_count() const { + return CountIf(test_suites_, TestSuiteFailed); +} + +// Gets the number of all test suites. +int UnitTestImpl::total_test_suite_count() const { + return static_cast(test_suites_.size()); +} + +// Gets the number of all test suites that contain at least one test +// that should run. +int UnitTestImpl::test_suite_to_run_count() const { + return CountIf(test_suites_, ShouldRunTestSuite); +} + +// Gets the number of successful tests. +int UnitTestImpl::successful_test_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::successful_test_count); +} + +// Gets the number of skipped tests. +int UnitTestImpl::skipped_test_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::skipped_test_count); +} + +// Gets the number of failed tests. +int UnitTestImpl::failed_test_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::failed_test_count); +} + +// Gets the number of disabled tests that will be reported in the XML report. +int UnitTestImpl::reportable_disabled_test_count() const { + return SumOverTestSuiteList(test_suites_, + &TestSuite::reportable_disabled_test_count); +} + +// Gets the number of disabled tests. +int UnitTestImpl::disabled_test_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::disabled_test_count); +} + +// Gets the number of tests to be printed in the XML report. +int UnitTestImpl::reportable_test_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::reportable_test_count); +} + +// Gets the number of all tests. +int UnitTestImpl::total_test_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::total_test_count); +} + +// Gets the number of tests that should run. +int UnitTestImpl::test_to_run_count() const { + return SumOverTestSuiteList(test_suites_, &TestSuite::test_to_run_count); +} + +// Returns the current OS stack trace as an std::string. +// +// The maximum number of stack frames to be included is specified by +// the gtest_stack_trace_depth flag. The skip_count parameter +// specifies the number of top frames to be skipped, which doesn't +// count against the number of frames to be included. +// +// For example, if Foo() calls Bar(), which in turn calls +// CurrentOsStackTraceExceptTop(1), Foo() will be included in the +// trace but Bar() and CurrentOsStackTraceExceptTop() won't. +std::string UnitTestImpl::CurrentOsStackTraceExceptTop(int skip_count) { + return os_stack_trace_getter()->CurrentStackTrace( + static_cast(GTEST_FLAG(stack_trace_depth)), + skip_count + 1 + // Skips the user-specified number of frames plus this function + // itself. + ); // NOLINT +} + +// Returns the current time in milliseconds. +TimeInMillis GetTimeInMillis() { +#if GTEST_OS_WINDOWS_MOBILE || defined(__BORLANDC__) + // Difference between 1970-01-01 and 1601-01-01 in milliseconds. + // http://analogous.blogspot.com/2005/04/epoch.html + const TimeInMillis kJavaEpochToWinFileTimeDelta = + static_cast(116444736UL) * 100000UL; + const DWORD kTenthMicrosInMilliSecond = 10000; + + SYSTEMTIME now_systime; + FILETIME now_filetime; + ULARGE_INTEGER now_int64; + GetSystemTime(&now_systime); + if (SystemTimeToFileTime(&now_systime, &now_filetime)) { + now_int64.LowPart = now_filetime.dwLowDateTime; + now_int64.HighPart = now_filetime.dwHighDateTime; + now_int64.QuadPart = (now_int64.QuadPart / kTenthMicrosInMilliSecond) - + kJavaEpochToWinFileTimeDelta; + return now_int64.QuadPart; + } + return 0; +#elif GTEST_OS_WINDOWS && !GTEST_HAS_GETTIMEOFDAY_ + __timeb64 now; + + // MSVC 8 deprecates _ftime64(), so we want to suppress warning 4996 + // (deprecated function) there. + GTEST_DISABLE_MSC_DEPRECATED_PUSH_() + _ftime64(&now); + GTEST_DISABLE_MSC_DEPRECATED_POP_() + + return static_cast(now.time) * 1000 + now.millitm; +#elif GTEST_HAS_GETTIMEOFDAY_ + struct timeval now; + gettimeofday(&now, nullptr); + return static_cast(now.tv_sec) * 1000 + now.tv_usec / 1000; +#else +# error "Don't know how to get the current time on your system." +#endif +} + +// Utilities + +// class String. + +#if GTEST_OS_WINDOWS_MOBILE +// Creates a UTF-16 wide string from the given ANSI string, allocating +// memory using new. The caller is responsible for deleting the return +// value using delete[]. Returns the wide string, or NULL if the +// input is NULL. +LPCWSTR String::AnsiToUtf16(const char* ansi) { + if (!ansi) return nullptr; + const int length = strlen(ansi); + const int unicode_length = + MultiByteToWideChar(CP_ACP, 0, ansi, length, nullptr, 0); + WCHAR* unicode = new WCHAR[unicode_length + 1]; + MultiByteToWideChar(CP_ACP, 0, ansi, length, + unicode, unicode_length); + unicode[unicode_length] = 0; + return unicode; +} + +// Creates an ANSI string from the given wide string, allocating +// memory using new. The caller is responsible for deleting the return +// value using delete[]. Returns the ANSI string, or NULL if the +// input is NULL. +const char* String::Utf16ToAnsi(LPCWSTR utf16_str) { + if (!utf16_str) return nullptr; + const int ansi_length = WideCharToMultiByte(CP_ACP, 0, utf16_str, -1, nullptr, + 0, nullptr, nullptr); + char* ansi = new char[ansi_length + 1]; + WideCharToMultiByte(CP_ACP, 0, utf16_str, -1, ansi, ansi_length, nullptr, + nullptr); + ansi[ansi_length] = 0; + return ansi; +} + +#endif // GTEST_OS_WINDOWS_MOBILE + +// Compares two C strings. Returns true iff they have the same content. +// +// Unlike strcmp(), this function can handle NULL argument(s). A NULL +// C string is considered different to any non-NULL C string, +// including the empty string. +bool String::CStringEquals(const char * lhs, const char * rhs) { + if (lhs == nullptr) return rhs == nullptr; + + if (rhs == nullptr) return false; + + return strcmp(lhs, rhs) == 0; +} + +#if GTEST_HAS_STD_WSTRING + +// Converts an array of wide chars to a narrow string using the UTF-8 +// encoding, and streams the result to the given Message object. +static void StreamWideCharsToMessage(const wchar_t* wstr, size_t length, + Message* msg) { + for (size_t i = 0; i != length; ) { // NOLINT + if (wstr[i] != L'\0') { + *msg << WideStringToUtf8(wstr + i, static_cast(length - i)); + while (i != length && wstr[i] != L'\0') + i++; + } else { + *msg << '\0'; + i++; + } + } +} + +#endif // GTEST_HAS_STD_WSTRING + +void SplitString(const ::std::string& str, char delimiter, + ::std::vector< ::std::string>* dest) { + ::std::vector< ::std::string> parsed; + ::std::string::size_type pos = 0; + while (::testing::internal::AlwaysTrue()) { + const ::std::string::size_type colon = str.find(delimiter, pos); + if (colon == ::std::string::npos) { + parsed.push_back(str.substr(pos)); + break; + } else { + parsed.push_back(str.substr(pos, colon - pos)); + pos = colon + 1; + } + } + dest->swap(parsed); +} + +} // namespace internal + +// Constructs an empty Message. +// We allocate the stringstream separately because otherwise each use of +// ASSERT/EXPECT in a procedure adds over 200 bytes to the procedure's +// stack frame leading to huge stack frames in some cases; gcc does not reuse +// the stack space. +Message::Message() : ss_(new ::std::stringstream) { + // By default, we want there to be enough precision when printing + // a double to a Message. + *ss_ << std::setprecision(std::numeric_limits::digits10 + 2); +} + +// These two overloads allow streaming a wide C string to a Message +// using the UTF-8 encoding. +Message& Message::operator <<(const wchar_t* wide_c_str) { + return *this << internal::String::ShowWideCString(wide_c_str); +} +Message& Message::operator <<(wchar_t* wide_c_str) { + return *this << internal::String::ShowWideCString(wide_c_str); +} + +#if GTEST_HAS_STD_WSTRING +// Converts the given wide string to a narrow string using the UTF-8 +// encoding, and streams the result to this Message object. +Message& Message::operator <<(const ::std::wstring& wstr) { + internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this); + return *this; +} +#endif // GTEST_HAS_STD_WSTRING + +// Gets the text streamed to this object so far as an std::string. +// Each '\0' character in the buffer is replaced with "\\0". +std::string Message::GetString() const { + return internal::StringStreamToString(ss_.get()); +} + +// AssertionResult constructors. +// Used in EXPECT_TRUE/FALSE(assertion_result). +AssertionResult::AssertionResult(const AssertionResult& other) + : success_(other.success_), + message_(other.message_.get() != nullptr + ? new ::std::string(*other.message_) + : static_cast< ::std::string*>(nullptr)) {} + +// Swaps two AssertionResults. +void AssertionResult::swap(AssertionResult& other) { + using std::swap; + swap(success_, other.success_); + swap(message_, other.message_); +} + +// Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE. +AssertionResult AssertionResult::operator!() const { + AssertionResult negation(!success_); + if (message_.get() != nullptr) negation << *message_; + return negation; +} + +// Makes a successful assertion result. +AssertionResult AssertionSuccess() { + return AssertionResult(true); +} + +// Makes a failed assertion result. +AssertionResult AssertionFailure() { + return AssertionResult(false); +} + +// Makes a failed assertion result with the given failure message. +// Deprecated; use AssertionFailure() << message. +AssertionResult AssertionFailure(const Message& message) { + return AssertionFailure() << message; +} + +namespace internal { + +namespace edit_distance { +std::vector CalculateOptimalEdits(const std::vector& left, + const std::vector& right) { + std::vector > costs( + left.size() + 1, std::vector(right.size() + 1)); + std::vector > best_move( + left.size() + 1, std::vector(right.size() + 1)); + + // Populate for empty right. + for (size_t l_i = 0; l_i < costs.size(); ++l_i) { + costs[l_i][0] = static_cast(l_i); + best_move[l_i][0] = kRemove; + } + // Populate for empty left. + for (size_t r_i = 1; r_i < costs[0].size(); ++r_i) { + costs[0][r_i] = static_cast(r_i); + best_move[0][r_i] = kAdd; + } + + for (size_t l_i = 0; l_i < left.size(); ++l_i) { + for (size_t r_i = 0; r_i < right.size(); ++r_i) { + if (left[l_i] == right[r_i]) { + // Found a match. Consume it. + costs[l_i + 1][r_i + 1] = costs[l_i][r_i]; + best_move[l_i + 1][r_i + 1] = kMatch; + continue; + } + + const double add = costs[l_i + 1][r_i]; + const double remove = costs[l_i][r_i + 1]; + const double replace = costs[l_i][r_i]; + if (add < remove && add < replace) { + costs[l_i + 1][r_i + 1] = add + 1; + best_move[l_i + 1][r_i + 1] = kAdd; + } else if (remove < add && remove < replace) { + costs[l_i + 1][r_i + 1] = remove + 1; + best_move[l_i + 1][r_i + 1] = kRemove; + } else { + // We make replace a little more expensive than add/remove to lower + // their priority. + costs[l_i + 1][r_i + 1] = replace + 1.00001; + best_move[l_i + 1][r_i + 1] = kReplace; + } + } + } + + // Reconstruct the best path. We do it in reverse order. + std::vector best_path; + for (size_t l_i = left.size(), r_i = right.size(); l_i > 0 || r_i > 0;) { + EditType move = best_move[l_i][r_i]; + best_path.push_back(move); + l_i -= move != kAdd; + r_i -= move != kRemove; + } + std::reverse(best_path.begin(), best_path.end()); + return best_path; +} + +namespace { + +// Helper class to convert string into ids with deduplication. +class InternalStrings { + public: + size_t GetId(const std::string& str) { + IdMap::iterator it = ids_.find(str); + if (it != ids_.end()) return it->second; + size_t id = ids_.size(); + return ids_[str] = id; + } + + private: + typedef std::map IdMap; + IdMap ids_; +}; + +} // namespace + +std::vector CalculateOptimalEdits( + const std::vector& left, + const std::vector& right) { + std::vector left_ids, right_ids; + { + InternalStrings intern_table; + for (size_t i = 0; i < left.size(); ++i) { + left_ids.push_back(intern_table.GetId(left[i])); + } + for (size_t i = 0; i < right.size(); ++i) { + right_ids.push_back(intern_table.GetId(right[i])); + } + } + return CalculateOptimalEdits(left_ids, right_ids); +} + +namespace { + +// Helper class that holds the state for one hunk and prints it out to the +// stream. +// It reorders adds/removes when possible to group all removes before all +// adds. It also adds the hunk header before printint into the stream. +class Hunk { + public: + Hunk(size_t left_start, size_t right_start) + : left_start_(left_start), + right_start_(right_start), + adds_(), + removes_(), + common_() {} + + void PushLine(char edit, const char* line) { + switch (edit) { + case ' ': + ++common_; + FlushEdits(); + hunk_.push_back(std::make_pair(' ', line)); + break; + case '-': + ++removes_; + hunk_removes_.push_back(std::make_pair('-', line)); + break; + case '+': + ++adds_; + hunk_adds_.push_back(std::make_pair('+', line)); + break; + } + } + + void PrintTo(std::ostream* os) { + PrintHeader(os); + FlushEdits(); + for (std::list >::const_iterator it = + hunk_.begin(); + it != hunk_.end(); ++it) { + *os << it->first << it->second << "\n"; + } + } + + bool has_edits() const { return adds_ || removes_; } + + private: + void FlushEdits() { + hunk_.splice(hunk_.end(), hunk_removes_); + hunk_.splice(hunk_.end(), hunk_adds_); + } + + // Print a unified diff header for one hunk. + // The format is + // "@@ -, +, @@" + // where the left/right parts are omitted if unnecessary. + void PrintHeader(std::ostream* ss) const { + *ss << "@@ "; + if (removes_) { + *ss << "-" << left_start_ << "," << (removes_ + common_); + } + if (removes_ && adds_) { + *ss << " "; + } + if (adds_) { + *ss << "+" << right_start_ << "," << (adds_ + common_); + } + *ss << " @@\n"; + } + + size_t left_start_, right_start_; + size_t adds_, removes_, common_; + std::list > hunk_, hunk_adds_, hunk_removes_; +}; + +} // namespace + +// Create a list of diff hunks in Unified diff format. +// Each hunk has a header generated by PrintHeader above plus a body with +// lines prefixed with ' ' for no change, '-' for deletion and '+' for +// addition. +// 'context' represents the desired unchanged prefix/suffix around the diff. +// If two hunks are close enough that their contexts overlap, then they are +// joined into one hunk. +std::string CreateUnifiedDiff(const std::vector& left, + const std::vector& right, + size_t context) { + const std::vector edits = CalculateOptimalEdits(left, right); + + size_t l_i = 0, r_i = 0, edit_i = 0; + std::stringstream ss; + while (edit_i < edits.size()) { + // Find first edit. + while (edit_i < edits.size() && edits[edit_i] == kMatch) { + ++l_i; + ++r_i; + ++edit_i; + } + + // Find the first line to include in the hunk. + const size_t prefix_context = std::min(l_i, context); + Hunk hunk(l_i - prefix_context + 1, r_i - prefix_context + 1); + for (size_t i = prefix_context; i > 0; --i) { + hunk.PushLine(' ', left[l_i - i].c_str()); + } + + // Iterate the edits until we found enough suffix for the hunk or the input + // is over. + size_t n_suffix = 0; + for (; edit_i < edits.size(); ++edit_i) { + if (n_suffix >= context) { + // Continue only if the next hunk is very close. + auto it = edits.begin() + static_cast(edit_i); + while (it != edits.end() && *it == kMatch) ++it; + if (it == edits.end() || + static_cast(it - edits.begin()) - edit_i >= context) { + // There is no next edit or it is too far away. + break; + } + } + + EditType edit = edits[edit_i]; + // Reset count when a non match is found. + n_suffix = edit == kMatch ? n_suffix + 1 : 0; + + if (edit == kMatch || edit == kRemove || edit == kReplace) { + hunk.PushLine(edit == kMatch ? ' ' : '-', left[l_i].c_str()); + } + if (edit == kAdd || edit == kReplace) { + hunk.PushLine('+', right[r_i].c_str()); + } + + // Advance indices, depending on edit type. + l_i += edit != kAdd; + r_i += edit != kRemove; + } + + if (!hunk.has_edits()) { + // We are done. We don't want this hunk. + break; + } + + hunk.PrintTo(&ss); + } + return ss.str(); +} + +} // namespace edit_distance + +namespace { + +// The string representation of the values received in EqFailure() are already +// escaped. Split them on escaped '\n' boundaries. Leave all other escaped +// characters the same. +std::vector SplitEscapedString(const std::string& str) { + std::vector lines; + size_t start = 0, end = str.size(); + if (end > 2 && str[0] == '"' && str[end - 1] == '"') { + ++start; + --end; + } + bool escaped = false; + for (size_t i = start; i + 1 < end; ++i) { + if (escaped) { + escaped = false; + if (str[i] == 'n') { + lines.push_back(str.substr(start, i - start - 1)); + start = i + 1; + } + } else { + escaped = str[i] == '\\'; + } + } + lines.push_back(str.substr(start, end - start)); + return lines; +} + +} // namespace + +// Constructs and returns the message for an equality assertion +// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure. +// +// The first four parameters are the expressions used in the assertion +// and their values, as strings. For example, for ASSERT_EQ(foo, bar) +// where foo is 5 and bar is 6, we have: +// +// lhs_expression: "foo" +// rhs_expression: "bar" +// lhs_value: "5" +// rhs_value: "6" +// +// The ignoring_case parameter is true iff the assertion is a +// *_STRCASEEQ*. When it's true, the string "Ignoring case" will +// be inserted into the message. +AssertionResult EqFailure(const char* lhs_expression, + const char* rhs_expression, + const std::string& lhs_value, + const std::string& rhs_value, + bool ignoring_case) { + Message msg; + msg << "Expected equality of these values:"; + msg << "\n " << lhs_expression; + if (lhs_value != lhs_expression) { + msg << "\n Which is: " << lhs_value; + } + msg << "\n " << rhs_expression; + if (rhs_value != rhs_expression) { + msg << "\n Which is: " << rhs_value; + } + + if (ignoring_case) { + msg << "\nIgnoring case"; + } + + if (!lhs_value.empty() && !rhs_value.empty()) { + const std::vector lhs_lines = + SplitEscapedString(lhs_value); + const std::vector rhs_lines = + SplitEscapedString(rhs_value); + if (lhs_lines.size() > 1 || rhs_lines.size() > 1) { + msg << "\nWith diff:\n" + << edit_distance::CreateUnifiedDiff(lhs_lines, rhs_lines); + } + } + + return AssertionFailure() << msg; +} + +// Constructs a failure message for Boolean assertions such as EXPECT_TRUE. +std::string GetBoolAssertionFailureMessage( + const AssertionResult& assertion_result, + const char* expression_text, + const char* actual_predicate_value, + const char* expected_predicate_value) { + const char* actual_message = assertion_result.message(); + Message msg; + msg << "Value of: " << expression_text + << "\n Actual: " << actual_predicate_value; + if (actual_message[0] != '\0') + msg << " (" << actual_message << ")"; + msg << "\nExpected: " << expected_predicate_value; + return msg.GetString(); +} + +// Helper function for implementing ASSERT_NEAR. +AssertionResult DoubleNearPredFormat(const char* expr1, + const char* expr2, + const char* abs_error_expr, + double val1, + double val2, + double abs_error) { + const double diff = fabs(val1 - val2); + if (diff <= abs_error) return AssertionSuccess(); + + return AssertionFailure() + << "The difference between " << expr1 << " and " << expr2 + << " is " << diff << ", which exceeds " << abs_error_expr << ", where\n" + << expr1 << " evaluates to " << val1 << ",\n" + << expr2 << " evaluates to " << val2 << ", and\n" + << abs_error_expr << " evaluates to " << abs_error << "."; +} + + +// Helper template for implementing FloatLE() and DoubleLE(). +template +AssertionResult FloatingPointLE(const char* expr1, + const char* expr2, + RawType val1, + RawType val2) { + // Returns success if val1 is less than val2, + if (val1 < val2) { + return AssertionSuccess(); + } + + // or if val1 is almost equal to val2. + const FloatingPoint lhs(val1), rhs(val2); + if (lhs.AlmostEquals(rhs)) { + return AssertionSuccess(); + } + + // Note that the above two checks will both fail if either val1 or + // val2 is NaN, as the IEEE floating-point standard requires that + // any predicate involving a NaN must return false. + + ::std::stringstream val1_ss; + val1_ss << std::setprecision(std::numeric_limits::digits10 + 2) + << val1; + + ::std::stringstream val2_ss; + val2_ss << std::setprecision(std::numeric_limits::digits10 + 2) + << val2; + + return AssertionFailure() + << "Expected: (" << expr1 << ") <= (" << expr2 << ")\n" + << " Actual: " << StringStreamToString(&val1_ss) << " vs " + << StringStreamToString(&val2_ss); +} + +} // namespace internal + +// Asserts that val1 is less than, or almost equal to, val2. Fails +// otherwise. In particular, it fails if either val1 or val2 is NaN. +AssertionResult FloatLE(const char* expr1, const char* expr2, + float val1, float val2) { + return internal::FloatingPointLE(expr1, expr2, val1, val2); +} + +// Asserts that val1 is less than, or almost equal to, val2. Fails +// otherwise. In particular, it fails if either val1 or val2 is NaN. +AssertionResult DoubleLE(const char* expr1, const char* expr2, + double val1, double val2) { + return internal::FloatingPointLE(expr1, expr2, val1, val2); +} + +namespace internal { + +// The helper function for {ASSERT|EXPECT}_EQ with int or enum +// arguments. +AssertionResult CmpHelperEQ(const char* lhs_expression, + const char* rhs_expression, + BiggestInt lhs, + BiggestInt rhs) { + if (lhs == rhs) { + return AssertionSuccess(); + } + + return EqFailure(lhs_expression, + rhs_expression, + FormatForComparisonFailureMessage(lhs, rhs), + FormatForComparisonFailureMessage(rhs, lhs), + false); +} + +// A macro for implementing the helper functions needed to implement +// ASSERT_?? and EXPECT_?? with integer or enum arguments. It is here +// just to avoid copy-and-paste of similar code. +#define GTEST_IMPL_CMP_HELPER_(op_name, op)\ +AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \ + BiggestInt val1, BiggestInt val2) {\ + if (val1 op val2) {\ + return AssertionSuccess();\ + } else {\ + return AssertionFailure() \ + << "Expected: (" << expr1 << ") " #op " (" << expr2\ + << "), actual: " << FormatForComparisonFailureMessage(val1, val2)\ + << " vs " << FormatForComparisonFailureMessage(val2, val1);\ + }\ +} + +// Implements the helper function for {ASSERT|EXPECT}_NE with int or +// enum arguments. +GTEST_IMPL_CMP_HELPER_(NE, !=) +// Implements the helper function for {ASSERT|EXPECT}_LE with int or +// enum arguments. +GTEST_IMPL_CMP_HELPER_(LE, <=) +// Implements the helper function for {ASSERT|EXPECT}_LT with int or +// enum arguments. +GTEST_IMPL_CMP_HELPER_(LT, < ) +// Implements the helper function for {ASSERT|EXPECT}_GE with int or +// enum arguments. +GTEST_IMPL_CMP_HELPER_(GE, >=) +// Implements the helper function for {ASSERT|EXPECT}_GT with int or +// enum arguments. +GTEST_IMPL_CMP_HELPER_(GT, > ) + +#undef GTEST_IMPL_CMP_HELPER_ + +// The helper function for {ASSERT|EXPECT}_STREQ. +AssertionResult CmpHelperSTREQ(const char* lhs_expression, + const char* rhs_expression, + const char* lhs, + const char* rhs) { + if (String::CStringEquals(lhs, rhs)) { + return AssertionSuccess(); + } + + return EqFailure(lhs_expression, + rhs_expression, + PrintToString(lhs), + PrintToString(rhs), + false); +} + +// The helper function for {ASSERT|EXPECT}_STRCASEEQ. +AssertionResult CmpHelperSTRCASEEQ(const char* lhs_expression, + const char* rhs_expression, + const char* lhs, + const char* rhs) { + if (String::CaseInsensitiveCStringEquals(lhs, rhs)) { + return AssertionSuccess(); + } + + return EqFailure(lhs_expression, + rhs_expression, + PrintToString(lhs), + PrintToString(rhs), + true); +} + +// The helper function for {ASSERT|EXPECT}_STRNE. +AssertionResult CmpHelperSTRNE(const char* s1_expression, + const char* s2_expression, + const char* s1, + const char* s2) { + if (!String::CStringEquals(s1, s2)) { + return AssertionSuccess(); + } else { + return AssertionFailure() << "Expected: (" << s1_expression << ") != (" + << s2_expression << "), actual: \"" + << s1 << "\" vs \"" << s2 << "\""; + } +} + +// The helper function for {ASSERT|EXPECT}_STRCASENE. +AssertionResult CmpHelperSTRCASENE(const char* s1_expression, + const char* s2_expression, + const char* s1, + const char* s2) { + if (!String::CaseInsensitiveCStringEquals(s1, s2)) { + return AssertionSuccess(); + } else { + return AssertionFailure() + << "Expected: (" << s1_expression << ") != (" + << s2_expression << ") (ignoring case), actual: \"" + << s1 << "\" vs \"" << s2 << "\""; + } +} + +} // namespace internal + +namespace { + +// Helper functions for implementing IsSubString() and IsNotSubstring(). + +// This group of overloaded functions return true iff needle is a +// substring of haystack. NULL is considered a substring of itself +// only. + +bool IsSubstringPred(const char* needle, const char* haystack) { + if (needle == nullptr || haystack == nullptr) return needle == haystack; + + return strstr(haystack, needle) != nullptr; +} + +bool IsSubstringPred(const wchar_t* needle, const wchar_t* haystack) { + if (needle == nullptr || haystack == nullptr) return needle == haystack; + + return wcsstr(haystack, needle) != nullptr; +} + +// StringType here can be either ::std::string or ::std::wstring. +template +bool IsSubstringPred(const StringType& needle, + const StringType& haystack) { + return haystack.find(needle) != StringType::npos; +} + +// This function implements either IsSubstring() or IsNotSubstring(), +// depending on the value of the expected_to_be_substring parameter. +// StringType here can be const char*, const wchar_t*, ::std::string, +// or ::std::wstring. +template +AssertionResult IsSubstringImpl( + bool expected_to_be_substring, + const char* needle_expr, const char* haystack_expr, + const StringType& needle, const StringType& haystack) { + if (IsSubstringPred(needle, haystack) == expected_to_be_substring) + return AssertionSuccess(); + + const bool is_wide_string = sizeof(needle[0]) > 1; + const char* const begin_string_quote = is_wide_string ? "L\"" : "\""; + return AssertionFailure() + << "Value of: " << needle_expr << "\n" + << " Actual: " << begin_string_quote << needle << "\"\n" + << "Expected: " << (expected_to_be_substring ? "" : "not ") + << "a substring of " << haystack_expr << "\n" + << "Which is: " << begin_string_quote << haystack << "\""; +} + +} // namespace + +// IsSubstring() and IsNotSubstring() check whether needle is a +// substring of haystack (NULL is considered a substring of itself +// only), and return an appropriate error message when they fail. + +AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const char* needle, const char* haystack) { + return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); +} + +AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const wchar_t* needle, const wchar_t* haystack) { + return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); +} + +AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const char* needle, const char* haystack) { + return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); +} + +AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const wchar_t* needle, const wchar_t* haystack) { + return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); +} + +AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::string& needle, const ::std::string& haystack) { + return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); +} + +AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::string& needle, const ::std::string& haystack) { + return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); +} + +#if GTEST_HAS_STD_WSTRING +AssertionResult IsSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::wstring& needle, const ::std::wstring& haystack) { + return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack); +} + +AssertionResult IsNotSubstring( + const char* needle_expr, const char* haystack_expr, + const ::std::wstring& needle, const ::std::wstring& haystack) { + return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack); +} +#endif // GTEST_HAS_STD_WSTRING + +namespace internal { + +#if GTEST_OS_WINDOWS + +namespace { + +// Helper function for IsHRESULT{SuccessFailure} predicates +AssertionResult HRESULTFailureHelper(const char* expr, + const char* expected, + long hr) { // NOLINT +# if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_TV_TITLE + + // Windows CE doesn't support FormatMessage. + const char error_text[] = ""; + +# else + + // Looks up the human-readable system message for the HRESULT code + // and since we're not passing any params to FormatMessage, we don't + // want inserts expanded. + const DWORD kFlags = FORMAT_MESSAGE_FROM_SYSTEM | + FORMAT_MESSAGE_IGNORE_INSERTS; + const DWORD kBufSize = 4096; + // Gets the system's human readable message string for this HRESULT. + char error_text[kBufSize] = { '\0' }; + DWORD message_length = ::FormatMessageA(kFlags, + 0, // no source, we're asking system + hr, // the error + 0, // no line width restrictions + error_text, // output buffer + kBufSize, // buf size + nullptr); // no arguments for inserts + // Trims tailing white space (FormatMessage leaves a trailing CR-LF) + for (; message_length && IsSpace(error_text[message_length - 1]); + --message_length) { + error_text[message_length - 1] = '\0'; + } + +# endif // GTEST_OS_WINDOWS_MOBILE + + const std::string error_hex("0x" + String::FormatHexInt(hr)); + return ::testing::AssertionFailure() + << "Expected: " << expr << " " << expected << ".\n" + << " Actual: " << error_hex << " " << error_text << "\n"; +} + +} // namespace + +AssertionResult IsHRESULTSuccess(const char* expr, long hr) { // NOLINT + if (SUCCEEDED(hr)) { + return AssertionSuccess(); + } + return HRESULTFailureHelper(expr, "succeeds", hr); +} + +AssertionResult IsHRESULTFailure(const char* expr, long hr) { // NOLINT + if (FAILED(hr)) { + return AssertionSuccess(); + } + return HRESULTFailureHelper(expr, "fails", hr); +} + +#endif // GTEST_OS_WINDOWS + +// Utility functions for encoding Unicode text (wide strings) in +// UTF-8. + +// A Unicode code-point can have up to 21 bits, and is encoded in UTF-8 +// like this: +// +// Code-point length Encoding +// 0 - 7 bits 0xxxxxxx +// 8 - 11 bits 110xxxxx 10xxxxxx +// 12 - 16 bits 1110xxxx 10xxxxxx 10xxxxxx +// 17 - 21 bits 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx + +// The maximum code-point a one-byte UTF-8 sequence can represent. +const UInt32 kMaxCodePoint1 = (static_cast(1) << 7) - 1; + +// The maximum code-point a two-byte UTF-8 sequence can represent. +const UInt32 kMaxCodePoint2 = (static_cast(1) << (5 + 6)) - 1; + +// The maximum code-point a three-byte UTF-8 sequence can represent. +const UInt32 kMaxCodePoint3 = (static_cast(1) << (4 + 2*6)) - 1; + +// The maximum code-point a four-byte UTF-8 sequence can represent. +const UInt32 kMaxCodePoint4 = (static_cast(1) << (3 + 3*6)) - 1; + +// Chops off the n lowest bits from a bit pattern. Returns the n +// lowest bits. As a side effect, the original bit pattern will be +// shifted to the right by n bits. +inline UInt32 ChopLowBits(UInt32* bits, int n) { + const UInt32 low_bits = *bits & ((static_cast(1) << n) - 1); + *bits >>= n; + return low_bits; +} + +// Converts a Unicode code point to a narrow string in UTF-8 encoding. +// code_point parameter is of type UInt32 because wchar_t may not be +// wide enough to contain a code point. +// If the code_point is not a valid Unicode code point +// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted +// to "(Invalid Unicode 0xXXXXXXXX)". +std::string CodePointToUtf8(UInt32 code_point) { + if (code_point > kMaxCodePoint4) { + return "(Invalid Unicode 0x" + String::FormatHexUInt32(code_point) + ")"; + } + + char str[5]; // Big enough for the largest valid code point. + if (code_point <= kMaxCodePoint1) { + str[1] = '\0'; + str[0] = static_cast(code_point); // 0xxxxxxx + } else if (code_point <= kMaxCodePoint2) { + str[2] = '\0'; + str[1] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx + str[0] = static_cast(0xC0 | code_point); // 110xxxxx + } else if (code_point <= kMaxCodePoint3) { + str[3] = '\0'; + str[2] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx + str[1] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx + str[0] = static_cast(0xE0 | code_point); // 1110xxxx + } else { // code_point <= kMaxCodePoint4 + str[4] = '\0'; + str[3] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx + str[2] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx + str[1] = static_cast(0x80 | ChopLowBits(&code_point, 6)); // 10xxxxxx + str[0] = static_cast(0xF0 | code_point); // 11110xxx + } + return str; +} + +// The following two functions only make sense if the system +// uses UTF-16 for wide string encoding. All supported systems +// with 16 bit wchar_t (Windows, Cygwin) do use UTF-16. + +// Determines if the arguments constitute UTF-16 surrogate pair +// and thus should be combined into a single Unicode code point +// using CreateCodePointFromUtf16SurrogatePair. +inline bool IsUtf16SurrogatePair(wchar_t first, wchar_t second) { + return sizeof(wchar_t) == 2 && + (first & 0xFC00) == 0xD800 && (second & 0xFC00) == 0xDC00; +} + +// Creates a Unicode code point from UTF16 surrogate pair. +inline UInt32 CreateCodePointFromUtf16SurrogatePair(wchar_t first, + wchar_t second) { + const auto first_u = static_cast(first); + const auto second_u = static_cast(second); + const UInt32 mask = (1 << 10) - 1; + return (sizeof(wchar_t) == 2) + ? (((first_u & mask) << 10) | (second_u & mask)) + 0x10000 + : + // This function should not be called when the condition is + // false, but we provide a sensible default in case it is. + first_u; +} + +// Converts a wide string to a narrow string in UTF-8 encoding. +// The wide string is assumed to have the following encoding: +// UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin) +// UTF-32 if sizeof(wchar_t) == 4 (on Linux) +// Parameter str points to a null-terminated wide string. +// Parameter num_chars may additionally limit the number +// of wchar_t characters processed. -1 is used when the entire string +// should be processed. +// If the string contains code points that are not valid Unicode code points +// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output +// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding +// and contains invalid UTF-16 surrogate pairs, values in those pairs +// will be encoded as individual Unicode characters from Basic Normal Plane. +std::string WideStringToUtf8(const wchar_t* str, int num_chars) { + if (num_chars == -1) + num_chars = static_cast(wcslen(str)); + + ::std::stringstream stream; + for (int i = 0; i < num_chars; ++i) { + UInt32 unicode_code_point; + + if (str[i] == L'\0') { + break; + } else if (i + 1 < num_chars && IsUtf16SurrogatePair(str[i], str[i + 1])) { + unicode_code_point = CreateCodePointFromUtf16SurrogatePair(str[i], + str[i + 1]); + i++; + } else { + unicode_code_point = static_cast(str[i]); + } + + stream << CodePointToUtf8(unicode_code_point); + } + return StringStreamToString(&stream); +} + +// Converts a wide C string to an std::string using the UTF-8 encoding. +// NULL will be converted to "(null)". +std::string String::ShowWideCString(const wchar_t * wide_c_str) { + if (wide_c_str == nullptr) return "(null)"; + + return internal::WideStringToUtf8(wide_c_str, -1); +} + +// Compares two wide C strings. Returns true iff they have the same +// content. +// +// Unlike wcscmp(), this function can handle NULL argument(s). A NULL +// C string is considered different to any non-NULL C string, +// including the empty string. +bool String::WideCStringEquals(const wchar_t * lhs, const wchar_t * rhs) { + if (lhs == nullptr) return rhs == nullptr; + + if (rhs == nullptr) return false; + + return wcscmp(lhs, rhs) == 0; +} + +// Helper function for *_STREQ on wide strings. +AssertionResult CmpHelperSTREQ(const char* lhs_expression, + const char* rhs_expression, + const wchar_t* lhs, + const wchar_t* rhs) { + if (String::WideCStringEquals(lhs, rhs)) { + return AssertionSuccess(); + } + + return EqFailure(lhs_expression, + rhs_expression, + PrintToString(lhs), + PrintToString(rhs), + false); +} + +// Helper function for *_STRNE on wide strings. +AssertionResult CmpHelperSTRNE(const char* s1_expression, + const char* s2_expression, + const wchar_t* s1, + const wchar_t* s2) { + if (!String::WideCStringEquals(s1, s2)) { + return AssertionSuccess(); + } + + return AssertionFailure() << "Expected: (" << s1_expression << ") != (" + << s2_expression << "), actual: " + << PrintToString(s1) + << " vs " << PrintToString(s2); +} + +// Compares two C strings, ignoring case. Returns true iff they have +// the same content. +// +// Unlike strcasecmp(), this function can handle NULL argument(s). A +// NULL C string is considered different to any non-NULL C string, +// including the empty string. +bool String::CaseInsensitiveCStringEquals(const char * lhs, const char * rhs) { + if (lhs == nullptr) return rhs == nullptr; + if (rhs == nullptr) return false; + return posix::StrCaseCmp(lhs, rhs) == 0; +} + + // Compares two wide C strings, ignoring case. Returns true iff they + // have the same content. + // + // Unlike wcscasecmp(), this function can handle NULL argument(s). + // A NULL C string is considered different to any non-NULL wide C string, + // including the empty string. + // NB: The implementations on different platforms slightly differ. + // On windows, this method uses _wcsicmp which compares according to LC_CTYPE + // environment variable. On GNU platform this method uses wcscasecmp + // which compares according to LC_CTYPE category of the current locale. + // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the + // current locale. +bool String::CaseInsensitiveWideCStringEquals(const wchar_t* lhs, + const wchar_t* rhs) { + if (lhs == nullptr) return rhs == nullptr; + + if (rhs == nullptr) return false; + +#if GTEST_OS_WINDOWS + return _wcsicmp(lhs, rhs) == 0; +#elif GTEST_OS_LINUX && !GTEST_OS_LINUX_ANDROID + return wcscasecmp(lhs, rhs) == 0; +#else + // Android, Mac OS X and Cygwin don't define wcscasecmp. + // Other unknown OSes may not define it either. + wint_t left, right; + do { + left = towlower(*lhs++); + right = towlower(*rhs++); + } while (left && left == right); + return left == right; +#endif // OS selector +} + +// Returns true iff str ends with the given suffix, ignoring case. +// Any string is considered to end with an empty suffix. +bool String::EndsWithCaseInsensitive( + const std::string& str, const std::string& suffix) { + const size_t str_len = str.length(); + const size_t suffix_len = suffix.length(); + return (str_len >= suffix_len) && + CaseInsensitiveCStringEquals(str.c_str() + str_len - suffix_len, + suffix.c_str()); +} + +// Formats an int value as "%02d". +std::string String::FormatIntWidth2(int value) { + std::stringstream ss; + ss << std::setfill('0') << std::setw(2) << value; + return ss.str(); +} + +// Formats an int value as "%X". +std::string String::FormatHexUInt32(UInt32 value) { + std::stringstream ss; + ss << std::hex << std::uppercase << value; + return ss.str(); +} + +// Formats an int value as "%X". +std::string String::FormatHexInt(int value) { + return FormatHexUInt32(static_cast(value)); +} + +// Formats a byte as "%02X". +std::string String::FormatByte(unsigned char value) { + std::stringstream ss; + ss << std::setfill('0') << std::setw(2) << std::hex << std::uppercase + << static_cast(value); + return ss.str(); +} + +// Converts the buffer in a stringstream to an std::string, converting NUL +// bytes to "\\0" along the way. +std::string StringStreamToString(::std::stringstream* ss) { + const ::std::string& str = ss->str(); + const char* const start = str.c_str(); + const char* const end = start + str.length(); + + std::string result; + result.reserve(static_cast(2 * (end - start))); + for (const char* ch = start; ch != end; ++ch) { + if (*ch == '\0') { + result += "\\0"; // Replaces NUL with "\\0"; + } else { + result += *ch; + } + } + + return result; +} + +// Appends the user-supplied message to the Google-Test-generated message. +std::string AppendUserMessage(const std::string& gtest_msg, + const Message& user_msg) { + // Appends the user message if it's non-empty. + const std::string user_msg_string = user_msg.GetString(); + if (user_msg_string.empty()) { + return gtest_msg; + } + + return gtest_msg + "\n" + user_msg_string; +} + +} // namespace internal + +// class TestResult + +// Creates an empty TestResult. +TestResult::TestResult() + : death_test_count_(0), + elapsed_time_(0) { +} + +// D'tor. +TestResult::~TestResult() { +} + +// Returns the i-th test part result among all the results. i can +// range from 0 to total_part_count() - 1. If i is not in that range, +// aborts the program. +const TestPartResult& TestResult::GetTestPartResult(int i) const { + if (i < 0 || i >= total_part_count()) + internal::posix::Abort(); + return test_part_results_.at(static_cast(i)); +} + +// Returns the i-th test property. i can range from 0 to +// test_property_count() - 1. If i is not in that range, aborts the +// program. +const TestProperty& TestResult::GetTestProperty(int i) const { + if (i < 0 || i >= test_property_count()) + internal::posix::Abort(); + return test_properties_.at(static_cast(i)); +} + +// Clears the test part results. +void TestResult::ClearTestPartResults() { + test_part_results_.clear(); +} + +// Adds a test part result to the list. +void TestResult::AddTestPartResult(const TestPartResult& test_part_result) { + test_part_results_.push_back(test_part_result); +} + +// Adds a test property to the list. If a property with the same key as the +// supplied property is already represented, the value of this test_property +// replaces the old value for that key. +void TestResult::RecordProperty(const std::string& xml_element, + const TestProperty& test_property) { + if (!ValidateTestProperty(xml_element, test_property)) { + return; + } + internal::MutexLock lock(&test_properites_mutex_); + const std::vector::iterator property_with_matching_key = + std::find_if(test_properties_.begin(), test_properties_.end(), + internal::TestPropertyKeyIs(test_property.key())); + if (property_with_matching_key == test_properties_.end()) { + test_properties_.push_back(test_property); + return; + } + property_with_matching_key->SetValue(test_property.value()); +} + +// The list of reserved attributes used in the element of XML +// output. +static const char* const kReservedTestSuitesAttributes[] = { + "disabled", + "errors", + "failures", + "name", + "random_seed", + "tests", + "time", + "timestamp" +}; + +// The list of reserved attributes used in the element of XML +// output. +static const char* const kReservedTestSuiteAttributes[] = { + "disabled", + "errors", + "failures", + "name", + "tests", + "time" +}; + +// The list of reserved attributes used in the element of XML output. +static const char* const kReservedTestCaseAttributes[] = { + "classname", "name", "status", "time", "type_param", + "value_param", "file", "line"}; + +// Use a slightly different set for allowed output to ensure existing tests can +// still RecordProperty("result") +static const char* const kReservedOutputTestCaseAttributes[] = { + "classname", "name", "status", "time", "type_param", + "value_param", "file", "line", "result"}; + +template +std::vector ArrayAsVector(const char* const (&array)[kSize]) { + return std::vector(array, array + kSize); +} + +static std::vector GetReservedAttributesForElement( + const std::string& xml_element) { + if (xml_element == "testsuites") { + return ArrayAsVector(kReservedTestSuitesAttributes); + } else if (xml_element == "testsuite") { + return ArrayAsVector(kReservedTestSuiteAttributes); + } else if (xml_element == "testcase") { + return ArrayAsVector(kReservedTestCaseAttributes); + } else { + GTEST_CHECK_(false) << "Unrecognized xml_element provided: " << xml_element; + } + // This code is unreachable but some compilers may not realizes that. + return std::vector(); +} + +// TODO(jdesprez): Merge the two getReserved attributes once skip is improved +static std::vector GetReservedOutputAttributesForElement( + const std::string& xml_element) { + if (xml_element == "testsuites") { + return ArrayAsVector(kReservedTestSuitesAttributes); + } else if (xml_element == "testsuite") { + return ArrayAsVector(kReservedTestSuiteAttributes); + } else if (xml_element == "testcase") { + return ArrayAsVector(kReservedOutputTestCaseAttributes); + } else { + GTEST_CHECK_(false) << "Unrecognized xml_element provided: " << xml_element; + } + // This code is unreachable but some compilers may not realizes that. + return std::vector(); +} + +static std::string FormatWordList(const std::vector& words) { + Message word_list; + for (size_t i = 0; i < words.size(); ++i) { + if (i > 0 && words.size() > 2) { + word_list << ", "; + } + if (i == words.size() - 1) { + word_list << "and "; + } + word_list << "'" << words[i] << "'"; + } + return word_list.GetString(); +} + +static bool ValidateTestPropertyName( + const std::string& property_name, + const std::vector& reserved_names) { + if (std::find(reserved_names.begin(), reserved_names.end(), property_name) != + reserved_names.end()) { + ADD_FAILURE() << "Reserved key used in RecordProperty(): " << property_name + << " (" << FormatWordList(reserved_names) + << " are reserved by " << GTEST_NAME_ << ")"; + return false; + } + return true; +} + +// Adds a failure if the key is a reserved attribute of the element named +// xml_element. Returns true if the property is valid. +bool TestResult::ValidateTestProperty(const std::string& xml_element, + const TestProperty& test_property) { + return ValidateTestPropertyName(test_property.key(), + GetReservedAttributesForElement(xml_element)); +} + +// Clears the object. +void TestResult::Clear() { + test_part_results_.clear(); + test_properties_.clear(); + death_test_count_ = 0; + elapsed_time_ = 0; +} + +// Returns true off the test part was skipped. +static bool TestPartSkipped(const TestPartResult& result) { + return result.skipped(); +} + +// Returns true iff the test was skipped. +bool TestResult::Skipped() const { + return !Failed() && CountIf(test_part_results_, TestPartSkipped) > 0; +} + +// Returns true iff the test failed. +bool TestResult::Failed() const { + for (int i = 0; i < total_part_count(); ++i) { + if (GetTestPartResult(i).failed()) + return true; + } + return false; +} + +// Returns true iff the test part fatally failed. +static bool TestPartFatallyFailed(const TestPartResult& result) { + return result.fatally_failed(); +} + +// Returns true iff the test fatally failed. +bool TestResult::HasFatalFailure() const { + return CountIf(test_part_results_, TestPartFatallyFailed) > 0; +} + +// Returns true iff the test part non-fatally failed. +static bool TestPartNonfatallyFailed(const TestPartResult& result) { + return result.nonfatally_failed(); +} + +// Returns true iff the test has a non-fatal failure. +bool TestResult::HasNonfatalFailure() const { + return CountIf(test_part_results_, TestPartNonfatallyFailed) > 0; +} + +// Gets the number of all test parts. This is the sum of the number +// of successful test parts and the number of failed test parts. +int TestResult::total_part_count() const { + return static_cast(test_part_results_.size()); +} + +// Returns the number of the test properties. +int TestResult::test_property_count() const { + return static_cast(test_properties_.size()); +} + +// class Test + +// Creates a Test object. + +// The c'tor saves the states of all flags. +Test::Test() + : gtest_flag_saver_(new GTEST_FLAG_SAVER_) { +} + +// The d'tor restores the states of all flags. The actual work is +// done by the d'tor of the gtest_flag_saver_ field, and thus not +// visible here. +Test::~Test() { +} + +// Sets up the test fixture. +// +// A sub-class may override this. +void Test::SetUp() { +} + +// Tears down the test fixture. +// +// A sub-class may override this. +void Test::TearDown() { +} + +// Allows user supplied key value pairs to be recorded for later output. +void Test::RecordProperty(const std::string& key, const std::string& value) { + UnitTest::GetInstance()->RecordProperty(key, value); +} + +// Allows user supplied key value pairs to be recorded for later output. +void Test::RecordProperty(const std::string& key, int value) { + Message value_message; + value_message << value; + RecordProperty(key, value_message.GetString().c_str()); +} + +namespace internal { + +void ReportFailureInUnknownLocation(TestPartResult::Type result_type, + const std::string& message) { + // This function is a friend of UnitTest and as such has access to + // AddTestPartResult. + UnitTest::GetInstance()->AddTestPartResult( + result_type, + nullptr, // No info about the source file where the exception occurred. + -1, // We have no info on which line caused the exception. + message, + ""); // No stack trace, either. +} + +} // namespace internal + +// Google Test requires all tests in the same test suite to use the same test +// fixture class. This function checks if the current test has the +// same fixture class as the first test in the current test suite. If +// yes, it returns true; otherwise it generates a Google Test failure and +// returns false. +bool Test::HasSameFixtureClass() { + internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); + const TestSuite* const test_suite = impl->current_test_suite(); + + // Info about the first test in the current test suite. + const TestInfo* const first_test_info = test_suite->test_info_list()[0]; + const internal::TypeId first_fixture_id = first_test_info->fixture_class_id_; + const char* const first_test_name = first_test_info->name(); + + // Info about the current test. + const TestInfo* const this_test_info = impl->current_test_info(); + const internal::TypeId this_fixture_id = this_test_info->fixture_class_id_; + const char* const this_test_name = this_test_info->name(); + + if (this_fixture_id != first_fixture_id) { + // Is the first test defined using TEST? + const bool first_is_TEST = first_fixture_id == internal::GetTestTypeId(); + // Is this test defined using TEST? + const bool this_is_TEST = this_fixture_id == internal::GetTestTypeId(); + + if (first_is_TEST || this_is_TEST) { + // Both TEST and TEST_F appear in same test suite, which is incorrect. + // Tell the user how to fix this. + + // Gets the name of the TEST and the name of the TEST_F. Note + // that first_is_TEST and this_is_TEST cannot both be true, as + // the fixture IDs are different for the two tests. + const char* const TEST_name = + first_is_TEST ? first_test_name : this_test_name; + const char* const TEST_F_name = + first_is_TEST ? this_test_name : first_test_name; + + ADD_FAILURE() + << "All tests in the same test suite must use the same test fixture\n" + << "class, so mixing TEST_F and TEST in the same test suite is\n" + << "illegal. In test suite " << this_test_info->test_suite_name() + << ",\n" + << "test " << TEST_F_name << " is defined using TEST_F but\n" + << "test " << TEST_name << " is defined using TEST. You probably\n" + << "want to change the TEST to TEST_F or move it to another test\n" + << "case."; + } else { + // Two fixture classes with the same name appear in two different + // namespaces, which is not allowed. Tell the user how to fix this. + ADD_FAILURE() + << "All tests in the same test suite must use the same test fixture\n" + << "class. However, in test suite " + << this_test_info->test_suite_name() << ",\n" + << "you defined test " << first_test_name << " and test " + << this_test_name << "\n" + << "using two different test fixture classes. This can happen if\n" + << "the two classes are from different namespaces or translation\n" + << "units and have the same name. You should probably rename one\n" + << "of the classes to put the tests into different test suites."; + } + return false; + } + + return true; +} + +#if GTEST_HAS_SEH + +// Adds an "exception thrown" fatal failure to the current test. This +// function returns its result via an output parameter pointer because VC++ +// prohibits creation of objects with destructors on stack in functions +// using __try (see error C2712). +static std::string* FormatSehExceptionMessage(DWORD exception_code, + const char* location) { + Message message; + message << "SEH exception with code 0x" << std::setbase(16) << + exception_code << std::setbase(10) << " thrown in " << location << "."; + + return new std::string(message.GetString()); +} + +#endif // GTEST_HAS_SEH + +namespace internal { + +#if GTEST_HAS_EXCEPTIONS + +// Adds an "exception thrown" fatal failure to the current test. +static std::string FormatCxxExceptionMessage(const char* description, + const char* location) { + Message message; + if (description != nullptr) { + message << "C++ exception with description \"" << description << "\""; + } else { + message << "Unknown C++ exception"; + } + message << " thrown in " << location << "."; + + return message.GetString(); +} + +static std::string PrintTestPartResultToString( + const TestPartResult& test_part_result); + +GoogleTestFailureException::GoogleTestFailureException( + const TestPartResult& failure) + : ::std::runtime_error(PrintTestPartResultToString(failure).c_str()) {} + +#endif // GTEST_HAS_EXCEPTIONS + +// We put these helper functions in the internal namespace as IBM's xlC +// compiler rejects the code if they were declared static. + +// Runs the given method and handles SEH exceptions it throws, when +// SEH is supported; returns the 0-value for type Result in case of an +// SEH exception. (Microsoft compilers cannot handle SEH and C++ +// exceptions in the same function. Therefore, we provide a separate +// wrapper function for handling SEH exceptions.) +template +Result HandleSehExceptionsInMethodIfSupported( + T* object, Result (T::*method)(), const char* location) { +#if GTEST_HAS_SEH + __try { + return (object->*method)(); + } __except (internal::UnitTestOptions::GTestShouldProcessSEH( // NOLINT + GetExceptionCode())) { + // We create the exception message on the heap because VC++ prohibits + // creation of objects with destructors on stack in functions using __try + // (see error C2712). + std::string* exception_message = FormatSehExceptionMessage( + GetExceptionCode(), location); + internal::ReportFailureInUnknownLocation(TestPartResult::kFatalFailure, + *exception_message); + delete exception_message; + return static_cast(0); + } +#else + (void)location; + return (object->*method)(); +#endif // GTEST_HAS_SEH +} + +// Runs the given method and catches and reports C++ and/or SEH-style +// exceptions, if they are supported; returns the 0-value for type +// Result in case of an SEH exception. +template +Result HandleExceptionsInMethodIfSupported( + T* object, Result (T::*method)(), const char* location) { + // NOTE: The user code can affect the way in which Google Test handles + // exceptions by setting GTEST_FLAG(catch_exceptions), but only before + // RUN_ALL_TESTS() starts. It is technically possible to check the flag + // after the exception is caught and either report or re-throw the + // exception based on the flag's value: + // + // try { + // // Perform the test method. + // } catch (...) { + // if (GTEST_FLAG(catch_exceptions)) + // // Report the exception as failure. + // else + // throw; // Re-throws the original exception. + // } + // + // However, the purpose of this flag is to allow the program to drop into + // the debugger when the exception is thrown. On most platforms, once the + // control enters the catch block, the exception origin information is + // lost and the debugger will stop the program at the point of the + // re-throw in this function -- instead of at the point of the original + // throw statement in the code under test. For this reason, we perform + // the check early, sacrificing the ability to affect Google Test's + // exception handling in the method where the exception is thrown. + if (internal::GetUnitTestImpl()->catch_exceptions()) { +#if GTEST_HAS_EXCEPTIONS + try { + return HandleSehExceptionsInMethodIfSupported(object, method, location); + } catch (const AssertionException&) { // NOLINT + // This failure was reported already. + } catch (const internal::GoogleTestFailureException&) { // NOLINT + // This exception type can only be thrown by a failed Google + // Test assertion with the intention of letting another testing + // framework catch it. Therefore we just re-throw it. + throw; + } catch (const std::exception& e) { // NOLINT + internal::ReportFailureInUnknownLocation( + TestPartResult::kFatalFailure, + FormatCxxExceptionMessage(e.what(), location)); + } catch (...) { // NOLINT + internal::ReportFailureInUnknownLocation( + TestPartResult::kFatalFailure, + FormatCxxExceptionMessage(nullptr, location)); + } + return static_cast(0); +#else + return HandleSehExceptionsInMethodIfSupported(object, method, location); +#endif // GTEST_HAS_EXCEPTIONS + } else { + return (object->*method)(); + } +} + +} // namespace internal + +// Runs the test and updates the test result. +void Test::Run() { + if (!HasSameFixtureClass()) return; + + internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); + impl->os_stack_trace_getter()->UponLeavingGTest(); + internal::HandleExceptionsInMethodIfSupported(this, &Test::SetUp, "SetUp()"); + // We will run the test only if SetUp() was successful and didn't call + // GTEST_SKIP(). + if (!HasFatalFailure() && !IsSkipped()) { + impl->os_stack_trace_getter()->UponLeavingGTest(); + internal::HandleExceptionsInMethodIfSupported( + this, &Test::TestBody, "the test body"); + } + + // However, we want to clean up as much as possible. Hence we will + // always call TearDown(), even if SetUp() or the test body has + // failed. + impl->os_stack_trace_getter()->UponLeavingGTest(); + internal::HandleExceptionsInMethodIfSupported( + this, &Test::TearDown, "TearDown()"); +} + +// Returns true iff the current test has a fatal failure. +bool Test::HasFatalFailure() { + return internal::GetUnitTestImpl()->current_test_result()->HasFatalFailure(); +} + +// Returns true iff the current test has a non-fatal failure. +bool Test::HasNonfatalFailure() { + return internal::GetUnitTestImpl()->current_test_result()-> + HasNonfatalFailure(); +} + +// Returns true iff the current test was skipped. +bool Test::IsSkipped() { + return internal::GetUnitTestImpl()->current_test_result()->Skipped(); +} + +// class TestInfo + +// Constructs a TestInfo object. It assumes ownership of the test factory +// object. +TestInfo::TestInfo(const std::string& a_test_suite_name, + const std::string& a_name, const char* a_type_param, + const char* a_value_param, + internal::CodeLocation a_code_location, + internal::TypeId fixture_class_id, + internal::TestFactoryBase* factory) + : test_suite_name_(a_test_suite_name), + name_(a_name), + type_param_(a_type_param ? new std::string(a_type_param) : nullptr), + value_param_(a_value_param ? new std::string(a_value_param) : nullptr), + location_(a_code_location), + fixture_class_id_(fixture_class_id), + should_run_(false), + is_disabled_(false), + matches_filter_(false), + factory_(factory), + result_() {} + +// Destructs a TestInfo object. +TestInfo::~TestInfo() { delete factory_; } + +namespace internal { + +// Creates a new TestInfo object and registers it with Google Test; +// returns the created object. +// +// Arguments: +// +// test_suite_name: name of the test suite +// name: name of the test +// type_param: the name of the test's type parameter, or NULL if +// this is not a typed or a type-parameterized test. +// value_param: text representation of the test's value parameter, +// or NULL if this is not a value-parameterized test. +// code_location: code location where the test is defined +// fixture_class_id: ID of the test fixture class +// set_up_tc: pointer to the function that sets up the test suite +// tear_down_tc: pointer to the function that tears down the test suite +// factory: pointer to the factory that creates a test object. +// The newly created TestInfo instance will assume +// ownership of the factory object. +TestInfo* MakeAndRegisterTestInfo( + const char* test_suite_name, const char* name, const char* type_param, + const char* value_param, CodeLocation code_location, + TypeId fixture_class_id, SetUpTestSuiteFunc set_up_tc, + TearDownTestSuiteFunc tear_down_tc, TestFactoryBase* factory) { + TestInfo* const test_info = + new TestInfo(test_suite_name, name, type_param, value_param, + code_location, fixture_class_id, factory); + GetUnitTestImpl()->AddTestInfo(set_up_tc, tear_down_tc, test_info); + return test_info; +} + +void ReportInvalidTestSuiteType(const char* test_suite_name, + CodeLocation code_location) { + Message errors; + errors + << "Attempted redefinition of test suite " << test_suite_name << ".\n" + << "All tests in the same test suite must use the same test fixture\n" + << "class. However, in test suite " << test_suite_name << ", you tried\n" + << "to define a test using a fixture class different from the one\n" + << "used earlier. This can happen if the two fixture classes are\n" + << "from different namespaces and have the same name. You should\n" + << "probably rename one of the classes to put the tests into different\n" + << "test suites."; + + GTEST_LOG_(ERROR) << FormatFileLocation(code_location.file.c_str(), + code_location.line) + << " " << errors.GetString(); +} +} // namespace internal + +namespace { + +// A predicate that checks the test name of a TestInfo against a known +// value. +// +// This is used for implementation of the TestSuite class only. We put +// it in the anonymous namespace to prevent polluting the outer +// namespace. +// +// TestNameIs is copyable. +class TestNameIs { + public: + // Constructor. + // + // TestNameIs has NO default constructor. + explicit TestNameIs(const char* name) + : name_(name) {} + + // Returns true iff the test name of test_info matches name_. + bool operator()(const TestInfo * test_info) const { + return test_info && test_info->name() == name_; + } + + private: + std::string name_; +}; + +} // namespace + +namespace internal { + +// This method expands all parameterized tests registered with macros TEST_P +// and INSTANTIATE_TEST_SUITE_P into regular tests and registers those. +// This will be done just once during the program runtime. +void UnitTestImpl::RegisterParameterizedTests() { + if (!parameterized_tests_registered_) { + parameterized_test_registry_.RegisterTests(); + parameterized_tests_registered_ = true; + } +} + +} // namespace internal + +// Creates the test object, runs it, records its result, and then +// deletes it. +void TestInfo::Run() { + if (!should_run_) return; + + // Tells UnitTest where to store test result. + internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); + impl->set_current_test_info(this); + + TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater(); + + // Notifies the unit test event listeners that a test is about to start. + repeater->OnTestStart(*this); + + const TimeInMillis start = internal::GetTimeInMillis(); + + impl->os_stack_trace_getter()->UponLeavingGTest(); + + // Creates the test object. + Test* const test = internal::HandleExceptionsInMethodIfSupported( + factory_, &internal::TestFactoryBase::CreateTest, + "the test fixture's constructor"); + + // Runs the test if the constructor didn't generate a fatal failure or invoke + // GTEST_SKIP(). + // Note that the object will not be null + if (!Test::HasFatalFailure() && !Test::IsSkipped()) { + // This doesn't throw as all user code that can throw are wrapped into + // exception handling code. + test->Run(); + } + + if (test != nullptr) { + // Deletes the test object. + impl->os_stack_trace_getter()->UponLeavingGTest(); + internal::HandleExceptionsInMethodIfSupported( + test, &Test::DeleteSelf_, "the test fixture's destructor"); + } + + result_.set_elapsed_time(internal::GetTimeInMillis() - start); + + // Notifies the unit test event listener that a test has just finished. + repeater->OnTestEnd(*this); + + // Tells UnitTest to stop associating assertion results to this + // test. + impl->set_current_test_info(nullptr); +} + +// class TestSuite + +// Gets the number of successful tests in this test suite. +int TestSuite::successful_test_count() const { + return CountIf(test_info_list_, TestPassed); +} + +// Gets the number of successful tests in this test suite. +int TestSuite::skipped_test_count() const { + return CountIf(test_info_list_, TestSkipped); +} + +// Gets the number of failed tests in this test suite. +int TestSuite::failed_test_count() const { + return CountIf(test_info_list_, TestFailed); +} + +// Gets the number of disabled tests that will be reported in the XML report. +int TestSuite::reportable_disabled_test_count() const { + return CountIf(test_info_list_, TestReportableDisabled); +} + +// Gets the number of disabled tests in this test suite. +int TestSuite::disabled_test_count() const { + return CountIf(test_info_list_, TestDisabled); +} + +// Gets the number of tests to be printed in the XML report. +int TestSuite::reportable_test_count() const { + return CountIf(test_info_list_, TestReportable); +} + +// Get the number of tests in this test suite that should run. +int TestSuite::test_to_run_count() const { + return CountIf(test_info_list_, ShouldRunTest); +} + +// Gets the number of all tests. +int TestSuite::total_test_count() const { + return static_cast(test_info_list_.size()); +} + +// Creates a TestSuite with the given name. +// +// Arguments: +// +// name: name of the test suite +// a_type_param: the name of the test suite's type parameter, or NULL if +// this is not a typed or a type-parameterized test suite. +// set_up_tc: pointer to the function that sets up the test suite +// tear_down_tc: pointer to the function that tears down the test suite +TestSuite::TestSuite(const char* a_name, const char* a_type_param, + internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc) + : name_(a_name), + type_param_(a_type_param ? new std::string(a_type_param) : nullptr), + set_up_tc_(set_up_tc), + tear_down_tc_(tear_down_tc), + should_run_(false), + elapsed_time_(0) {} + +// Destructor of TestSuite. +TestSuite::~TestSuite() { + // Deletes every Test in the collection. + ForEach(test_info_list_, internal::Delete); +} + +// Returns the i-th test among all the tests. i can range from 0 to +// total_test_count() - 1. If i is not in that range, returns NULL. +const TestInfo* TestSuite::GetTestInfo(int i) const { + const int index = GetElementOr(test_indices_, i, -1); + return index < 0 ? nullptr : test_info_list_[static_cast(index)]; +} + +// Returns the i-th test among all the tests. i can range from 0 to +// total_test_count() - 1. If i is not in that range, returns NULL. +TestInfo* TestSuite::GetMutableTestInfo(int i) { + const int index = GetElementOr(test_indices_, i, -1); + return index < 0 ? nullptr : test_info_list_[static_cast(index)]; +} + +// Adds a test to this test suite. Will delete the test upon +// destruction of the TestSuite object. +void TestSuite::AddTestInfo(TestInfo* test_info) { + test_info_list_.push_back(test_info); + test_indices_.push_back(static_cast(test_indices_.size())); +} + +// Runs every test in this TestSuite. +void TestSuite::Run() { + if (!should_run_) return; + + internal::UnitTestImpl* const impl = internal::GetUnitTestImpl(); + impl->set_current_test_suite(this); + + TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater(); + + // Call both legacy and the new API + repeater->OnTestSuiteStart(*this); +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI + repeater->OnTestCaseStart(*this); +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI + + impl->os_stack_trace_getter()->UponLeavingGTest(); + internal::HandleExceptionsInMethodIfSupported( + this, &TestSuite::RunSetUpTestSuite, "SetUpTestSuite()"); + + const internal::TimeInMillis start = internal::GetTimeInMillis(); + for (int i = 0; i < total_test_count(); i++) { + GetMutableTestInfo(i)->Run(); + } + elapsed_time_ = internal::GetTimeInMillis() - start; + + impl->os_stack_trace_getter()->UponLeavingGTest(); + internal::HandleExceptionsInMethodIfSupported( + this, &TestSuite::RunTearDownTestSuite, "TearDownTestSuite()"); + + // Call both legacy and the new API + repeater->OnTestSuiteEnd(*this); +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI + repeater->OnTestCaseEnd(*this); +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI + + impl->set_current_test_suite(nullptr); +} + +// Clears the results of all tests in this test suite. +void TestSuite::ClearResult() { + ad_hoc_test_result_.Clear(); + ForEach(test_info_list_, TestInfo::ClearTestResult); +} + +// Shuffles the tests in this test suite. +void TestSuite::ShuffleTests(internal::Random* random) { + Shuffle(random, &test_indices_); +} + +// Restores the test order to before the first shuffle. +void TestSuite::UnshuffleTests() { + for (size_t i = 0; i < test_indices_.size(); i++) { + test_indices_[i] = static_cast(i); + } +} + +// Formats a countable noun. Depending on its quantity, either the +// singular form or the plural form is used. e.g. +// +// FormatCountableNoun(1, "formula", "formuli") returns "1 formula". +// FormatCountableNoun(5, "book", "books") returns "5 books". +static std::string FormatCountableNoun(int count, + const char * singular_form, + const char * plural_form) { + return internal::StreamableToString(count) + " " + + (count == 1 ? singular_form : plural_form); +} + +// Formats the count of tests. +static std::string FormatTestCount(int test_count) { + return FormatCountableNoun(test_count, "test", "tests"); +} + +// Formats the count of test suites. +static std::string FormatTestSuiteCount(int test_suite_count) { + return FormatCountableNoun(test_suite_count, "test suite", "test suites"); +} + +// Converts a TestPartResult::Type enum to human-friendly string +// representation. Both kNonFatalFailure and kFatalFailure are translated +// to "Failure", as the user usually doesn't care about the difference +// between the two when viewing the test result. +static const char * TestPartResultTypeToString(TestPartResult::Type type) { + switch (type) { + case TestPartResult::kSkip: + return "Skipped"; + case TestPartResult::kSuccess: + return "Success"; + + case TestPartResult::kNonFatalFailure: + case TestPartResult::kFatalFailure: +#ifdef _MSC_VER + return "error: "; +#else + return "Failure\n"; +#endif + default: + return "Unknown result type"; + } +} + +namespace internal { + +// Prints a TestPartResult to an std::string. +static std::string PrintTestPartResultToString( + const TestPartResult& test_part_result) { + return (Message() + << internal::FormatFileLocation(test_part_result.file_name(), + test_part_result.line_number()) + << " " << TestPartResultTypeToString(test_part_result.type()) + << test_part_result.message()).GetString(); +} + +// Prints a TestPartResult. +static void PrintTestPartResult(const TestPartResult& test_part_result) { + const std::string& result = + PrintTestPartResultToString(test_part_result); + printf("%s\n", result.c_str()); + fflush(stdout); + // If the test program runs in Visual Studio or a debugger, the + // following statements add the test part result message to the Output + // window such that the user can double-click on it to jump to the + // corresponding source code location; otherwise they do nothing. +#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE + // We don't call OutputDebugString*() on Windows Mobile, as printing + // to stdout is done by OutputDebugString() there already - we don't + // want the same message printed twice. + ::OutputDebugStringA(result.c_str()); + ::OutputDebugStringA("\n"); +#endif +} + +// class PrettyUnitTestResultPrinter +#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \ + !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT && !GTEST_OS_WINDOWS_MINGW + +// Returns the character attribute for the given color. +static WORD GetColorAttribute(GTestColor color) { + switch (color) { + case COLOR_RED: return FOREGROUND_RED; + case COLOR_GREEN: return FOREGROUND_GREEN; + case COLOR_YELLOW: return FOREGROUND_RED | FOREGROUND_GREEN; + default: return 0; + } +} + +static int GetBitOffset(WORD color_mask) { + if (color_mask == 0) return 0; + + int bitOffset = 0; + while ((color_mask & 1) == 0) { + color_mask >>= 1; + ++bitOffset; + } + return bitOffset; +} + +static WORD GetNewColor(GTestColor color, WORD old_color_attrs) { + // Let's reuse the BG + static const WORD background_mask = BACKGROUND_BLUE | BACKGROUND_GREEN | + BACKGROUND_RED | BACKGROUND_INTENSITY; + static const WORD foreground_mask = FOREGROUND_BLUE | FOREGROUND_GREEN | + FOREGROUND_RED | FOREGROUND_INTENSITY; + const WORD existing_bg = old_color_attrs & background_mask; + + WORD new_color = + GetColorAttribute(color) | existing_bg | FOREGROUND_INTENSITY; + static const int bg_bitOffset = GetBitOffset(background_mask); + static const int fg_bitOffset = GetBitOffset(foreground_mask); + + if (((new_color & background_mask) >> bg_bitOffset) == + ((new_color & foreground_mask) >> fg_bitOffset)) { + new_color ^= FOREGROUND_INTENSITY; // invert intensity + } + return new_color; +} + +#else + +// Returns the ANSI color code for the given color. COLOR_DEFAULT is +// an invalid input. +static const char* GetAnsiColorCode(GTestColor color) { + switch (color) { + case COLOR_RED: return "1"; + case COLOR_GREEN: return "2"; + case COLOR_YELLOW: return "3"; + default: + return nullptr; + } +} + +#endif // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE + +// Returns true iff Google Test should use colors in the output. +bool ShouldUseColor(bool stdout_is_tty) { + const char* const gtest_color = GTEST_FLAG(color).c_str(); + + if (String::CaseInsensitiveCStringEquals(gtest_color, "auto")) { +#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MINGW + // On Windows the TERM variable is usually not set, but the + // console there does support colors. + return stdout_is_tty; +#else + // On non-Windows platforms, we rely on the TERM variable. + const char* const term = posix::GetEnv("TERM"); + const bool term_supports_color = + String::CStringEquals(term, "xterm") || + String::CStringEquals(term, "xterm-color") || + String::CStringEquals(term, "xterm-256color") || + String::CStringEquals(term, "screen") || + String::CStringEquals(term, "screen-256color") || + String::CStringEquals(term, "tmux") || + String::CStringEquals(term, "tmux-256color") || + String::CStringEquals(term, "rxvt-unicode") || + String::CStringEquals(term, "rxvt-unicode-256color") || + String::CStringEquals(term, "linux") || + String::CStringEquals(term, "cygwin"); + return stdout_is_tty && term_supports_color; +#endif // GTEST_OS_WINDOWS + } + + return String::CaseInsensitiveCStringEquals(gtest_color, "yes") || + String::CaseInsensitiveCStringEquals(gtest_color, "true") || + String::CaseInsensitiveCStringEquals(gtest_color, "t") || + String::CStringEquals(gtest_color, "1"); + // We take "yes", "true", "t", and "1" as meaning "yes". If the + // value is neither one of these nor "auto", we treat it as "no" to + // be conservative. +} + +// Helpers for printing colored strings to stdout. Note that on Windows, we +// cannot simply emit special characters and have the terminal change colors. +// This routine must actually emit the characters rather than return a string +// that would be colored when printed, as can be done on Linux. +void ColoredPrintf(GTestColor color, const char* fmt, ...) { + va_list args; + va_start(args, fmt); + +#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_ZOS || GTEST_OS_IOS || \ + GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT + const bool use_color = AlwaysFalse(); +#else + static const bool in_color_mode = + ShouldUseColor(posix::IsATTY(posix::FileNo(stdout)) != 0); + const bool use_color = in_color_mode && (color != COLOR_DEFAULT); +#endif // GTEST_OS_WINDOWS_MOBILE || GTEST_OS_ZOS + + if (!use_color) { + vprintf(fmt, args); + va_end(args); + return; + } + +#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \ + !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT && !GTEST_OS_WINDOWS_MINGW + const HANDLE stdout_handle = GetStdHandle(STD_OUTPUT_HANDLE); + + // Gets the current text color. + CONSOLE_SCREEN_BUFFER_INFO buffer_info; + GetConsoleScreenBufferInfo(stdout_handle, &buffer_info); + const WORD old_color_attrs = buffer_info.wAttributes; + const WORD new_color = GetNewColor(color, old_color_attrs); + + // We need to flush the stream buffers into the console before each + // SetConsoleTextAttribute call lest it affect the text that is already + // printed but has not yet reached the console. + fflush(stdout); + SetConsoleTextAttribute(stdout_handle, new_color); + + vprintf(fmt, args); + + fflush(stdout); + // Restores the text color. + SetConsoleTextAttribute(stdout_handle, old_color_attrs); +#else + printf("\033[0;3%sm", GetAnsiColorCode(color)); + vprintf(fmt, args); + printf("\033[m"); // Resets the terminal to default. +#endif // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE + va_end(args); +} + +// Text printed in Google Test's text output and --gtest_list_tests +// output to label the type parameter and value parameter for a test. +static const char kTypeParamLabel[] = "TypeParam"; +static const char kValueParamLabel[] = "GetParam()"; + +static void PrintFullTestCommentIfPresent(const TestInfo& test_info) { + const char* const type_param = test_info.type_param(); + const char* const value_param = test_info.value_param(); + + if (type_param != nullptr || value_param != nullptr) { + printf(", where "); + if (type_param != nullptr) { + printf("%s = %s", kTypeParamLabel, type_param); + if (value_param != nullptr) printf(" and "); + } + if (value_param != nullptr) { + printf("%s = %s", kValueParamLabel, value_param); + } + } +} + +// This class implements the TestEventListener interface. +// +// Class PrettyUnitTestResultPrinter is copyable. +class PrettyUnitTestResultPrinter : public TestEventListener { + public: + PrettyUnitTestResultPrinter() {} + static void PrintTestName(const char* test_suite, const char* test) { + printf("%s.%s", test_suite, test); + } + + // The following methods override what's in the TestEventListener class. + void OnTestProgramStart(const UnitTest& /*unit_test*/) override {} + void OnTestIterationStart(const UnitTest& unit_test, int iteration) override; + void OnEnvironmentsSetUpStart(const UnitTest& unit_test) override; + void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) override {} + void OnTestCaseStart(const TestSuite& test_suite) override; + void OnTestStart(const TestInfo& test_info) override; + void OnTestPartResult(const TestPartResult& result) override; + void OnTestEnd(const TestInfo& test_info) override; + void OnTestCaseEnd(const TestSuite& test_suite) override; + void OnEnvironmentsTearDownStart(const UnitTest& unit_test) override; + void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) override {} + void OnTestIterationEnd(const UnitTest& unit_test, int iteration) override; + void OnTestProgramEnd(const UnitTest& /*unit_test*/) override {} + + private: + static void PrintFailedTests(const UnitTest& unit_test); + static void PrintSkippedTests(const UnitTest& unit_test); +}; + + // Fired before each iteration of tests starts. +void PrettyUnitTestResultPrinter::OnTestIterationStart( + const UnitTest& unit_test, int iteration) { + if (GTEST_FLAG(repeat) != 1) + printf("\nRepeating all tests (iteration %d) . . .\n\n", iteration + 1); + + const char* const filter = GTEST_FLAG(filter).c_str(); + + // Prints the filter if it's not *. This reminds the user that some + // tests may be skipped. + if (!String::CStringEquals(filter, kUniversalFilter)) { + ColoredPrintf(COLOR_YELLOW, + "Note: %s filter = %s\n", GTEST_NAME_, filter); + } + + if (internal::ShouldShard(kTestTotalShards, kTestShardIndex, false)) { + const Int32 shard_index = Int32FromEnvOrDie(kTestShardIndex, -1); + ColoredPrintf(COLOR_YELLOW, + "Note: This is test shard %d of %s.\n", + static_cast(shard_index) + 1, + internal::posix::GetEnv(kTestTotalShards)); + } + + if (GTEST_FLAG(shuffle)) { + ColoredPrintf(COLOR_YELLOW, + "Note: Randomizing tests' orders with a seed of %d .\n", + unit_test.random_seed()); + } + + ColoredPrintf(COLOR_GREEN, "[==========] "); + printf("Running %s from %s.\n", + FormatTestCount(unit_test.test_to_run_count()).c_str(), + FormatTestSuiteCount(unit_test.test_suite_to_run_count()).c_str()); + fflush(stdout); +} + +void PrettyUnitTestResultPrinter::OnEnvironmentsSetUpStart( + const UnitTest& /*unit_test*/) { + ColoredPrintf(COLOR_GREEN, "[----------] "); + printf("Global test environment set-up.\n"); + fflush(stdout); +} + +void PrettyUnitTestResultPrinter::OnTestCaseStart(const TestSuite& test_suite) { + const std::string counts = + FormatCountableNoun(test_suite.test_to_run_count(), "test", "tests"); + ColoredPrintf(COLOR_GREEN, "[----------] "); + printf("%s from %s", counts.c_str(), test_suite.name()); + if (test_suite.type_param() == nullptr) { + printf("\n"); + } else { + printf(", where %s = %s\n", kTypeParamLabel, test_suite.type_param()); + } + fflush(stdout); +} + +void PrettyUnitTestResultPrinter::OnTestStart(const TestInfo& test_info) { + ColoredPrintf(COLOR_GREEN, "[ RUN ] "); + PrintTestName(test_info.test_suite_name(), test_info.name()); + printf("\n"); + fflush(stdout); +} + +// Called after an assertion failure. +void PrettyUnitTestResultPrinter::OnTestPartResult( + const TestPartResult& result) { + switch (result.type()) { + // If the test part succeeded, or was skipped, + // we don't need to do anything. + case TestPartResult::kSkip: + case TestPartResult::kSuccess: + return; + default: + // Print failure message from the assertion + // (e.g. expected this and got that). + PrintTestPartResult(result); + fflush(stdout); + } +} + +void PrettyUnitTestResultPrinter::OnTestEnd(const TestInfo& test_info) { + if (test_info.result()->Passed()) { + ColoredPrintf(COLOR_GREEN, "[ OK ] "); + } else if (test_info.result()->Skipped()) { + ColoredPrintf(COLOR_GREEN, "[ SKIPPED ] "); + } else { + ColoredPrintf(COLOR_RED, "[ FAILED ] "); + } + PrintTestName(test_info.test_suite_name(), test_info.name()); + if (test_info.result()->Failed()) + PrintFullTestCommentIfPresent(test_info); + + if (GTEST_FLAG(print_time)) { + printf(" (%s ms)\n", internal::StreamableToString( + test_info.result()->elapsed_time()).c_str()); + } else { + printf("\n"); + } + fflush(stdout); +} + +void PrettyUnitTestResultPrinter::OnTestCaseEnd(const TestSuite& test_suite) { + if (!GTEST_FLAG(print_time)) return; + + const std::string counts = + FormatCountableNoun(test_suite.test_to_run_count(), "test", "tests"); + ColoredPrintf(COLOR_GREEN, "[----------] "); + printf("%s from %s (%s ms total)\n\n", counts.c_str(), test_suite.name(), + internal::StreamableToString(test_suite.elapsed_time()).c_str()); + fflush(stdout); +} + +void PrettyUnitTestResultPrinter::OnEnvironmentsTearDownStart( + const UnitTest& /*unit_test*/) { + ColoredPrintf(COLOR_GREEN, "[----------] "); + printf("Global test environment tear-down\n"); + fflush(stdout); +} + +// Internal helper for printing the list of failed tests. +void PrettyUnitTestResultPrinter::PrintFailedTests(const UnitTest& unit_test) { + const int failed_test_count = unit_test.failed_test_count(); + if (failed_test_count == 0) { + return; + } + + for (int i = 0; i < unit_test.total_test_suite_count(); ++i) { + const TestSuite& test_suite = *unit_test.GetTestSuite(i); + if (!test_suite.should_run() || (test_suite.failed_test_count() == 0)) { + continue; + } + for (int j = 0; j < test_suite.total_test_count(); ++j) { + const TestInfo& test_info = *test_suite.GetTestInfo(j); + if (!test_info.should_run() || !test_info.result()->Failed()) { + continue; + } + ColoredPrintf(COLOR_RED, "[ FAILED ] "); + printf("%s.%s", test_suite.name(), test_info.name()); + PrintFullTestCommentIfPresent(test_info); + printf("\n"); + } + } +} + +// Internal helper for printing the list of skipped tests. +void PrettyUnitTestResultPrinter::PrintSkippedTests(const UnitTest& unit_test) { + const int skipped_test_count = unit_test.skipped_test_count(); + if (skipped_test_count == 0) { + return; + } + + for (int i = 0; i < unit_test.total_test_suite_count(); ++i) { + const TestSuite& test_suite = *unit_test.GetTestSuite(i); + if (!test_suite.should_run() || (test_suite.skipped_test_count() == 0)) { + continue; + } + for (int j = 0; j < test_suite.total_test_count(); ++j) { + const TestInfo& test_info = *test_suite.GetTestInfo(j); + if (!test_info.should_run() || !test_info.result()->Skipped()) { + continue; + } + ColoredPrintf(COLOR_GREEN, "[ SKIPPED ] "); + printf("%s.%s", test_suite.name(), test_info.name()); + printf("\n"); + } + } +} + +void PrettyUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test, + int /*iteration*/) { + ColoredPrintf(COLOR_GREEN, "[==========] "); + printf("%s from %s ran.", + FormatTestCount(unit_test.test_to_run_count()).c_str(), + FormatTestSuiteCount(unit_test.test_suite_to_run_count()).c_str()); + if (GTEST_FLAG(print_time)) { + printf(" (%s ms total)", + internal::StreamableToString(unit_test.elapsed_time()).c_str()); + } + printf("\n"); + ColoredPrintf(COLOR_GREEN, "[ PASSED ] "); + printf("%s.\n", FormatTestCount(unit_test.successful_test_count()).c_str()); + + const int skipped_test_count = unit_test.skipped_test_count(); + if (skipped_test_count > 0) { + ColoredPrintf(COLOR_GREEN, "[ SKIPPED ] "); + printf("%s, listed below:\n", FormatTestCount(skipped_test_count).c_str()); + PrintSkippedTests(unit_test); + } + + int num_failures = unit_test.failed_test_count(); + if (!unit_test.Passed()) { + const int failed_test_count = unit_test.failed_test_count(); + ColoredPrintf(COLOR_RED, "[ FAILED ] "); + printf("%s, listed below:\n", FormatTestCount(failed_test_count).c_str()); + PrintFailedTests(unit_test); + printf("\n%2d FAILED %s\n", num_failures, + num_failures == 1 ? "TEST" : "TESTS"); + } + + int num_disabled = unit_test.reportable_disabled_test_count(); + if (num_disabled && !GTEST_FLAG(also_run_disabled_tests)) { + if (!num_failures) { + printf("\n"); // Add a spacer if no FAILURE banner is displayed. + } + ColoredPrintf(COLOR_YELLOW, + " YOU HAVE %d DISABLED %s\n\n", + num_disabled, + num_disabled == 1 ? "TEST" : "TESTS"); + } + // Ensure that Google Test output is printed before, e.g., heapchecker output. + fflush(stdout); +} + +// End PrettyUnitTestResultPrinter + +// class TestEventRepeater +// +// This class forwards events to other event listeners. +class TestEventRepeater : public TestEventListener { + public: + TestEventRepeater() : forwarding_enabled_(true) {} + ~TestEventRepeater() override; + void Append(TestEventListener *listener); + TestEventListener* Release(TestEventListener* listener); + + // Controls whether events will be forwarded to listeners_. Set to false + // in death test child processes. + bool forwarding_enabled() const { return forwarding_enabled_; } + void set_forwarding_enabled(bool enable) { forwarding_enabled_ = enable; } + + void OnTestProgramStart(const UnitTest& unit_test) override; + void OnTestIterationStart(const UnitTest& unit_test, int iteration) override; + void OnEnvironmentsSetUpStart(const UnitTest& unit_test) override; + void OnEnvironmentsSetUpEnd(const UnitTest& unit_test) override; +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI + void OnTestCaseStart(const TestSuite& parameter) override; +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI + void OnTestSuiteStart(const TestSuite& parameter) override; + void OnTestStart(const TestInfo& test_info) override; + void OnTestPartResult(const TestPartResult& result) override; + void OnTestEnd(const TestInfo& test_info) override; +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI + void OnTestCaseEnd(const TestSuite& parameter) override; +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI + void OnTestSuiteEnd(const TestSuite& parameter) override; + void OnEnvironmentsTearDownStart(const UnitTest& unit_test) override; + void OnEnvironmentsTearDownEnd(const UnitTest& unit_test) override; + void OnTestIterationEnd(const UnitTest& unit_test, int iteration) override; + void OnTestProgramEnd(const UnitTest& unit_test) override; + + private: + // Controls whether events will be forwarded to listeners_. Set to false + // in death test child processes. + bool forwarding_enabled_; + // The list of listeners that receive events. + std::vector listeners_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventRepeater); +}; + +TestEventRepeater::~TestEventRepeater() { + ForEach(listeners_, Delete); +} + +void TestEventRepeater::Append(TestEventListener *listener) { + listeners_.push_back(listener); +} + +TestEventListener* TestEventRepeater::Release(TestEventListener *listener) { + for (size_t i = 0; i < listeners_.size(); ++i) { + if (listeners_[i] == listener) { + listeners_.erase(listeners_.begin() + static_cast(i)); + return listener; + } + } + + return nullptr; +} + +// Since most methods are very similar, use macros to reduce boilerplate. +// This defines a member that forwards the call to all listeners. +#define GTEST_REPEATER_METHOD_(Name, Type) \ +void TestEventRepeater::Name(const Type& parameter) { \ + if (forwarding_enabled_) { \ + for (size_t i = 0; i < listeners_.size(); i++) { \ + listeners_[i]->Name(parameter); \ + } \ + } \ +} +// This defines a member that forwards the call to all listeners in reverse +// order. +#define GTEST_REVERSE_REPEATER_METHOD_(Name, Type) \ + void TestEventRepeater::Name(const Type& parameter) { \ + if (forwarding_enabled_) { \ + for (size_t i = listeners_.size(); i != 0; i--) { \ + listeners_[i - 1]->Name(parameter); \ + } \ + } \ + } + +GTEST_REPEATER_METHOD_(OnTestProgramStart, UnitTest) +GTEST_REPEATER_METHOD_(OnEnvironmentsSetUpStart, UnitTest) +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +GTEST_REPEATER_METHOD_(OnTestCaseStart, TestSuite) +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +GTEST_REPEATER_METHOD_(OnTestSuiteStart, TestSuite) +GTEST_REPEATER_METHOD_(OnTestStart, TestInfo) +GTEST_REPEATER_METHOD_(OnTestPartResult, TestPartResult) +GTEST_REPEATER_METHOD_(OnEnvironmentsTearDownStart, UnitTest) +GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsSetUpEnd, UnitTest) +GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsTearDownEnd, UnitTest) +GTEST_REVERSE_REPEATER_METHOD_(OnTestEnd, TestInfo) +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +GTEST_REVERSE_REPEATER_METHOD_(OnTestCaseEnd, TestSuite) +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +GTEST_REVERSE_REPEATER_METHOD_(OnTestSuiteEnd, TestSuite) +GTEST_REVERSE_REPEATER_METHOD_(OnTestProgramEnd, UnitTest) + +#undef GTEST_REPEATER_METHOD_ +#undef GTEST_REVERSE_REPEATER_METHOD_ + +void TestEventRepeater::OnTestIterationStart(const UnitTest& unit_test, + int iteration) { + if (forwarding_enabled_) { + for (size_t i = 0; i < listeners_.size(); i++) { + listeners_[i]->OnTestIterationStart(unit_test, iteration); + } + } +} + +void TestEventRepeater::OnTestIterationEnd(const UnitTest& unit_test, + int iteration) { + if (forwarding_enabled_) { + for (size_t i = listeners_.size(); i > 0; i--) { + listeners_[i - 1]->OnTestIterationEnd(unit_test, iteration); + } + } +} + +// End TestEventRepeater + +// This class generates an XML output file. +class XmlUnitTestResultPrinter : public EmptyTestEventListener { + public: + explicit XmlUnitTestResultPrinter(const char* output_file); + + void OnTestIterationEnd(const UnitTest& unit_test, int iteration) override; + void ListTestsMatchingFilter(const std::vector& test_suites); + + // Prints an XML summary of all unit tests. + static void PrintXmlTestsList(std::ostream* stream, + const std::vector& test_suites); + + private: + // Is c a whitespace character that is normalized to a space character + // when it appears in an XML attribute value? + static bool IsNormalizableWhitespace(char c) { + return c == 0x9 || c == 0xA || c == 0xD; + } + + // May c appear in a well-formed XML document? + static bool IsValidXmlCharacter(char c) { + return IsNormalizableWhitespace(c) || c >= 0x20; + } + + // Returns an XML-escaped copy of the input string str. If + // is_attribute is true, the text is meant to appear as an attribute + // value, and normalizable whitespace is preserved by replacing it + // with character references. + static std::string EscapeXml(const std::string& str, bool is_attribute); + + // Returns the given string with all characters invalid in XML removed. + static std::string RemoveInvalidXmlCharacters(const std::string& str); + + // Convenience wrapper around EscapeXml when str is an attribute value. + static std::string EscapeXmlAttribute(const std::string& str) { + return EscapeXml(str, true); + } + + // Convenience wrapper around EscapeXml when str is not an attribute value. + static std::string EscapeXmlText(const char* str) { + return EscapeXml(str, false); + } + + // Verifies that the given attribute belongs to the given element and + // streams the attribute as XML. + static void OutputXmlAttribute(std::ostream* stream, + const std::string& element_name, + const std::string& name, + const std::string& value); + + // Streams an XML CDATA section, escaping invalid CDATA sequences as needed. + static void OutputXmlCDataSection(::std::ostream* stream, const char* data); + + // Streams an XML representation of a TestInfo object. + static void OutputXmlTestInfo(::std::ostream* stream, + const char* test_suite_name, + const TestInfo& test_info); + + // Prints an XML representation of a TestSuite object + static void PrintXmlTestSuite(::std::ostream* stream, + const TestSuite& test_suite); + + // Prints an XML summary of unit_test to output stream out. + static void PrintXmlUnitTest(::std::ostream* stream, + const UnitTest& unit_test); + + // Produces a string representing the test properties in a result as space + // delimited XML attributes based on the property key="value" pairs. + // When the std::string is not empty, it includes a space at the beginning, + // to delimit this attribute from prior attributes. + static std::string TestPropertiesAsXmlAttributes(const TestResult& result); + + // Streams an XML representation of the test properties of a TestResult + // object. + static void OutputXmlTestProperties(std::ostream* stream, + const TestResult& result); + + // The output file. + const std::string output_file_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(XmlUnitTestResultPrinter); +}; + +// Creates a new XmlUnitTestResultPrinter. +XmlUnitTestResultPrinter::XmlUnitTestResultPrinter(const char* output_file) + : output_file_(output_file) { + if (output_file_.empty()) { + GTEST_LOG_(FATAL) << "XML output file may not be null"; + } +} + +// Called after the unit test ends. +void XmlUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test, + int /*iteration*/) { + FILE* xmlout = OpenFileForWriting(output_file_); + std::stringstream stream; + PrintXmlUnitTest(&stream, unit_test); + fprintf(xmlout, "%s", StringStreamToString(&stream).c_str()); + fclose(xmlout); +} + +void XmlUnitTestResultPrinter::ListTestsMatchingFilter( + const std::vector& test_suites) { + FILE* xmlout = OpenFileForWriting(output_file_); + std::stringstream stream; + PrintXmlTestsList(&stream, test_suites); + fprintf(xmlout, "%s", StringStreamToString(&stream).c_str()); + fclose(xmlout); +} + +// Returns an XML-escaped copy of the input string str. If is_attribute +// is true, the text is meant to appear as an attribute value, and +// normalizable whitespace is preserved by replacing it with character +// references. +// +// Invalid XML characters in str, if any, are stripped from the output. +// It is expected that most, if not all, of the text processed by this +// module will consist of ordinary English text. +// If this module is ever modified to produce version 1.1 XML output, +// most invalid characters can be retained using character references. +std::string XmlUnitTestResultPrinter::EscapeXml( + const std::string& str, bool is_attribute) { + Message m; + + for (size_t i = 0; i < str.size(); ++i) { + const char ch = str[i]; + switch (ch) { + case '<': + m << "<"; + break; + case '>': + m << ">"; + break; + case '&': + m << "&"; + break; + case '\'': + if (is_attribute) + m << "'"; + else + m << '\''; + break; + case '"': + if (is_attribute) + m << """; + else + m << '"'; + break; + default: + if (IsValidXmlCharacter(ch)) { + if (is_attribute && IsNormalizableWhitespace(ch)) + m << "&#x" << String::FormatByte(static_cast(ch)) + << ";"; + else + m << ch; + } + break; + } + } + + return m.GetString(); +} + +// Returns the given string with all characters invalid in XML removed. +// Currently invalid characters are dropped from the string. An +// alternative is to replace them with certain characters such as . or ?. +std::string XmlUnitTestResultPrinter::RemoveInvalidXmlCharacters( + const std::string& str) { + std::string output; + output.reserve(str.size()); + for (std::string::const_iterator it = str.begin(); it != str.end(); ++it) + if (IsValidXmlCharacter(*it)) + output.push_back(*it); + + return output; +} + +// The following routines generate an XML representation of a UnitTest +// object. +// GOOGLETEST_CM0009 DO NOT DELETE +// +// This is how Google Test concepts map to the DTD: +// +// <-- corresponds to a UnitTest object +// <-- corresponds to a TestSuite object +// <-- corresponds to a TestInfo object +// ... +// ... +// ... +// <-- individual assertion failures +// +// +// + +// Formats the given time in milliseconds as seconds. +std::string FormatTimeInMillisAsSeconds(TimeInMillis ms) { + ::std::stringstream ss; + ss << (static_cast(ms) * 1e-3); + return ss.str(); +} + +static bool PortableLocaltime(time_t seconds, struct tm* out) { +#if defined(_MSC_VER) + return localtime_s(out, &seconds) == 0; +#elif defined(__MINGW32__) || defined(__MINGW64__) + // MINGW provides neither localtime_r nor localtime_s, but uses + // Windows' localtime(), which has a thread-local tm buffer. + struct tm* tm_ptr = localtime(&seconds); // NOLINT + if (tm_ptr == nullptr) return false; + *out = *tm_ptr; + return true; +#else + return localtime_r(&seconds, out) != nullptr; +#endif +} + +// Converts the given epoch time in milliseconds to a date string in the ISO +// 8601 format, without the timezone information. +std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms) { + struct tm time_struct; + if (!PortableLocaltime(static_cast(ms / 1000), &time_struct)) + return ""; + // YYYY-MM-DDThh:mm:ss + return StreamableToString(time_struct.tm_year + 1900) + "-" + + String::FormatIntWidth2(time_struct.tm_mon + 1) + "-" + + String::FormatIntWidth2(time_struct.tm_mday) + "T" + + String::FormatIntWidth2(time_struct.tm_hour) + ":" + + String::FormatIntWidth2(time_struct.tm_min) + ":" + + String::FormatIntWidth2(time_struct.tm_sec); +} + +// Streams an XML CDATA section, escaping invalid CDATA sequences as needed. +void XmlUnitTestResultPrinter::OutputXmlCDataSection(::std::ostream* stream, + const char* data) { + const char* segment = data; + *stream << ""); + if (next_segment != nullptr) { + stream->write( + segment, static_cast(next_segment - segment)); + *stream << "]]>]]>"); + } else { + *stream << segment; + break; + } + } + *stream << "]]>"; +} + +void XmlUnitTestResultPrinter::OutputXmlAttribute( + std::ostream* stream, + const std::string& element_name, + const std::string& name, + const std::string& value) { + const std::vector& allowed_names = + GetReservedOutputAttributesForElement(element_name); + + GTEST_CHECK_(std::find(allowed_names.begin(), allowed_names.end(), name) != + allowed_names.end()) + << "Attribute " << name << " is not allowed for element <" << element_name + << ">."; + + *stream << " " << name << "=\"" << EscapeXmlAttribute(value) << "\""; +} + +// Prints an XML representation of a TestInfo object. +void XmlUnitTestResultPrinter::OutputXmlTestInfo(::std::ostream* stream, + const char* test_suite_name, + const TestInfo& test_info) { + const TestResult& result = *test_info.result(); + const std::string kTestsuite = "testcase"; + + if (test_info.is_in_another_shard()) { + return; + } + + *stream << " \n"; + return; + } + + OutputXmlAttribute(stream, kTestsuite, "status", + test_info.should_run() ? "run" : "notrun"); + OutputXmlAttribute(stream, kTestsuite, "result", + test_info.should_run() + ? (result.Skipped() ? "skipped" : "completed") + : "suppressed"); + OutputXmlAttribute(stream, kTestsuite, "time", + FormatTimeInMillisAsSeconds(result.elapsed_time())); + OutputXmlAttribute(stream, kTestsuite, "classname", test_suite_name); + + int failures = 0; + for (int i = 0; i < result.total_part_count(); ++i) { + const TestPartResult& part = result.GetTestPartResult(i); + if (part.failed()) { + if (++failures == 1) { + *stream << ">\n"; + } + const std::string location = + internal::FormatCompilerIndependentFileLocation(part.file_name(), + part.line_number()); + const std::string summary = location + "\n" + part.summary(); + *stream << " "; + const std::string detail = location + "\n" + part.message(); + OutputXmlCDataSection(stream, RemoveInvalidXmlCharacters(detail).c_str()); + *stream << "\n"; + } + } + + if (failures == 0 && result.test_property_count() == 0) { + *stream << " />\n"; + } else { + if (failures == 0) { + *stream << ">\n"; + } + OutputXmlTestProperties(stream, result); + *stream << " \n"; + } +} + +// Prints an XML representation of a TestSuite object +void XmlUnitTestResultPrinter::PrintXmlTestSuite(std::ostream* stream, + const TestSuite& test_suite) { + const std::string kTestsuite = "testsuite"; + *stream << " <" << kTestsuite; + OutputXmlAttribute(stream, kTestsuite, "name", test_suite.name()); + OutputXmlAttribute(stream, kTestsuite, "tests", + StreamableToString(test_suite.reportable_test_count())); + if (!GTEST_FLAG(list_tests)) { + OutputXmlAttribute(stream, kTestsuite, "failures", + StreamableToString(test_suite.failed_test_count())); + OutputXmlAttribute( + stream, kTestsuite, "disabled", + StreamableToString(test_suite.reportable_disabled_test_count())); + OutputXmlAttribute(stream, kTestsuite, "errors", "0"); + OutputXmlAttribute(stream, kTestsuite, "time", + FormatTimeInMillisAsSeconds(test_suite.elapsed_time())); + *stream << TestPropertiesAsXmlAttributes(test_suite.ad_hoc_test_result()); + } + *stream << ">\n"; + for (int i = 0; i < test_suite.total_test_count(); ++i) { + if (test_suite.GetTestInfo(i)->is_reportable()) + OutputXmlTestInfo(stream, test_suite.name(), *test_suite.GetTestInfo(i)); + } + *stream << " \n"; +} + +// Prints an XML summary of unit_test to output stream out. +void XmlUnitTestResultPrinter::PrintXmlUnitTest(std::ostream* stream, + const UnitTest& unit_test) { + const std::string kTestsuites = "testsuites"; + + *stream << "\n"; + *stream << "<" << kTestsuites; + + OutputXmlAttribute(stream, kTestsuites, "tests", + StreamableToString(unit_test.reportable_test_count())); + OutputXmlAttribute(stream, kTestsuites, "failures", + StreamableToString(unit_test.failed_test_count())); + OutputXmlAttribute( + stream, kTestsuites, "disabled", + StreamableToString(unit_test.reportable_disabled_test_count())); + OutputXmlAttribute(stream, kTestsuites, "errors", "0"); + OutputXmlAttribute( + stream, kTestsuites, "timestamp", + FormatEpochTimeInMillisAsIso8601(unit_test.start_timestamp())); + OutputXmlAttribute(stream, kTestsuites, "time", + FormatTimeInMillisAsSeconds(unit_test.elapsed_time())); + + if (GTEST_FLAG(shuffle)) { + OutputXmlAttribute(stream, kTestsuites, "random_seed", + StreamableToString(unit_test.random_seed())); + } + *stream << TestPropertiesAsXmlAttributes(unit_test.ad_hoc_test_result()); + + OutputXmlAttribute(stream, kTestsuites, "name", "AllTests"); + *stream << ">\n"; + + for (int i = 0; i < unit_test.total_test_suite_count(); ++i) { + if (unit_test.GetTestSuite(i)->reportable_test_count() > 0) + PrintXmlTestSuite(stream, *unit_test.GetTestSuite(i)); + } + *stream << "\n"; +} + +void XmlUnitTestResultPrinter::PrintXmlTestsList( + std::ostream* stream, const std::vector& test_suites) { + const std::string kTestsuites = "testsuites"; + + *stream << "\n"; + *stream << "<" << kTestsuites; + + int total_tests = 0; + for (auto test_suite : test_suites) { + total_tests += test_suite->total_test_count(); + } + OutputXmlAttribute(stream, kTestsuites, "tests", + StreamableToString(total_tests)); + OutputXmlAttribute(stream, kTestsuites, "name", "AllTests"); + *stream << ">\n"; + + for (auto test_suite : test_suites) { + PrintXmlTestSuite(stream, *test_suite); + } + *stream << "\n"; +} + +// Produces a string representing the test properties in a result as space +// delimited XML attributes based on the property key="value" pairs. +std::string XmlUnitTestResultPrinter::TestPropertiesAsXmlAttributes( + const TestResult& result) { + Message attributes; + for (int i = 0; i < result.test_property_count(); ++i) { + const TestProperty& property = result.GetTestProperty(i); + attributes << " " << property.key() << "=" + << "\"" << EscapeXmlAttribute(property.value()) << "\""; + } + return attributes.GetString(); +} + +void XmlUnitTestResultPrinter::OutputXmlTestProperties( + std::ostream* stream, const TestResult& result) { + const std::string kProperties = "properties"; + const std::string kProperty = "property"; + + if (result.test_property_count() <= 0) { + return; + } + + *stream << "<" << kProperties << ">\n"; + for (int i = 0; i < result.test_property_count(); ++i) { + const TestProperty& property = result.GetTestProperty(i); + *stream << "<" << kProperty; + *stream << " name=\"" << EscapeXmlAttribute(property.key()) << "\""; + *stream << " value=\"" << EscapeXmlAttribute(property.value()) << "\""; + *stream << "/>\n"; + } + *stream << "\n"; +} + +// End XmlUnitTestResultPrinter + +// This class generates an JSON output file. +class JsonUnitTestResultPrinter : public EmptyTestEventListener { + public: + explicit JsonUnitTestResultPrinter(const char* output_file); + + void OnTestIterationEnd(const UnitTest& unit_test, int iteration) override; + + // Prints an JSON summary of all unit tests. + static void PrintJsonTestList(::std::ostream* stream, + const std::vector& test_suites); + + private: + // Returns an JSON-escaped copy of the input string str. + static std::string EscapeJson(const std::string& str); + + //// Verifies that the given attribute belongs to the given element and + //// streams the attribute as JSON. + static void OutputJsonKey(std::ostream* stream, + const std::string& element_name, + const std::string& name, + const std::string& value, + const std::string& indent, + bool comma = true); + static void OutputJsonKey(std::ostream* stream, + const std::string& element_name, + const std::string& name, + int value, + const std::string& indent, + bool comma = true); + + // Streams a JSON representation of a TestInfo object. + static void OutputJsonTestInfo(::std::ostream* stream, + const char* test_suite_name, + const TestInfo& test_info); + + // Prints a JSON representation of a TestSuite object + static void PrintJsonTestSuite(::std::ostream* stream, + const TestSuite& test_suite); + + // Prints a JSON summary of unit_test to output stream out. + static void PrintJsonUnitTest(::std::ostream* stream, + const UnitTest& unit_test); + + // Produces a string representing the test properties in a result as + // a JSON dictionary. + static std::string TestPropertiesAsJson(const TestResult& result, + const std::string& indent); + + // The output file. + const std::string output_file_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(JsonUnitTestResultPrinter); +}; + +// Creates a new JsonUnitTestResultPrinter. +JsonUnitTestResultPrinter::JsonUnitTestResultPrinter(const char* output_file) + : output_file_(output_file) { + if (output_file_.empty()) { + GTEST_LOG_(FATAL) << "JSON output file may not be null"; + } +} + +void JsonUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test, + int /*iteration*/) { + FILE* jsonout = OpenFileForWriting(output_file_); + std::stringstream stream; + PrintJsonUnitTest(&stream, unit_test); + fprintf(jsonout, "%s", StringStreamToString(&stream).c_str()); + fclose(jsonout); +} + +// Returns an JSON-escaped copy of the input string str. +std::string JsonUnitTestResultPrinter::EscapeJson(const std::string& str) { + Message m; + + for (size_t i = 0; i < str.size(); ++i) { + const char ch = str[i]; + switch (ch) { + case '\\': + case '"': + case '/': + m << '\\' << ch; + break; + case '\b': + m << "\\b"; + break; + case '\t': + m << "\\t"; + break; + case '\n': + m << "\\n"; + break; + case '\f': + m << "\\f"; + break; + case '\r': + m << "\\r"; + break; + default: + if (ch < ' ') { + m << "\\u00" << String::FormatByte(static_cast(ch)); + } else { + m << ch; + } + break; + } + } + + return m.GetString(); +} + +// The following routines generate an JSON representation of a UnitTest +// object. + +// Formats the given time in milliseconds as seconds. +static std::string FormatTimeInMillisAsDuration(TimeInMillis ms) { + ::std::stringstream ss; + ss << (static_cast(ms) * 1e-3) << "s"; + return ss.str(); +} + +// Converts the given epoch time in milliseconds to a date string in the +// RFC3339 format, without the timezone information. +static std::string FormatEpochTimeInMillisAsRFC3339(TimeInMillis ms) { + struct tm time_struct; + if (!PortableLocaltime(static_cast(ms / 1000), &time_struct)) + return ""; + // YYYY-MM-DDThh:mm:ss + return StreamableToString(time_struct.tm_year + 1900) + "-" + + String::FormatIntWidth2(time_struct.tm_mon + 1) + "-" + + String::FormatIntWidth2(time_struct.tm_mday) + "T" + + String::FormatIntWidth2(time_struct.tm_hour) + ":" + + String::FormatIntWidth2(time_struct.tm_min) + ":" + + String::FormatIntWidth2(time_struct.tm_sec) + "Z"; +} + +static inline std::string Indent(size_t width) { + return std::string(width, ' '); +} + +void JsonUnitTestResultPrinter::OutputJsonKey( + std::ostream* stream, + const std::string& element_name, + const std::string& name, + const std::string& value, + const std::string& indent, + bool comma) { + const std::vector& allowed_names = + GetReservedOutputAttributesForElement(element_name); + + GTEST_CHECK_(std::find(allowed_names.begin(), allowed_names.end(), name) != + allowed_names.end()) + << "Key \"" << name << "\" is not allowed for value \"" << element_name + << "\"."; + + *stream << indent << "\"" << name << "\": \"" << EscapeJson(value) << "\""; + if (comma) + *stream << ",\n"; +} + +void JsonUnitTestResultPrinter::OutputJsonKey( + std::ostream* stream, + const std::string& element_name, + const std::string& name, + int value, + const std::string& indent, + bool comma) { + const std::vector& allowed_names = + GetReservedOutputAttributesForElement(element_name); + + GTEST_CHECK_(std::find(allowed_names.begin(), allowed_names.end(), name) != + allowed_names.end()) + << "Key \"" << name << "\" is not allowed for value \"" << element_name + << "\"."; + + *stream << indent << "\"" << name << "\": " << StreamableToString(value); + if (comma) + *stream << ",\n"; +} + +// Prints a JSON representation of a TestInfo object. +void JsonUnitTestResultPrinter::OutputJsonTestInfo(::std::ostream* stream, + const char* test_suite_name, + const TestInfo& test_info) { + const TestResult& result = *test_info.result(); + const std::string kTestsuite = "testcase"; + const std::string kIndent = Indent(10); + + *stream << Indent(8) << "{\n"; + OutputJsonKey(stream, kTestsuite, "name", test_info.name(), kIndent); + + if (test_info.value_param() != nullptr) { + OutputJsonKey(stream, kTestsuite, "value_param", test_info.value_param(), + kIndent); + } + if (test_info.type_param() != nullptr) { + OutputJsonKey(stream, kTestsuite, "type_param", test_info.type_param(), + kIndent); + } + if (GTEST_FLAG(list_tests)) { + OutputJsonKey(stream, kTestsuite, "file", test_info.file(), kIndent); + OutputJsonKey(stream, kTestsuite, "line", test_info.line(), kIndent, false); + *stream << "\n" << Indent(8) << "}"; + return; + } + + OutputJsonKey(stream, kTestsuite, "status", + test_info.should_run() ? "RUN" : "NOTRUN", kIndent); + OutputJsonKey(stream, kTestsuite, "result", + test_info.should_run() + ? (result.Skipped() ? "SKIPPED" : "COMPLETED") + : "SUPPRESSED", + kIndent); + OutputJsonKey(stream, kTestsuite, "time", + FormatTimeInMillisAsDuration(result.elapsed_time()), kIndent); + OutputJsonKey(stream, kTestsuite, "classname", test_suite_name, kIndent, + false); + *stream << TestPropertiesAsJson(result, kIndent); + + int failures = 0; + for (int i = 0; i < result.total_part_count(); ++i) { + const TestPartResult& part = result.GetTestPartResult(i); + if (part.failed()) { + *stream << ",\n"; + if (++failures == 1) { + *stream << kIndent << "\"" << "failures" << "\": [\n"; + } + const std::string location = + internal::FormatCompilerIndependentFileLocation(part.file_name(), + part.line_number()); + const std::string message = EscapeJson(location + "\n" + part.message()); + *stream << kIndent << " {\n" + << kIndent << " \"failure\": \"" << message << "\",\n" + << kIndent << " \"type\": \"\"\n" + << kIndent << " }"; + } + } + + if (failures > 0) + *stream << "\n" << kIndent << "]"; + *stream << "\n" << Indent(8) << "}"; +} + +// Prints an JSON representation of a TestSuite object +void JsonUnitTestResultPrinter::PrintJsonTestSuite( + std::ostream* stream, const TestSuite& test_suite) { + const std::string kTestsuite = "testsuite"; + const std::string kIndent = Indent(6); + + *stream << Indent(4) << "{\n"; + OutputJsonKey(stream, kTestsuite, "name", test_suite.name(), kIndent); + OutputJsonKey(stream, kTestsuite, "tests", test_suite.reportable_test_count(), + kIndent); + if (!GTEST_FLAG(list_tests)) { + OutputJsonKey(stream, kTestsuite, "failures", + test_suite.failed_test_count(), kIndent); + OutputJsonKey(stream, kTestsuite, "disabled", + test_suite.reportable_disabled_test_count(), kIndent); + OutputJsonKey(stream, kTestsuite, "errors", 0, kIndent); + OutputJsonKey(stream, kTestsuite, "time", + FormatTimeInMillisAsDuration(test_suite.elapsed_time()), + kIndent, false); + *stream << TestPropertiesAsJson(test_suite.ad_hoc_test_result(), kIndent) + << ",\n"; + } + + *stream << kIndent << "\"" << kTestsuite << "\": [\n"; + + bool comma = false; + for (int i = 0; i < test_suite.total_test_count(); ++i) { + if (test_suite.GetTestInfo(i)->is_reportable()) { + if (comma) { + *stream << ",\n"; + } else { + comma = true; + } + OutputJsonTestInfo(stream, test_suite.name(), *test_suite.GetTestInfo(i)); + } + } + *stream << "\n" << kIndent << "]\n" << Indent(4) << "}"; +} + +// Prints a JSON summary of unit_test to output stream out. +void JsonUnitTestResultPrinter::PrintJsonUnitTest(std::ostream* stream, + const UnitTest& unit_test) { + const std::string kTestsuites = "testsuites"; + const std::string kIndent = Indent(2); + *stream << "{\n"; + + OutputJsonKey(stream, kTestsuites, "tests", unit_test.reportable_test_count(), + kIndent); + OutputJsonKey(stream, kTestsuites, "failures", unit_test.failed_test_count(), + kIndent); + OutputJsonKey(stream, kTestsuites, "disabled", + unit_test.reportable_disabled_test_count(), kIndent); + OutputJsonKey(stream, kTestsuites, "errors", 0, kIndent); + if (GTEST_FLAG(shuffle)) { + OutputJsonKey(stream, kTestsuites, "random_seed", unit_test.random_seed(), + kIndent); + } + OutputJsonKey(stream, kTestsuites, "timestamp", + FormatEpochTimeInMillisAsRFC3339(unit_test.start_timestamp()), + kIndent); + OutputJsonKey(stream, kTestsuites, "time", + FormatTimeInMillisAsDuration(unit_test.elapsed_time()), kIndent, + false); + + *stream << TestPropertiesAsJson(unit_test.ad_hoc_test_result(), kIndent) + << ",\n"; + + OutputJsonKey(stream, kTestsuites, "name", "AllTests", kIndent); + *stream << kIndent << "\"" << kTestsuites << "\": [\n"; + + bool comma = false; + for (int i = 0; i < unit_test.total_test_suite_count(); ++i) { + if (unit_test.GetTestSuite(i)->reportable_test_count() > 0) { + if (comma) { + *stream << ",\n"; + } else { + comma = true; + } + PrintJsonTestSuite(stream, *unit_test.GetTestSuite(i)); + } + } + + *stream << "\n" << kIndent << "]\n" << "}\n"; +} + +void JsonUnitTestResultPrinter::PrintJsonTestList( + std::ostream* stream, const std::vector& test_suites) { + const std::string kTestsuites = "testsuites"; + const std::string kIndent = Indent(2); + *stream << "{\n"; + int total_tests = 0; + for (auto test_suite : test_suites) { + total_tests += test_suite->total_test_count(); + } + OutputJsonKey(stream, kTestsuites, "tests", total_tests, kIndent); + + OutputJsonKey(stream, kTestsuites, "name", "AllTests", kIndent); + *stream << kIndent << "\"" << kTestsuites << "\": [\n"; + + for (size_t i = 0; i < test_suites.size(); ++i) { + if (i != 0) { + *stream << ",\n"; + } + PrintJsonTestSuite(stream, *test_suites[i]); + } + + *stream << "\n" + << kIndent << "]\n" + << "}\n"; +} +// Produces a string representing the test properties in a result as +// a JSON dictionary. +std::string JsonUnitTestResultPrinter::TestPropertiesAsJson( + const TestResult& result, const std::string& indent) { + Message attributes; + for (int i = 0; i < result.test_property_count(); ++i) { + const TestProperty& property = result.GetTestProperty(i); + attributes << ",\n" << indent << "\"" << property.key() << "\": " + << "\"" << EscapeJson(property.value()) << "\""; + } + return attributes.GetString(); +} + +// End JsonUnitTestResultPrinter + +#if GTEST_CAN_STREAM_RESULTS_ + +// Checks if str contains '=', '&', '%' or '\n' characters. If yes, +// replaces them by "%xx" where xx is their hexadecimal value. For +// example, replaces "=" with "%3D". This algorithm is O(strlen(str)) +// in both time and space -- important as the input str may contain an +// arbitrarily long test failure message and stack trace. +std::string StreamingListener::UrlEncode(const char* str) { + std::string result; + result.reserve(strlen(str) + 1); + for (char ch = *str; ch != '\0'; ch = *++str) { + switch (ch) { + case '%': + case '=': + case '&': + case '\n': + result.append("%" + String::FormatByte(static_cast(ch))); + break; + default: + result.push_back(ch); + break; + } + } + return result; +} + +void StreamingListener::SocketWriter::MakeConnection() { + GTEST_CHECK_(sockfd_ == -1) + << "MakeConnection() can't be called when there is already a connection."; + + addrinfo hints; + memset(&hints, 0, sizeof(hints)); + hints.ai_family = AF_UNSPEC; // To allow both IPv4 and IPv6 addresses. + hints.ai_socktype = SOCK_STREAM; + addrinfo* servinfo = nullptr; + + // Use the getaddrinfo() to get a linked list of IP addresses for + // the given host name. + const int error_num = getaddrinfo( + host_name_.c_str(), port_num_.c_str(), &hints, &servinfo); + if (error_num != 0) { + GTEST_LOG_(WARNING) << "stream_result_to: getaddrinfo() failed: " + << gai_strerror(error_num); + } + + // Loop through all the results and connect to the first we can. + for (addrinfo* cur_addr = servinfo; sockfd_ == -1 && cur_addr != nullptr; + cur_addr = cur_addr->ai_next) { + sockfd_ = socket( + cur_addr->ai_family, cur_addr->ai_socktype, cur_addr->ai_protocol); + if (sockfd_ != -1) { + // Connect the client socket to the server socket. + if (connect(sockfd_, cur_addr->ai_addr, cur_addr->ai_addrlen) == -1) { + close(sockfd_); + sockfd_ = -1; + } + } + } + + freeaddrinfo(servinfo); // all done with this structure + + if (sockfd_ == -1) { + GTEST_LOG_(WARNING) << "stream_result_to: failed to connect to " + << host_name_ << ":" << port_num_; + } +} + +// End of class Streaming Listener +#endif // GTEST_CAN_STREAM_RESULTS__ + +// class OsStackTraceGetter + +const char* const OsStackTraceGetterInterface::kElidedFramesMarker = + "... " GTEST_NAME_ " internal frames ..."; + +std::string OsStackTraceGetter::CurrentStackTrace(int max_depth, int skip_count) + GTEST_LOCK_EXCLUDED_(mutex_) { +#if GTEST_HAS_ABSL + std::string result; + + if (max_depth <= 0) { + return result; + } + + max_depth = std::min(max_depth, kMaxStackTraceDepth); + + std::vector raw_stack(max_depth); + // Skips the frames requested by the caller, plus this function. + const int raw_stack_size = + absl::GetStackTrace(&raw_stack[0], max_depth, skip_count + 1); + + void* caller_frame = nullptr; + { + MutexLock lock(&mutex_); + caller_frame = caller_frame_; + } + + for (int i = 0; i < raw_stack_size; ++i) { + if (raw_stack[i] == caller_frame && + !GTEST_FLAG(show_internal_stack_frames)) { + // Add a marker to the trace and stop adding frames. + absl::StrAppend(&result, kElidedFramesMarker, "\n"); + break; + } + + char tmp[1024]; + const char* symbol = "(unknown)"; + if (absl::Symbolize(raw_stack[i], tmp, sizeof(tmp))) { + symbol = tmp; + } + + char line[1024]; + snprintf(line, sizeof(line), " %p: %s\n", raw_stack[i], symbol); + result += line; + } + + return result; + +#else // !GTEST_HAS_ABSL + static_cast(max_depth); + static_cast(skip_count); + return ""; +#endif // GTEST_HAS_ABSL +} + +void OsStackTraceGetter::UponLeavingGTest() GTEST_LOCK_EXCLUDED_(mutex_) { +#if GTEST_HAS_ABSL + void* caller_frame = nullptr; + if (absl::GetStackTrace(&caller_frame, 1, 3) <= 0) { + caller_frame = nullptr; + } + + MutexLock lock(&mutex_); + caller_frame_ = caller_frame; +#endif // GTEST_HAS_ABSL +} + +// A helper class that creates the premature-exit file in its +// constructor and deletes the file in its destructor. +class ScopedPrematureExitFile { + public: + explicit ScopedPrematureExitFile(const char* premature_exit_filepath) + : premature_exit_filepath_(premature_exit_filepath ? + premature_exit_filepath : "") { + // If a path to the premature-exit file is specified... + if (!premature_exit_filepath_.empty()) { + // create the file with a single "0" character in it. I/O + // errors are ignored as there's nothing better we can do and we + // don't want to fail the test because of this. + FILE* pfile = posix::FOpen(premature_exit_filepath, "w"); + fwrite("0", 1, 1, pfile); + fclose(pfile); + } + } + + ~ScopedPrematureExitFile() { + if (!premature_exit_filepath_.empty()) { + int retval = remove(premature_exit_filepath_.c_str()); + if (retval) { + GTEST_LOG_(ERROR) << "Failed to remove premature exit filepath \"" + << premature_exit_filepath_ << "\" with error " + << retval; + } + } + } + + private: + const std::string premature_exit_filepath_; + + GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedPrematureExitFile); +}; + +} // namespace internal + +// class TestEventListeners + +TestEventListeners::TestEventListeners() + : repeater_(new internal::TestEventRepeater()), + default_result_printer_(nullptr), + default_xml_generator_(nullptr) {} + +TestEventListeners::~TestEventListeners() { delete repeater_; } + +// Returns the standard listener responsible for the default console +// output. Can be removed from the listeners list to shut down default +// console output. Note that removing this object from the listener list +// with Release transfers its ownership to the user. +void TestEventListeners::Append(TestEventListener* listener) { + repeater_->Append(listener); +} + +// Removes the given event listener from the list and returns it. It then +// becomes the caller's responsibility to delete the listener. Returns +// NULL if the listener is not found in the list. +TestEventListener* TestEventListeners::Release(TestEventListener* listener) { + if (listener == default_result_printer_) + default_result_printer_ = nullptr; + else if (listener == default_xml_generator_) + default_xml_generator_ = nullptr; + return repeater_->Release(listener); +} + +// Returns repeater that broadcasts the TestEventListener events to all +// subscribers. +TestEventListener* TestEventListeners::repeater() { return repeater_; } + +// Sets the default_result_printer attribute to the provided listener. +// The listener is also added to the listener list and previous +// default_result_printer is removed from it and deleted. The listener can +// also be NULL in which case it will not be added to the list. Does +// nothing if the previous and the current listener objects are the same. +void TestEventListeners::SetDefaultResultPrinter(TestEventListener* listener) { + if (default_result_printer_ != listener) { + // It is an error to pass this method a listener that is already in the + // list. + delete Release(default_result_printer_); + default_result_printer_ = listener; + if (listener != nullptr) Append(listener); + } +} + +// Sets the default_xml_generator attribute to the provided listener. The +// listener is also added to the listener list and previous +// default_xml_generator is removed from it and deleted. The listener can +// also be NULL in which case it will not be added to the list. Does +// nothing if the previous and the current listener objects are the same. +void TestEventListeners::SetDefaultXmlGenerator(TestEventListener* listener) { + if (default_xml_generator_ != listener) { + // It is an error to pass this method a listener that is already in the + // list. + delete Release(default_xml_generator_); + default_xml_generator_ = listener; + if (listener != nullptr) Append(listener); + } +} + +// Controls whether events will be forwarded by the repeater to the +// listeners in the list. +bool TestEventListeners::EventForwardingEnabled() const { + return repeater_->forwarding_enabled(); +} + +void TestEventListeners::SuppressEventForwarding() { + repeater_->set_forwarding_enabled(false); +} + +// class UnitTest + +// Gets the singleton UnitTest object. The first time this method is +// called, a UnitTest object is constructed and returned. Consecutive +// calls will return the same object. +// +// We don't protect this under mutex_ as a user is not supposed to +// call this before main() starts, from which point on the return +// value will never change. +UnitTest* UnitTest::GetInstance() { + // CodeGear C++Builder insists on a public destructor for the + // default implementation. Use this implementation to keep good OO + // design with private destructor. + +#if defined(__BORLANDC__) + static UnitTest* const instance = new UnitTest; + return instance; +#else + static UnitTest instance; + return &instance; +#endif // defined(__BORLANDC__) +} + +// Gets the number of successful test suites. +int UnitTest::successful_test_suite_count() const { + return impl()->successful_test_suite_count(); +} + +// Gets the number of failed test suites. +int UnitTest::failed_test_suite_count() const { + return impl()->failed_test_suite_count(); +} + +// Gets the number of all test suites. +int UnitTest::total_test_suite_count() const { + return impl()->total_test_suite_count(); +} + +// Gets the number of all test suites that contain at least one test +// that should run. +int UnitTest::test_suite_to_run_count() const { + return impl()->test_suite_to_run_count(); +} + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +int UnitTest::successful_test_case_count() const { + return impl()->successful_test_suite_count(); +} +int UnitTest::failed_test_case_count() const { + return impl()->failed_test_suite_count(); +} +int UnitTest::total_test_case_count() const { + return impl()->total_test_suite_count(); +} +int UnitTest::test_case_to_run_count() const { + return impl()->test_suite_to_run_count(); +} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +// Gets the number of successful tests. +int UnitTest::successful_test_count() const { + return impl()->successful_test_count(); +} + +// Gets the number of skipped tests. +int UnitTest::skipped_test_count() const { + return impl()->skipped_test_count(); +} + +// Gets the number of failed tests. +int UnitTest::failed_test_count() const { return impl()->failed_test_count(); } + +// Gets the number of disabled tests that will be reported in the XML report. +int UnitTest::reportable_disabled_test_count() const { + return impl()->reportable_disabled_test_count(); +} + +// Gets the number of disabled tests. +int UnitTest::disabled_test_count() const { + return impl()->disabled_test_count(); +} + +// Gets the number of tests to be printed in the XML report. +int UnitTest::reportable_test_count() const { + return impl()->reportable_test_count(); +} + +// Gets the number of all tests. +int UnitTest::total_test_count() const { return impl()->total_test_count(); } + +// Gets the number of tests that should run. +int UnitTest::test_to_run_count() const { return impl()->test_to_run_count(); } + +// Gets the time of the test program start, in ms from the start of the +// UNIX epoch. +internal::TimeInMillis UnitTest::start_timestamp() const { + return impl()->start_timestamp(); +} + +// Gets the elapsed time, in milliseconds. +internal::TimeInMillis UnitTest::elapsed_time() const { + return impl()->elapsed_time(); +} + +// Returns true iff the unit test passed (i.e. all test suites passed). +bool UnitTest::Passed() const { return impl()->Passed(); } + +// Returns true iff the unit test failed (i.e. some test suite failed +// or something outside of all tests failed). +bool UnitTest::Failed() const { return impl()->Failed(); } + +// Gets the i-th test suite among all the test suites. i can range from 0 to +// total_test_suite_count() - 1. If i is not in that range, returns NULL. +const TestSuite* UnitTest::GetTestSuite(int i) const { + return impl()->GetTestSuite(i); +} + +// Legacy API is deprecated but still available +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +const TestCase* UnitTest::GetTestCase(int i) const { + return impl()->GetTestCase(i); +} +#endif // GTEST_REMOVE_LEGACY_TEST_CASEAPI_ + +// Returns the TestResult containing information on test failures and +// properties logged outside of individual test suites. +const TestResult& UnitTest::ad_hoc_test_result() const { + return *impl()->ad_hoc_test_result(); +} + +// Gets the i-th test suite among all the test suites. i can range from 0 to +// total_test_suite_count() - 1. If i is not in that range, returns NULL. +TestSuite* UnitTest::GetMutableTestSuite(int i) { + return impl()->GetMutableSuiteCase(i); +} + +// Returns the list of event listeners that can be used to track events +// inside Google Test. +TestEventListeners& UnitTest::listeners() { + return *impl()->listeners(); +} + +// Registers and returns a global test environment. When a test +// program is run, all global test environments will be set-up in the +// order they were registered. After all tests in the program have +// finished, all global test environments will be torn-down in the +// *reverse* order they were registered. +// +// The UnitTest object takes ownership of the given environment. +// +// We don't protect this under mutex_, as we only support calling it +// from the main thread. +Environment* UnitTest::AddEnvironment(Environment* env) { + if (env == nullptr) { + return nullptr; + } + + impl_->environments().push_back(env); + return env; +} + +// Adds a TestPartResult to the current TestResult object. All Google Test +// assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) eventually call +// this to report their results. The user code should use the +// assertion macros instead of calling this directly. +void UnitTest::AddTestPartResult( + TestPartResult::Type result_type, + const char* file_name, + int line_number, + const std::string& message, + const std::string& os_stack_trace) GTEST_LOCK_EXCLUDED_(mutex_) { + Message msg; + msg << message; + + internal::MutexLock lock(&mutex_); + if (impl_->gtest_trace_stack().size() > 0) { + msg << "\n" << GTEST_NAME_ << " trace:"; + + for (size_t i = impl_->gtest_trace_stack().size(); i > 0; --i) { + const internal::TraceInfo& trace = impl_->gtest_trace_stack()[i - 1]; + msg << "\n" << internal::FormatFileLocation(trace.file, trace.line) + << " " << trace.message; + } + } + + if (os_stack_trace.c_str() != nullptr && !os_stack_trace.empty()) { + msg << internal::kStackTraceMarker << os_stack_trace; + } + + const TestPartResult result = TestPartResult( + result_type, file_name, line_number, msg.GetString().c_str()); + impl_->GetTestPartResultReporterForCurrentThread()-> + ReportTestPartResult(result); + + if (result_type != TestPartResult::kSuccess && + result_type != TestPartResult::kSkip) { + // gtest_break_on_failure takes precedence over + // gtest_throw_on_failure. This allows a user to set the latter + // in the code (perhaps in order to use Google Test assertions + // with another testing framework) and specify the former on the + // command line for debugging. + if (GTEST_FLAG(break_on_failure)) { +#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT + // Using DebugBreak on Windows allows gtest to still break into a debugger + // when a failure happens and both the --gtest_break_on_failure and + // the --gtest_catch_exceptions flags are specified. + DebugBreak(); +#elif (!defined(__native_client__)) && \ + ((defined(__clang__) || defined(__GNUC__)) && \ + (defined(__x86_64__) || defined(__i386__))) + // with clang/gcc we can achieve the same effect on x86 by invoking int3 + asm("int3"); +#else + // Dereference nullptr through a volatile pointer to prevent the compiler + // from removing. We use this rather than abort() or __builtin_trap() for + // portability: some debuggers don't correctly trap abort(). + *static_cast(nullptr) = 1; +#endif // GTEST_OS_WINDOWS + } else if (GTEST_FLAG(throw_on_failure)) { +#if GTEST_HAS_EXCEPTIONS + throw internal::GoogleTestFailureException(result); +#else + // We cannot call abort() as it generates a pop-up in debug mode + // that cannot be suppressed in VC 7.1 or below. + exit(1); +#endif + } + } +} + +// Adds a TestProperty to the current TestResult object when invoked from +// inside a test, to current TestSuite's ad_hoc_test_result_ when invoked +// from SetUpTestSuite or TearDownTestSuite, or to the global property set +// when invoked elsewhere. If the result already contains a property with +// the same key, the value will be updated. +void UnitTest::RecordProperty(const std::string& key, + const std::string& value) { + impl_->RecordProperty(TestProperty(key, value)); +} + +// Runs all tests in this UnitTest object and prints the result. +// Returns 0 if successful, or 1 otherwise. +// +// We don't protect this under mutex_, as we only support calling it +// from the main thread. +int UnitTest::Run() { + const bool in_death_test_child_process = + internal::GTEST_FLAG(internal_run_death_test).length() > 0; + + // Google Test implements this protocol for catching that a test + // program exits before returning control to Google Test: + // + // 1. Upon start, Google Test creates a file whose absolute path + // is specified by the environment variable + // TEST_PREMATURE_EXIT_FILE. + // 2. When Google Test has finished its work, it deletes the file. + // + // This allows a test runner to set TEST_PREMATURE_EXIT_FILE before + // running a Google-Test-based test program and check the existence + // of the file at the end of the test execution to see if it has + // exited prematurely. + + // If we are in the child process of a death test, don't + // create/delete the premature exit file, as doing so is unnecessary + // and will confuse the parent process. Otherwise, create/delete + // the file upon entering/leaving this function. If the program + // somehow exits before this function has a chance to return, the + // premature-exit file will be left undeleted, causing a test runner + // that understands the premature-exit-file protocol to report the + // test as having failed. + const internal::ScopedPrematureExitFile premature_exit_file( + in_death_test_child_process + ? nullptr + : internal::posix::GetEnv("TEST_PREMATURE_EXIT_FILE")); + + // Captures the value of GTEST_FLAG(catch_exceptions). This value will be + // used for the duration of the program. + impl()->set_catch_exceptions(GTEST_FLAG(catch_exceptions)); + +#if GTEST_OS_WINDOWS + // Either the user wants Google Test to catch exceptions thrown by the + // tests or this is executing in the context of death test child + // process. In either case the user does not want to see pop-up dialogs + // about crashes - they are expected. + if (impl()->catch_exceptions() || in_death_test_child_process) { +# if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT + // SetErrorMode doesn't exist on CE. + SetErrorMode(SEM_FAILCRITICALERRORS | SEM_NOALIGNMENTFAULTEXCEPT | + SEM_NOGPFAULTERRORBOX | SEM_NOOPENFILEERRORBOX); +# endif // !GTEST_OS_WINDOWS_MOBILE + +# if (defined(_MSC_VER) || GTEST_OS_WINDOWS_MINGW) && !GTEST_OS_WINDOWS_MOBILE + // Death test children can be terminated with _abort(). On Windows, + // _abort() can show a dialog with a warning message. This forces the + // abort message to go to stderr instead. + _set_error_mode(_OUT_TO_STDERR); +# endif + +# if defined(_MSC_VER) && !GTEST_OS_WINDOWS_MOBILE + // In the debug version, Visual Studio pops up a separate dialog + // offering a choice to debug the aborted program. We need to suppress + // this dialog or it will pop up for every EXPECT/ASSERT_DEATH statement + // executed. Google Test will notify the user of any unexpected + // failure via stderr. + if (!GTEST_FLAG(break_on_failure)) + _set_abort_behavior( + 0x0, // Clear the following flags: + _WRITE_ABORT_MSG | _CALL_REPORTFAULT); // pop-up window, core dump. +# endif + } +#endif // GTEST_OS_WINDOWS + + return internal::HandleExceptionsInMethodIfSupported( + impl(), + &internal::UnitTestImpl::RunAllTests, + "auxiliary test code (environments or event listeners)") ? 0 : 1; +} + +// Returns the working directory when the first TEST() or TEST_F() was +// executed. +const char* UnitTest::original_working_dir() const { + return impl_->original_working_dir_.c_str(); +} + +// Returns the TestSuite object for the test that's currently running, +// or NULL if no test is running. +const TestSuite* UnitTest::current_test_suite() const + GTEST_LOCK_EXCLUDED_(mutex_) { + internal::MutexLock lock(&mutex_); + return impl_->current_test_suite(); +} + +// Legacy API is still available but deprecated +#ifndef GTEST_REMOVE_LEGACY_TEST_CASEAPI_ +const TestCase* UnitTest::current_test_case() const + GTEST_LOCK_EXCLUDED_(mutex_) { + internal::MutexLock lock(&mutex_); + return impl_->current_test_suite(); +} +#endif + +// Returns the TestInfo object for the test that's currently running, +// or NULL if no test is running. +const TestInfo* UnitTest::current_test_info() const + GTEST_LOCK_EXCLUDED_(mutex_) { + internal::MutexLock lock(&mutex_); + return impl_->current_test_info(); +} + +// Returns the random seed used at the start of the current test run. +int UnitTest::random_seed() const { return impl_->random_seed(); } + +// Returns ParameterizedTestSuiteRegistry object used to keep track of +// value-parameterized tests and instantiate and register them. +internal::ParameterizedTestSuiteRegistry& +UnitTest::parameterized_test_registry() GTEST_LOCK_EXCLUDED_(mutex_) { + return impl_->parameterized_test_registry(); +} + +// Creates an empty UnitTest. +UnitTest::UnitTest() { + impl_ = new internal::UnitTestImpl(this); +} + +// Destructor of UnitTest. +UnitTest::~UnitTest() { + delete impl_; +} + +// Pushes a trace defined by SCOPED_TRACE() on to the per-thread +// Google Test trace stack. +void UnitTest::PushGTestTrace(const internal::TraceInfo& trace) + GTEST_LOCK_EXCLUDED_(mutex_) { + internal::MutexLock lock(&mutex_); + impl_->gtest_trace_stack().push_back(trace); +} + +// Pops a trace from the per-thread Google Test trace stack. +void UnitTest::PopGTestTrace() + GTEST_LOCK_EXCLUDED_(mutex_) { + internal::MutexLock lock(&mutex_); + impl_->gtest_trace_stack().pop_back(); +} + +namespace internal { + +UnitTestImpl::UnitTestImpl(UnitTest* parent) + : parent_(parent), + GTEST_DISABLE_MSC_WARNINGS_PUSH_(4355 /* using this in initializer */) + default_global_test_part_result_reporter_(this), + default_per_thread_test_part_result_reporter_(this), + GTEST_DISABLE_MSC_WARNINGS_POP_() global_test_part_result_repoter_( + &default_global_test_part_result_reporter_), + per_thread_test_part_result_reporter_( + &default_per_thread_test_part_result_reporter_), + parameterized_test_registry_(), + parameterized_tests_registered_(false), + last_death_test_suite_(-1), + current_test_suite_(nullptr), + current_test_info_(nullptr), + ad_hoc_test_result_(), + os_stack_trace_getter_(nullptr), + post_flag_parse_init_performed_(false), + random_seed_(0), // Will be overridden by the flag before first use. + random_(0), // Will be reseeded before first use. + start_timestamp_(0), + elapsed_time_(0), +#if GTEST_HAS_DEATH_TEST + death_test_factory_(new DefaultDeathTestFactory), +#endif + // Will be overridden by the flag before first use. + catch_exceptions_(false) { + listeners()->SetDefaultResultPrinter(new PrettyUnitTestResultPrinter); +} + +UnitTestImpl::~UnitTestImpl() { + // Deletes every TestSuite. + ForEach(test_suites_, internal::Delete); + + // Deletes every Environment. + ForEach(environments_, internal::Delete); + + delete os_stack_trace_getter_; +} + +// Adds a TestProperty to the current TestResult object when invoked in a +// context of a test, to current test suite's ad_hoc_test_result when invoke +// from SetUpTestSuite/TearDownTestSuite, or to the global property set +// otherwise. If the result already contains a property with the same key, +// the value will be updated. +void UnitTestImpl::RecordProperty(const TestProperty& test_property) { + std::string xml_element; + TestResult* test_result; // TestResult appropriate for property recording. + + if (current_test_info_ != nullptr) { + xml_element = "testcase"; + test_result = &(current_test_info_->result_); + } else if (current_test_suite_ != nullptr) { + xml_element = "testsuite"; + test_result = &(current_test_suite_->ad_hoc_test_result_); + } else { + xml_element = "testsuites"; + test_result = &ad_hoc_test_result_; + } + test_result->RecordProperty(xml_element, test_property); +} + +#if GTEST_HAS_DEATH_TEST +// Disables event forwarding if the control is currently in a death test +// subprocess. Must not be called before InitGoogleTest. +void UnitTestImpl::SuppressTestEventsIfInSubprocess() { + if (internal_run_death_test_flag_.get() != nullptr) + listeners()->SuppressEventForwarding(); +} +#endif // GTEST_HAS_DEATH_TEST + +// Initializes event listeners performing XML output as specified by +// UnitTestOptions. Must not be called before InitGoogleTest. +void UnitTestImpl::ConfigureXmlOutput() { + const std::string& output_format = UnitTestOptions::GetOutputFormat(); + if (output_format == "xml") { + listeners()->SetDefaultXmlGenerator(new XmlUnitTestResultPrinter( + UnitTestOptions::GetAbsolutePathToOutputFile().c_str())); + } else if (output_format == "json") { + listeners()->SetDefaultXmlGenerator(new JsonUnitTestResultPrinter( + UnitTestOptions::GetAbsolutePathToOutputFile().c_str())); + } else if (output_format != "") { + GTEST_LOG_(WARNING) << "WARNING: unrecognized output format \"" + << output_format << "\" ignored."; + } +} + +#if GTEST_CAN_STREAM_RESULTS_ +// Initializes event listeners for streaming test results in string form. +// Must not be called before InitGoogleTest. +void UnitTestImpl::ConfigureStreamingOutput() { + const std::string& target = GTEST_FLAG(stream_result_to); + if (!target.empty()) { + const size_t pos = target.find(':'); + if (pos != std::string::npos) { + listeners()->Append(new StreamingListener(target.substr(0, pos), + target.substr(pos+1))); + } else { + GTEST_LOG_(WARNING) << "unrecognized streaming target \"" << target + << "\" ignored."; + } + } +} +#endif // GTEST_CAN_STREAM_RESULTS_ + +// Performs initialization dependent upon flag values obtained in +// ParseGoogleTestFlagsOnly. Is called from InitGoogleTest after the call to +// ParseGoogleTestFlagsOnly. In case a user neglects to call InitGoogleTest +// this function is also called from RunAllTests. Since this function can be +// called more than once, it has to be idempotent. +void UnitTestImpl::PostFlagParsingInit() { + // Ensures that this function does not execute more than once. + if (!post_flag_parse_init_performed_) { + post_flag_parse_init_performed_ = true; + +#if defined(GTEST_CUSTOM_TEST_EVENT_LISTENER_) + // Register to send notifications about key process state changes. + listeners()->Append(new GTEST_CUSTOM_TEST_EVENT_LISTENER_()); +#endif // defined(GTEST_CUSTOM_TEST_EVENT_LISTENER_) + +#if GTEST_HAS_DEATH_TEST + InitDeathTestSubprocessControlInfo(); + SuppressTestEventsIfInSubprocess(); +#endif // GTEST_HAS_DEATH_TEST + + // Registers parameterized tests. This makes parameterized tests + // available to the UnitTest reflection API without running + // RUN_ALL_TESTS. + RegisterParameterizedTests(); + + // Configures listeners for XML output. This makes it possible for users + // to shut down the default XML output before invoking RUN_ALL_TESTS. + ConfigureXmlOutput(); + +#if GTEST_CAN_STREAM_RESULTS_ + // Configures listeners for streaming test results to the specified server. + ConfigureStreamingOutput(); +#endif // GTEST_CAN_STREAM_RESULTS_ + +#if GTEST_HAS_ABSL + if (GTEST_FLAG(install_failure_signal_handler)) { + absl::FailureSignalHandlerOptions options; + absl::InstallFailureSignalHandler(options); + } +#endif // GTEST_HAS_ABSL + } +} + +// A predicate that checks the name of a TestSuite against a known +// value. +// +// This is used for implementation of the UnitTest class only. We put +// it in the anonymous namespace to prevent polluting the outer +// namespace. +// +// TestSuiteNameIs is copyable. +class TestSuiteNameIs { + public: + // Constructor. + explicit TestSuiteNameIs(const std::string& name) : name_(name) {} + + // Returns true iff the name of test_suite matches name_. + bool operator()(const TestSuite* test_suite) const { + return test_suite != nullptr && + strcmp(test_suite->name(), name_.c_str()) == 0; + } + + private: + std::string name_; +}; + +// Finds and returns a TestSuite with the given name. If one doesn't +// exist, creates one and returns it. It's the CALLER'S +// RESPONSIBILITY to ensure that this function is only called WHEN THE +// TESTS ARE NOT SHUFFLED. +// +// Arguments: +// +// test_suite_name: name of the test suite +// type_param: the name of the test suite's type parameter, or NULL if +// this is not a typed or a type-parameterized test suite. +// set_up_tc: pointer to the function that sets up the test suite +// tear_down_tc: pointer to the function that tears down the test suite +TestSuite* UnitTestImpl::GetTestSuite( + const char* test_suite_name, const char* type_param, + internal::SetUpTestSuiteFunc set_up_tc, + internal::TearDownTestSuiteFunc tear_down_tc) { + // Can we find a TestSuite with the given name? + const auto test_suite = + std::find_if(test_suites_.rbegin(), test_suites_.rend(), + TestSuiteNameIs(test_suite_name)); + + if (test_suite != test_suites_.rend()) return *test_suite; + + // No. Let's create one. + auto* const new_test_suite = + new TestSuite(test_suite_name, type_param, set_up_tc, tear_down_tc); + + // Is this a death test suite? + if (internal::UnitTestOptions::MatchesFilter(test_suite_name, + kDeathTestSuiteFilter)) { + // Yes. Inserts the test suite after the last death test suite + // defined so far. This only works when the test suites haven't + // been shuffled. Otherwise we may end up running a death test + // after a non-death test. + ++last_death_test_suite_; + test_suites_.insert(test_suites_.begin() + last_death_test_suite_, + new_test_suite); + } else { + // No. Appends to the end of the list. + test_suites_.push_back(new_test_suite); + } + + test_suite_indices_.push_back(static_cast(test_suite_indices_.size())); + return new_test_suite; +} + +// Helpers for setting up / tearing down the given environment. They +// are for use in the ForEach() function. +static void SetUpEnvironment(Environment* env) { env->SetUp(); } +static void TearDownEnvironment(Environment* env) { env->TearDown(); } + +// Runs all tests in this UnitTest object, prints the result, and +// returns true if all tests are successful. If any exception is +// thrown during a test, the test is considered to be failed, but the +// rest of the tests will still be run. +// +// When parameterized tests are enabled, it expands and registers +// parameterized tests first in RegisterParameterizedTests(). +// All other functions called from RunAllTests() may safely assume that +// parameterized tests are ready to be counted and run. +bool UnitTestImpl::RunAllTests() { + // True iff Google Test is initialized before RUN_ALL_TESTS() is called. + const bool gtest_is_initialized_before_run_all_tests = GTestIsInitialized(); + + // Do not run any test if the --help flag was specified. + if (g_help_flag) + return true; + + // Repeats the call to the post-flag parsing initialization in case the + // user didn't call InitGoogleTest. + PostFlagParsingInit(); + + // Even if sharding is not on, test runners may want to use the + // GTEST_SHARD_STATUS_FILE to query whether the test supports the sharding + // protocol. + internal::WriteToShardStatusFileIfNeeded(); + + // True iff we are in a subprocess for running a thread-safe-style + // death test. + bool in_subprocess_for_death_test = false; + +#if GTEST_HAS_DEATH_TEST + in_subprocess_for_death_test = + (internal_run_death_test_flag_.get() != nullptr); +# if defined(GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_) + if (in_subprocess_for_death_test) { + GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_(); + } +# endif // defined(GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_) +#endif // GTEST_HAS_DEATH_TEST + + const bool should_shard = ShouldShard(kTestTotalShards, kTestShardIndex, + in_subprocess_for_death_test); + + // Compares the full test names with the filter to decide which + // tests to run. + const bool has_tests_to_run = FilterTests(should_shard + ? HONOR_SHARDING_PROTOCOL + : IGNORE_SHARDING_PROTOCOL) > 0; + + // Lists the tests and exits if the --gtest_list_tests flag was specified. + if (GTEST_FLAG(list_tests)) { + // This must be called *after* FilterTests() has been called. + ListTestsMatchingFilter(); + return true; + } + + random_seed_ = GTEST_FLAG(shuffle) ? + GetRandomSeedFromFlag(GTEST_FLAG(random_seed)) : 0; + + // True iff at least one test has failed. + bool failed = false; + + TestEventListener* repeater = listeners()->repeater(); + + start_timestamp_ = GetTimeInMillis(); + repeater->OnTestProgramStart(*parent_); + + // How many times to repeat the tests? We don't want to repeat them + // when we are inside the subprocess of a death test. + const int repeat = in_subprocess_for_death_test ? 1 : GTEST_FLAG(repeat); + // Repeats forever if the repeat count is negative. + const bool gtest_repeat_forever = repeat < 0; + for (int i = 0; gtest_repeat_forever || i != repeat; i++) { + // We want to preserve failures generated by ad-hoc test + // assertions executed before RUN_ALL_TESTS(). + ClearNonAdHocTestResult(); + + const TimeInMillis start = GetTimeInMillis(); + + // Shuffles test suites and tests if requested. + if (has_tests_to_run && GTEST_FLAG(shuffle)) { + random()->Reseed(static_cast(random_seed_)); + // This should be done before calling OnTestIterationStart(), + // such that a test event listener can see the actual test order + // in the event. + ShuffleTests(); + } + + // Tells the unit test event listeners that the tests are about to start. + repeater->OnTestIterationStart(*parent_, i); + + // Runs each test suite if there is at least one test to run. + if (has_tests_to_run) { + // Sets up all environments beforehand. + repeater->OnEnvironmentsSetUpStart(*parent_); + ForEach(environments_, SetUpEnvironment); + repeater->OnEnvironmentsSetUpEnd(*parent_); + + // Runs the tests only if there was no fatal failure or skip triggered + // during global set-up. + if (Test::IsSkipped()) { + // Emit diagnostics when global set-up calls skip, as it will not be + // emitted by default. + TestResult& test_result = + *internal::GetUnitTestImpl()->current_test_result(); + for (int j = 0; j < test_result.total_part_count(); ++j) { + const TestPartResult& test_part_result = + test_result.GetTestPartResult(j); + if (test_part_result.type() == TestPartResult::kSkip) { + const std::string& result = test_part_result.message(); + printf("%s\n", result.c_str()); + } + } + fflush(stdout); + } else if (!Test::HasFatalFailure()) { + for (int test_index = 0; test_index < total_test_suite_count(); + test_index++) { + GetMutableSuiteCase(test_index)->Run(); + } + } + + // Tears down all environments in reverse order afterwards. + repeater->OnEnvironmentsTearDownStart(*parent_); + std::for_each(environments_.rbegin(), environments_.rend(), + TearDownEnvironment); + repeater->OnEnvironmentsTearDownEnd(*parent_); + } + + elapsed_time_ = GetTimeInMillis() - start; + + // Tells the unit test event listener that the tests have just finished. + repeater->OnTestIterationEnd(*parent_, i); + + // Gets the result and clears it. + if (!Passed()) { + failed = true; + } + + // Restores the original test order after the iteration. This + // allows the user to quickly repro a failure that happens in the + // N-th iteration without repeating the first (N - 1) iterations. + // This is not enclosed in "if (GTEST_FLAG(shuffle)) { ... }", in + // case the user somehow changes the value of the flag somewhere + // (it's always safe to unshuffle the tests). + UnshuffleTests(); + + if (GTEST_FLAG(shuffle)) { + // Picks a new random seed for each iteration. + random_seed_ = GetNextRandomSeed(random_seed_); + } + } + + repeater->OnTestProgramEnd(*parent_); + + if (!gtest_is_initialized_before_run_all_tests) { + ColoredPrintf( + COLOR_RED, + "\nIMPORTANT NOTICE - DO NOT IGNORE:\n" + "This test program did NOT call " GTEST_INIT_GOOGLE_TEST_NAME_ + "() before calling RUN_ALL_TESTS(). This is INVALID. Soon " GTEST_NAME_ + " will start to enforce the valid usage. " + "Please fix it ASAP, or IT WILL START TO FAIL.\n"); // NOLINT +#if GTEST_FOR_GOOGLE_ + ColoredPrintf(COLOR_RED, + "For more details, see http://wiki/Main/ValidGUnitMain.\n"); +#endif // GTEST_FOR_GOOGLE_ + } + + return !failed; +} + +// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file +// if the variable is present. If a file already exists at this location, this +// function will write over it. If the variable is present, but the file cannot +// be created, prints an error and exits. +void WriteToShardStatusFileIfNeeded() { + const char* const test_shard_file = posix::GetEnv(kTestShardStatusFile); + if (test_shard_file != nullptr) { + FILE* const file = posix::FOpen(test_shard_file, "w"); + if (file == nullptr) { + ColoredPrintf(COLOR_RED, + "Could not write to the test shard status file \"%s\" " + "specified by the %s environment variable.\n", + test_shard_file, kTestShardStatusFile); + fflush(stdout); + exit(EXIT_FAILURE); + } + fclose(file); + } +} + +// Checks whether sharding is enabled by examining the relevant +// environment variable values. If the variables are present, +// but inconsistent (i.e., shard_index >= total_shards), prints +// an error and exits. If in_subprocess_for_death_test, sharding is +// disabled because it must only be applied to the original test +// process. Otherwise, we could filter out death tests we intended to execute. +bool ShouldShard(const char* total_shards_env, + const char* shard_index_env, + bool in_subprocess_for_death_test) { + if (in_subprocess_for_death_test) { + return false; + } + + const Int32 total_shards = Int32FromEnvOrDie(total_shards_env, -1); + const Int32 shard_index = Int32FromEnvOrDie(shard_index_env, -1); + + if (total_shards == -1 && shard_index == -1) { + return false; + } else if (total_shards == -1 && shard_index != -1) { + const Message msg = Message() + << "Invalid environment variables: you have " + << kTestShardIndex << " = " << shard_index + << ", but have left " << kTestTotalShards << " unset.\n"; + ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str()); + fflush(stdout); + exit(EXIT_FAILURE); + } else if (total_shards != -1 && shard_index == -1) { + const Message msg = Message() + << "Invalid environment variables: you have " + << kTestTotalShards << " = " << total_shards + << ", but have left " << kTestShardIndex << " unset.\n"; + ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str()); + fflush(stdout); + exit(EXIT_FAILURE); + } else if (shard_index < 0 || shard_index >= total_shards) { + const Message msg = Message() + << "Invalid environment variables: we require 0 <= " + << kTestShardIndex << " < " << kTestTotalShards + << ", but you have " << kTestShardIndex << "=" << shard_index + << ", " << kTestTotalShards << "=" << total_shards << ".\n"; + ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str()); + fflush(stdout); + exit(EXIT_FAILURE); + } + + return total_shards > 1; +} + +// Parses the environment variable var as an Int32. If it is unset, +// returns default_val. If it is not an Int32, prints an error +// and aborts. +Int32 Int32FromEnvOrDie(const char* var, Int32 default_val) { + const char* str_val = posix::GetEnv(var); + if (str_val == nullptr) { + return default_val; + } + + Int32 result; + if (!ParseInt32(Message() << "The value of environment variable " << var, + str_val, &result)) { + exit(EXIT_FAILURE); + } + return result; +} + +// Given the total number of shards, the shard index, and the test id, +// returns true iff the test should be run on this shard. The test id is +// some arbitrary but unique non-negative integer assigned to each test +// method. Assumes that 0 <= shard_index < total_shards. +bool ShouldRunTestOnShard(int total_shards, int shard_index, int test_id) { + return (test_id % total_shards) == shard_index; +} + +// Compares the name of each test with the user-specified filter to +// decide whether the test should be run, then records the result in +// each TestSuite and TestInfo object. +// If shard_tests == true, further filters tests based on sharding +// variables in the environment - see +// https://github.com/google/googletest/blob/master/googletest/docs/advanced.md +// . Returns the number of tests that should run. +int UnitTestImpl::FilterTests(ReactionToSharding shard_tests) { + const Int32 total_shards = shard_tests == HONOR_SHARDING_PROTOCOL ? + Int32FromEnvOrDie(kTestTotalShards, -1) : -1; + const Int32 shard_index = shard_tests == HONOR_SHARDING_PROTOCOL ? + Int32FromEnvOrDie(kTestShardIndex, -1) : -1; + + // num_runnable_tests are the number of tests that will + // run across all shards (i.e., match filter and are not disabled). + // num_selected_tests are the number of tests to be run on + // this shard. + int num_runnable_tests = 0; + int num_selected_tests = 0; + for (auto* test_suite : test_suites_) { + const std::string& test_suite_name = test_suite->name(); + test_suite->set_should_run(false); + + for (size_t j = 0; j < test_suite->test_info_list().size(); j++) { + TestInfo* const test_info = test_suite->test_info_list()[j]; + const std::string test_name(test_info->name()); + // A test is disabled if test suite name or test name matches + // kDisableTestFilter. + const bool is_disabled = internal::UnitTestOptions::MatchesFilter( + test_suite_name, kDisableTestFilter) || + internal::UnitTestOptions::MatchesFilter( + test_name, kDisableTestFilter); + test_info->is_disabled_ = is_disabled; + + const bool matches_filter = internal::UnitTestOptions::FilterMatchesTest( + test_suite_name, test_name); + test_info->matches_filter_ = matches_filter; + + const bool is_runnable = + (GTEST_FLAG(also_run_disabled_tests) || !is_disabled) && + matches_filter; + + const bool is_in_another_shard = + shard_tests != IGNORE_SHARDING_PROTOCOL && + !ShouldRunTestOnShard(total_shards, shard_index, num_runnable_tests); + test_info->is_in_another_shard_ = is_in_another_shard; + const bool is_selected = is_runnable && !is_in_another_shard; + + num_runnable_tests += is_runnable; + num_selected_tests += is_selected; + + test_info->should_run_ = is_selected; + test_suite->set_should_run(test_suite->should_run() || is_selected); + } + } + return num_selected_tests; +} + +// Prints the given C-string on a single line by replacing all '\n' +// characters with string "\\n". If the output takes more than +// max_length characters, only prints the first max_length characters +// and "...". +static void PrintOnOneLine(const char* str, int max_length) { + if (str != nullptr) { + for (int i = 0; *str != '\0'; ++str) { + if (i >= max_length) { + printf("..."); + break; + } + if (*str == '\n') { + printf("\\n"); + i += 2; + } else { + printf("%c", *str); + ++i; + } + } + } +} + +// Prints the names of the tests matching the user-specified filter flag. +void UnitTestImpl::ListTestsMatchingFilter() { + // Print at most this many characters for each type/value parameter. + const int kMaxParamLength = 250; + + for (auto* test_suite : test_suites_) { + bool printed_test_suite_name = false; + + for (size_t j = 0; j < test_suite->test_info_list().size(); j++) { + const TestInfo* const test_info = test_suite->test_info_list()[j]; + if (test_info->matches_filter_) { + if (!printed_test_suite_name) { + printed_test_suite_name = true; + printf("%s.", test_suite->name()); + if (test_suite->type_param() != nullptr) { + printf(" # %s = ", kTypeParamLabel); + // We print the type parameter on a single line to make + // the output easy to parse by a program. + PrintOnOneLine(test_suite->type_param(), kMaxParamLength); + } + printf("\n"); + } + printf(" %s", test_info->name()); + if (false && test_info->value_param() != nullptr) { + printf(" # %s = ", kValueParamLabel); + // We print the value parameter on a single line to make the + // output easy to parse by a program. + PrintOnOneLine(test_info->value_param(), kMaxParamLength); + } + printf("\n"); + } + } + } + fflush(stdout); + const std::string& output_format = UnitTestOptions::GetOutputFormat(); + if (output_format == "xml" || output_format == "json") { + FILE* fileout = OpenFileForWriting( + UnitTestOptions::GetAbsolutePathToOutputFile().c_str()); + std::stringstream stream; + if (output_format == "xml") { + XmlUnitTestResultPrinter( + UnitTestOptions::GetAbsolutePathToOutputFile().c_str()) + .PrintXmlTestsList(&stream, test_suites_); + } else if (output_format == "json") { + JsonUnitTestResultPrinter( + UnitTestOptions::GetAbsolutePathToOutputFile().c_str()) + .PrintJsonTestList(&stream, test_suites_); + } + fprintf(fileout, "%s", StringStreamToString(&stream).c_str()); + fclose(fileout); + } +} + +// Sets the OS stack trace getter. +// +// Does nothing if the input and the current OS stack trace getter are +// the same; otherwise, deletes the old getter and makes the input the +// current getter. +void UnitTestImpl::set_os_stack_trace_getter( + OsStackTraceGetterInterface* getter) { + if (os_stack_trace_getter_ != getter) { + delete os_stack_trace_getter_; + os_stack_trace_getter_ = getter; + } +} + +// Returns the current OS stack trace getter if it is not NULL; +// otherwise, creates an OsStackTraceGetter, makes it the current +// getter, and returns it. +OsStackTraceGetterInterface* UnitTestImpl::os_stack_trace_getter() { + if (os_stack_trace_getter_ == nullptr) { +#ifdef GTEST_OS_STACK_TRACE_GETTER_ + os_stack_trace_getter_ = new GTEST_OS_STACK_TRACE_GETTER_; +#else + os_stack_trace_getter_ = new OsStackTraceGetter; +#endif // GTEST_OS_STACK_TRACE_GETTER_ + } + + return os_stack_trace_getter_; +} + +// Returns the most specific TestResult currently running. +TestResult* UnitTestImpl::current_test_result() { + if (current_test_info_ != nullptr) { + return ¤t_test_info_->result_; + } + if (current_test_suite_ != nullptr) { + return ¤t_test_suite_->ad_hoc_test_result_; + } + return &ad_hoc_test_result_; +} + +// Shuffles all test suites, and the tests within each test suite, +// making sure that death tests are still run first. +void UnitTestImpl::ShuffleTests() { + // Shuffles the death test suites. + ShuffleRange(random(), 0, last_death_test_suite_ + 1, &test_suite_indices_); + + // Shuffles the non-death test suites. + ShuffleRange(random(), last_death_test_suite_ + 1, + static_cast(test_suites_.size()), &test_suite_indices_); + + // Shuffles the tests inside each test suite. + for (auto& test_suite : test_suites_) { + test_suite->ShuffleTests(random()); + } +} + +// Restores the test suites and tests to their order before the first shuffle. +void UnitTestImpl::UnshuffleTests() { + for (size_t i = 0; i < test_suites_.size(); i++) { + // Unshuffles the tests in each test suite. + test_suites_[i]->UnshuffleTests(); + // Resets the index of each test suite. + test_suite_indices_[i] = static_cast(i); + } +} + +// Returns the current OS stack trace as an std::string. +// +// The maximum number of stack frames to be included is specified by +// the gtest_stack_trace_depth flag. The skip_count parameter +// specifies the number of top frames to be skipped, which doesn't +// count against the number of frames to be included. +// +// For example, if Foo() calls Bar(), which in turn calls +// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in +// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't. +std::string GetCurrentOsStackTraceExceptTop(UnitTest* /*unit_test*/, + int skip_count) { + // We pass skip_count + 1 to skip this wrapper function in addition + // to what the user really wants to skip. + return GetUnitTestImpl()->CurrentOsStackTraceExceptTop(skip_count + 1); +} + +// Used by the GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_ macro to +// suppress unreachable code warnings. +namespace { +class ClassUniqueToAlwaysTrue {}; +} + +bool IsTrue(bool condition) { return condition; } + +bool AlwaysTrue() { +#if GTEST_HAS_EXCEPTIONS + // This condition is always false so AlwaysTrue() never actually throws, + // but it makes the compiler think that it may throw. + if (IsTrue(false)) + throw ClassUniqueToAlwaysTrue(); +#endif // GTEST_HAS_EXCEPTIONS + return true; +} + +// If *pstr starts with the given prefix, modifies *pstr to be right +// past the prefix and returns true; otherwise leaves *pstr unchanged +// and returns false. None of pstr, *pstr, and prefix can be NULL. +bool SkipPrefix(const char* prefix, const char** pstr) { + const size_t prefix_len = strlen(prefix); + if (strncmp(*pstr, prefix, prefix_len) == 0) { + *pstr += prefix_len; + return true; + } + return false; +} + +// Parses a string as a command line flag. The string should have +// the format "--flag=value". When def_optional is true, the "=value" +// part can be omitted. +// +// Returns the value of the flag, or NULL if the parsing failed. +static const char* ParseFlagValue(const char* str, const char* flag, + bool def_optional) { + // str and flag must not be NULL. + if (str == nullptr || flag == nullptr) return nullptr; + + // The flag must start with "--" followed by GTEST_FLAG_PREFIX_. + const std::string flag_str = std::string("--") + GTEST_FLAG_PREFIX_ + flag; + const size_t flag_len = flag_str.length(); + if (strncmp(str, flag_str.c_str(), flag_len) != 0) return nullptr; + + // Skips the flag name. + const char* flag_end = str + flag_len; + + // When def_optional is true, it's OK to not have a "=value" part. + if (def_optional && (flag_end[0] == '\0')) { + return flag_end; + } + + // If def_optional is true and there are more characters after the + // flag name, or if def_optional is false, there must be a '=' after + // the flag name. + if (flag_end[0] != '=') return nullptr; + + // Returns the string after "=". + return flag_end + 1; +} + +// Parses a string for a bool flag, in the form of either +// "--flag=value" or "--flag". +// +// In the former case, the value is taken as true as long as it does +// not start with '0', 'f', or 'F'. +// +// In the latter case, the value is taken as true. +// +// On success, stores the value of the flag in *value, and returns +// true. On failure, returns false without changing *value. +static bool ParseBoolFlag(const char* str, const char* flag, bool* value) { + // Gets the value of the flag as a string. + const char* const value_str = ParseFlagValue(str, flag, true); + + // Aborts if the parsing failed. + if (value_str == nullptr) return false; + + // Converts the string value to a bool. + *value = !(*value_str == '0' || *value_str == 'f' || *value_str == 'F'); + return true; +} + +// Parses a string for an Int32 flag, in the form of +// "--flag=value". +// +// On success, stores the value of the flag in *value, and returns +// true. On failure, returns false without changing *value. +bool ParseInt32Flag(const char* str, const char* flag, Int32* value) { + // Gets the value of the flag as a string. + const char* const value_str = ParseFlagValue(str, flag, false); + + // Aborts if the parsing failed. + if (value_str == nullptr) return false; + + // Sets *value to the value of the flag. + return ParseInt32(Message() << "The value of flag --" << flag, + value_str, value); +} + +// Parses a string for a string flag, in the form of +// "--flag=value". +// +// On success, stores the value of the flag in *value, and returns +// true. On failure, returns false without changing *value. +template +static bool ParseStringFlag(const char* str, const char* flag, String* value) { + // Gets the value of the flag as a string. + const char* const value_str = ParseFlagValue(str, flag, false); + + // Aborts if the parsing failed. + if (value_str == nullptr) return false; + + // Sets *value to the value of the flag. + *value = value_str; + return true; +} + +// Determines whether a string has a prefix that Google Test uses for its +// flags, i.e., starts with GTEST_FLAG_PREFIX_ or GTEST_FLAG_PREFIX_DASH_. +// If Google Test detects that a command line flag has its prefix but is not +// recognized, it will print its help message. Flags starting with +// GTEST_INTERNAL_PREFIX_ followed by "internal_" are considered Google Test +// internal flags and do not trigger the help message. +static bool HasGoogleTestFlagPrefix(const char* str) { + return (SkipPrefix("--", &str) || + SkipPrefix("-", &str) || + SkipPrefix("/", &str)) && + !SkipPrefix(GTEST_FLAG_PREFIX_ "internal_", &str) && + (SkipPrefix(GTEST_FLAG_PREFIX_, &str) || + SkipPrefix(GTEST_FLAG_PREFIX_DASH_, &str)); +} + +// Prints a string containing code-encoded text. The following escape +// sequences can be used in the string to control the text color: +// +// @@ prints a single '@' character. +// @R changes the color to red. +// @G changes the color to green. +// @Y changes the color to yellow. +// @D changes to the default terminal text color. +// +static void PrintColorEncoded(const char* str) { + GTestColor color = COLOR_DEFAULT; // The current color. + + // Conceptually, we split the string into segments divided by escape + // sequences. Then we print one segment at a time. At the end of + // each iteration, the str pointer advances to the beginning of the + // next segment. + for (;;) { + const char* p = strchr(str, '@'); + if (p == nullptr) { + ColoredPrintf(color, "%s", str); + return; + } + + ColoredPrintf(color, "%s", std::string(str, p).c_str()); + + const char ch = p[1]; + str = p + 2; + if (ch == '@') { + ColoredPrintf(color, "@"); + } else if (ch == 'D') { + color = COLOR_DEFAULT; + } else if (ch == 'R') { + color = COLOR_RED; + } else if (ch == 'G') { + color = COLOR_GREEN; + } else if (ch == 'Y') { + color = COLOR_YELLOW; + } else { + --str; + } + } +} + +static const char kColorEncodedHelpMessage[] = +"This program contains tests written using " GTEST_NAME_ ". You can use the\n" +"following command line flags to control its behavior:\n" +"\n" +"Test Selection:\n" +" @G--" GTEST_FLAG_PREFIX_ "list_tests@D\n" +" List the names of all tests instead of running them. The name of\n" +" TEST(Foo, Bar) is \"Foo.Bar\".\n" +" @G--" GTEST_FLAG_PREFIX_ "filter=@YPOSTIVE_PATTERNS" + "[@G-@YNEGATIVE_PATTERNS]@D\n" +" Run only the tests whose name matches one of the positive patterns but\n" +" none of the negative patterns. '?' matches any single character; '*'\n" +" matches any substring; ':' separates two patterns.\n" +" @G--" GTEST_FLAG_PREFIX_ "also_run_disabled_tests@D\n" +" Run all disabled tests too.\n" +"\n" +"Test Execution:\n" +" @G--" GTEST_FLAG_PREFIX_ "repeat=@Y[COUNT]@D\n" +" Run the tests repeatedly; use a negative count to repeat forever.\n" +" @G--" GTEST_FLAG_PREFIX_ "shuffle@D\n" +" Randomize tests' orders on every iteration.\n" +" @G--" GTEST_FLAG_PREFIX_ "random_seed=@Y[NUMBER]@D\n" +" Random number seed to use for shuffling test orders (between 1 and\n" +" 99999, or 0 to use a seed based on the current time).\n" +"\n" +"Test Output:\n" +" @G--" GTEST_FLAG_PREFIX_ "color=@Y(@Gyes@Y|@Gno@Y|@Gauto@Y)@D\n" +" Enable/disable colored output. The default is @Gauto@D.\n" +" -@G-" GTEST_FLAG_PREFIX_ "print_time=0@D\n" +" Don't print the elapsed time of each test.\n" +" @G--" GTEST_FLAG_PREFIX_ "output=@Y(@Gjson@Y|@Gxml@Y)[@G:@YDIRECTORY_PATH@G" + GTEST_PATH_SEP_ "@Y|@G:@YFILE_PATH]@D\n" +" Generate a JSON or XML report in the given directory or with the given\n" +" file name. @YFILE_PATH@D defaults to @Gtest_detail.xml@D.\n" +# if GTEST_CAN_STREAM_RESULTS_ +" @G--" GTEST_FLAG_PREFIX_ "stream_result_to=@YHOST@G:@YPORT@D\n" +" Stream test results to the given server.\n" +# endif // GTEST_CAN_STREAM_RESULTS_ +"\n" +"Assertion Behavior:\n" +# if GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS +" @G--" GTEST_FLAG_PREFIX_ "death_test_style=@Y(@Gfast@Y|@Gthreadsafe@Y)@D\n" +" Set the default death test style.\n" +# endif // GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS +" @G--" GTEST_FLAG_PREFIX_ "break_on_failure@D\n" +" Turn assertion failures into debugger break-points.\n" +" @G--" GTEST_FLAG_PREFIX_ "throw_on_failure@D\n" +" Turn assertion failures into C++ exceptions for use by an external\n" +" test framework.\n" +" @G--" GTEST_FLAG_PREFIX_ "catch_exceptions=0@D\n" +" Do not report exceptions as test failures. Instead, allow them\n" +" to crash the program or throw a pop-up (on Windows).\n" +"\n" +"Except for @G--" GTEST_FLAG_PREFIX_ "list_tests@D, you can alternatively set " + "the corresponding\n" +"environment variable of a flag (all letters in upper-case). For example, to\n" +"disable colored text output, you can either specify @G--" GTEST_FLAG_PREFIX_ + "color=no@D or set\n" +"the @G" GTEST_FLAG_PREFIX_UPPER_ "COLOR@D environment variable to @Gno@D.\n" +"\n" +"For more information, please read the " GTEST_NAME_ " documentation at\n" +"@G" GTEST_PROJECT_URL_ "@D. If you find a bug in " GTEST_NAME_ "\n" +"(not one in your own code or tests), please report it to\n" +"@G<" GTEST_DEV_EMAIL_ ">@D.\n"; + +static bool ParseGoogleTestFlag(const char* const arg) { + return ParseBoolFlag(arg, kAlsoRunDisabledTestsFlag, + >EST_FLAG(also_run_disabled_tests)) || + ParseBoolFlag(arg, kBreakOnFailureFlag, + >EST_FLAG(break_on_failure)) || + ParseBoolFlag(arg, kCatchExceptionsFlag, + >EST_FLAG(catch_exceptions)) || + ParseStringFlag(arg, kColorFlag, >EST_FLAG(color)) || + ParseStringFlag(arg, kDeathTestStyleFlag, + >EST_FLAG(death_test_style)) || + ParseBoolFlag(arg, kDeathTestUseFork, + >EST_FLAG(death_test_use_fork)) || + ParseStringFlag(arg, kFilterFlag, >EST_FLAG(filter)) || + ParseStringFlag(arg, kInternalRunDeathTestFlag, + >EST_FLAG(internal_run_death_test)) || + ParseBoolFlag(arg, kListTestsFlag, >EST_FLAG(list_tests)) || + ParseStringFlag(arg, kOutputFlag, >EST_FLAG(output)) || + ParseBoolFlag(arg, kPrintTimeFlag, >EST_FLAG(print_time)) || + ParseBoolFlag(arg, kPrintUTF8Flag, >EST_FLAG(print_utf8)) || + ParseInt32Flag(arg, kRandomSeedFlag, >EST_FLAG(random_seed)) || + ParseInt32Flag(arg, kRepeatFlag, >EST_FLAG(repeat)) || + ParseBoolFlag(arg, kShuffleFlag, >EST_FLAG(shuffle)) || + ParseInt32Flag(arg, kStackTraceDepthFlag, + >EST_FLAG(stack_trace_depth)) || + ParseStringFlag(arg, kStreamResultToFlag, + >EST_FLAG(stream_result_to)) || + ParseBoolFlag(arg, kThrowOnFailureFlag, + >EST_FLAG(throw_on_failure)); +} + +#if GTEST_USE_OWN_FLAGFILE_FLAG_ +static void LoadFlagsFromFile(const std::string& path) { + FILE* flagfile = posix::FOpen(path.c_str(), "r"); + if (!flagfile) { + GTEST_LOG_(FATAL) << "Unable to open file \"" << GTEST_FLAG(flagfile) + << "\""; + } + std::string contents(ReadEntireFile(flagfile)); + posix::FClose(flagfile); + std::vector lines; + SplitString(contents, '\n', &lines); + for (size_t i = 0; i < lines.size(); ++i) { + if (lines[i].empty()) + continue; + if (!ParseGoogleTestFlag(lines[i].c_str())) + g_help_flag = true; + } +} +#endif // GTEST_USE_OWN_FLAGFILE_FLAG_ + +// Parses the command line for Google Test flags, without initializing +// other parts of Google Test. The type parameter CharType can be +// instantiated to either char or wchar_t. +template +void ParseGoogleTestFlagsOnlyImpl(int* argc, CharType** argv) { + for (int i = 1; i < *argc; i++) { + const std::string arg_string = StreamableToString(argv[i]); + const char* const arg = arg_string.c_str(); + + using internal::ParseBoolFlag; + using internal::ParseInt32Flag; + using internal::ParseStringFlag; + + bool remove_flag = false; + if (ParseGoogleTestFlag(arg)) { + remove_flag = true; +#if GTEST_USE_OWN_FLAGFILE_FLAG_ + } else if (ParseStringFlag(arg, kFlagfileFlag, >EST_FLAG(flagfile))) { + LoadFlagsFromFile(GTEST_FLAG(flagfile)); + remove_flag = true; +#endif // GTEST_USE_OWN_FLAGFILE_FLAG_ + } else if (arg_string == "--help" || arg_string == "-h" || + arg_string == "-?" || arg_string == "/?" || + HasGoogleTestFlagPrefix(arg)) { + // Both help flag and unrecognized Google Test flags (excluding + // internal ones) trigger help display. + g_help_flag = true; + } + + if (remove_flag) { + // Shift the remainder of the argv list left by one. Note + // that argv has (*argc + 1) elements, the last one always being + // NULL. The following loop moves the trailing NULL element as + // well. + for (int j = i; j != *argc; j++) { + argv[j] = argv[j + 1]; + } + + // Decrements the argument count. + (*argc)--; + + // We also need to decrement the iterator as we just removed + // an element. + i--; + } + } + + if (g_help_flag) { + // We print the help here instead of in RUN_ALL_TESTS(), as the + // latter may not be called at all if the user is using Google + // Test with another testing framework. + PrintColorEncoded(kColorEncodedHelpMessage); + } +} + +// Parses the command line for Google Test flags, without initializing +// other parts of Google Test. +void ParseGoogleTestFlagsOnly(int* argc, char** argv) { + ParseGoogleTestFlagsOnlyImpl(argc, argv); + + // Fix the value of *_NSGetArgc() on macOS, but iff + // *_NSGetArgv() == argv + // Only applicable to char** version of argv +#if GTEST_OS_MAC +#ifndef GTEST_OS_IOS + if (*_NSGetArgv() == argv) { + *_NSGetArgc() = *argc; + } +#endif +#endif +} +void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv) { + ParseGoogleTestFlagsOnlyImpl(argc, argv); +} + +// The internal implementation of InitGoogleTest(). +// +// The type parameter CharType can be instantiated to either char or +// wchar_t. +template +void InitGoogleTestImpl(int* argc, CharType** argv) { + // We don't want to run the initialization code twice. + if (GTestIsInitialized()) return; + + if (*argc <= 0) return; + + g_argvs.clear(); + for (int i = 0; i != *argc; i++) { + g_argvs.push_back(StreamableToString(argv[i])); + } + +#if GTEST_HAS_ABSL + absl::InitializeSymbolizer(g_argvs[0].c_str()); +#endif // GTEST_HAS_ABSL + + ParseGoogleTestFlagsOnly(argc, argv); + GetUnitTestImpl()->PostFlagParsingInit(); +} + +} // namespace internal + +// Initializes Google Test. This must be called before calling +// RUN_ALL_TESTS(). In particular, it parses a command line for the +// flags that Google Test recognizes. Whenever a Google Test flag is +// seen, it is removed from argv, and *argc is decremented. +// +// No value is returned. Instead, the Google Test flag variables are +// updated. +// +// Calling the function for the second time has no user-visible effect. +void InitGoogleTest(int* argc, char** argv) { +#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) + GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(argc, argv); +#else // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) + internal::InitGoogleTestImpl(argc, argv); +#endif // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) +} + +// This overloaded version can be used in Windows programs compiled in +// UNICODE mode. +void InitGoogleTest(int* argc, wchar_t** argv) { +#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) + GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(argc, argv); +#else // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) + internal::InitGoogleTestImpl(argc, argv); +#endif // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) +} + +// This overloaded version can be used on Arduino/embedded platforms where +// there is no argc/argv. +void InitGoogleTest() { + // Since Arduino doesn't have a command line, fake out the argc/argv arguments + int argc = 1; + const auto arg0 = "dummy"; + char* argv0 = const_cast(arg0); + char** argv = &argv0; + +#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) + GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(&argc, argv); +#else // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) + internal::InitGoogleTestImpl(&argc, argv); +#endif // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_) +} + +std::string TempDir() { +#if defined(GTEST_CUSTOM_TEMPDIR_FUNCTION_) + return GTEST_CUSTOM_TEMPDIR_FUNCTION_(); +#endif + +#if GTEST_OS_WINDOWS_MOBILE + return "\\temp\\"; +#elif GTEST_OS_WINDOWS + const char* temp_dir = internal::posix::GetEnv("TEMP"); + if (temp_dir == nullptr || temp_dir[0] == '\0') + return "\\temp\\"; + else if (temp_dir[strlen(temp_dir) - 1] == '\\') + return temp_dir; + else + return std::string(temp_dir) + "\\"; +#elif GTEST_OS_LINUX_ANDROID + return "/sdcard/"; +#else + return "/tmp/"; +#endif // GTEST_OS_WINDOWS_MOBILE +} + +// Class ScopedTrace + +// Pushes the given source file location and message onto a per-thread +// trace stack maintained by Google Test. +void ScopedTrace::PushTrace(const char* file, int line, std::string message) { + internal::TraceInfo trace; + trace.file = file; + trace.line = line; + trace.message.swap(message); + + UnitTest::GetInstance()->PushGTestTrace(trace); +} + +// Pops the info pushed by the c'tor. +ScopedTrace::~ScopedTrace() + GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) { + UnitTest::GetInstance()->PopGTestTrace(); +} + +} // namespace testing diff --git a/deps/googletest/src/gtest_main.cc b/deps/googletest/src/gtest_main.cc new file mode 100644 index 000000000..f6e1dd96f --- /dev/null +++ b/deps/googletest/src/gtest_main.cc @@ -0,0 +1,47 @@ +// Copyright 2006, Google Inc. +// All rights reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include +#include "gtest/gtest.h" + +#ifdef ARDUINO +void setup() { + testing::InitGoogleTest(); +} + +void loop() { RUN_ALL_TESTS(); } + +#else + +GTEST_API_ int main(int argc, char **argv) { + printf("Running main() from %s\n", __FILE__); + testing::InitGoogleTest(&argc, argv); + return RUN_ALL_TESTS(); +} +#endif diff --git a/docs/CMakeLists.txt b/docs/CMakeLists.txt new file mode 100644 index 000000000..28c409ab2 --- /dev/null +++ b/docs/CMakeLists.txt @@ -0,0 +1,49 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +find_package(Sphinx REQUIRED) + +# configured documentation tools and intermediate build results +set(BINARY_BUILD_DIR "${CMAKE_CURRENT_BINARY_DIR}/_build") + +# Sphinx cache with pickled ReST documents +set(SPHINX_CACHE_DIR "${CMAKE_CURRENT_BINARY_DIR}/_doctrees") + +# HTML output directory +set(SPHINX_HTML_DIR "${CMAKE_CURRENT_BINARY_DIR}/Documentation/html") + +configure_file( + "${CMAKE_CURRENT_SOURCE_DIR}/conf.py.in" + "${BINARY_BUILD_DIR}/conf.py" + @ONLY) + +add_custom_target(Documentation ALL + ${SPHINX_EXECUTABLE} + -Q -b html + -c "${BINARY_BUILD_DIR}" + -d "${SPHINX_CACHE_DIR}" + "${CMAKE_CURRENT_SOURCE_DIR}" + "${SPHINX_HTML_DIR}" + COMMENT "Building HTML documentation with Sphinx") + +install(DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/Documentation/html + DESTINATION "docs" + PATTERN "_static" EXCLUDE + PATTERN "_sources" EXCLUDE +) diff --git a/docs/README.md b/docs/README.md new file mode 100644 index 000000000..454d442ec --- /dev/null +++ b/docs/README.md @@ -0,0 +1,8 @@ +### oneMKL documentation + +This folder contains oneMKL documentation in reStructuredText (rST) format. + +The documentation build step is skipped by default. +To enable building documentation: + +- Set `-DBUILD_DOC=ON` when building with cmake (for more information see [Build with CMake directly](../README.md#building-with-cmake)) diff --git a/docs/conf.py b/docs/conf.py new file mode 100644 index 000000000..5b12eb498 --- /dev/null +++ b/docs/conf.py @@ -0,0 +1,190 @@ +# -*- coding: utf-8 -*- +# +# Configuration file for the Sphinx documentation builder. +# +# This file does only contain a selection of the most common options. For a +# full list see the documentation: +# http://www.sphinx-doc.org/en/master/config + +# -- Path setup -------------------------------------------------------------- + +# If extensions (or modules to document with autodoc) are in another directory, +# add these directories to sys.path here. If the directory is relative to the +# documentation root, use os.path.abspath to make it absolute, like shown here. +# +# import os +# import sys +# sys.path.insert(0, os.path.abspath('.')) + + +# -- Project information ----------------------------------------------------- + +project = 'oneAPI Math Kernel Library Interfaces' +copyright = '2020, Intel Corporation' +author = 'Intel Corporation' + +# The short X.Y version +version = '' +# The full version, including alpha/beta/rc tags +release = '0.1' + + +# -- General configuration --------------------------------------------------- + +# If your documentation needs a minimal Sphinx version, state it here. +# +# needs_sphinx = '1.0' + +# Add any Sphinx extension module names here, as strings. They can be +# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom +# ones. +extensions = [ + 'sphinx.ext.autodoc', + 'sphinx.ext.doctest', + 'sphinx.ext.intersphinx', + 'sphinx.ext.todo', +] + +# Add any paths that contain templates here, relative to this directory. +templates_path = ['_templates'] + +# The suffix(es) of source filenames. +# You can specify multiple suffix as a list of string: +# +# source_suffix = ['.rst', '.md'] +source_suffix = '.rst' + +# The master toctree document. +master_doc = 'index' + +# The language for content autogenerated by Sphinx. Refer to documentation +# for a list of supported languages. +# +# This is also used if you do content translation via gettext catalogs. +# Usually you set "language" from the command line for these cases. +language = None + +# List of patterns, relative to source directory, that match files and +# directories to ignore when looking for source files. +# This pattern also affects html_static_path and html_extra_path. +exclude_patterns = [] + +# The name of the Pygments (syntax highlighting) style to use. +pygments_style = None + + +# -- Options for HTML output ------------------------------------------------- + +# The theme to use for HTML and HTML Help pages. See the documentation for +# a list of builtin themes. +# +html_theme = 'sphinx_rtd_theme' + +# Theme options are theme-specific and customize the look and feel of a theme +# further. For a list of options available for each theme, see the +# documentation. +# +# html_theme_options = {} + +# Add any paths that contain custom static files (such as style sheets) here, +# relative to this directory. They are copied after the builtin static files, +# so a file named "default.css" will overwrite the builtin "default.css". +html_static_path = ['_static'] + +# Custom sidebar templates, must be a dictionary that maps document names +# to template names. +# +# The default sidebars (for documents that don't match any pattern) are +# defined by theme itself. Builtin themes are using these templates by +# default: ``['localtoc.html', 'relations.html', 'sourcelink.html', +# 'searchbox.html']``. +# +# html_sidebars = {} + + +# -- Options for HTMLHelp output --------------------------------------------- + +# Output file base name for HTML help builder. +htmlhelp_basename = 'reSTTemplatedoc' + + +# -- Options for LaTeX output ------------------------------------------------ + +latex_elements = { + # The paper size ('letterpaper' or 'a4paper'). + # + # 'papersize': 'letterpaper', + + # The font size ('10pt', '11pt' or '12pt'). + # + # 'pointsize': '10pt', + + # Additional stuff for the LaTeX preamble. + # + # 'preamble': '', + + # Latex figure (float) alignment + # + # 'figure_align': 'htbp', +} + +# Grouping the document tree into LaTeX files. List of tuples +# (source start file, target name, title, +# author, documentclass [howto, manual, or own class]). +latex_documents = [ + (master_doc, 'reSTTemplate.tex', 'reST Template Documentation', + 'Ben Fitch', 'manual'), +] + + +# -- Options for manual page output ------------------------------------------ + +# One entry per manual page. List of tuples +# (source start file, name, description, authors, manual section). +man_pages = [ + (master_doc, 'resttemplate', 'reST Template Documentation', + [author], 1) +] + + +# -- Options for Texinfo output ---------------------------------------------- + +# Grouping the document tree into Texinfo files. List of tuples +# (source start file, target name, title, author, +# dir menu entry, description, category) +texinfo_documents = [ + (master_doc, 'reSTTemplate', 'reST Template Documentation', + author, 'reSTTemplate', 'One line description of project.', + 'Miscellaneous'), +] + + +# -- Options for Epub output ------------------------------------------------- + +# Bibliographic Dublin Core info. +epub_title = project + +# The unique identifier of the text. This can be a ISBN number +# or the project homepage. +# +# epub_identifier = '' + +# A unique identification for the text. +# +# epub_uid = '' + +# A list of files that should not be packed into the epub file. +epub_exclude_files = ['search.html'] + + +# -- Extension configuration ------------------------------------------------- + +# -- Options for intersphinx extension --------------------------------------- + +# Example configuration for intersphinx: refer to the Python standard library. +intersphinx_mapping = {'https://docs.python.org/': None} + +# -- Options for todo extension ---------------------------------------------- + +# If true, `todo` and `todoList` produce output, else they produce nothing. +todo_include_todos = True diff --git a/docs/conf.py.in b/docs/conf.py.in new file mode 100644 index 000000000..793c9b87c --- /dev/null +++ b/docs/conf.py.in @@ -0,0 +1,190 @@ +# -*- coding: utf-8 -*- +# +# Configuration file for the Sphinx documentation builder. +# +# This file does only contain a selection of the most common options. For a +# full list see the documentation: +# http://www.sphinx-doc.org/en/master/config + +# -- Path setup -------------------------------------------------------------- + +# If extensions (or modules to document with autodoc) are in another directory, +# add these directories to sys.path here. If the directory is relative to the +# documentation root, use os.path.abspath to make it absolute, like shown here. +# +# import os +# import sys +# sys.path.insert(0, os.path.abspath('.')) + + +# -- Project information ----------------------------------------------------- + +project = 'oneAPI Math Kernel Library Interfaces' +copyright = '2020, Intel Corporation' +author = 'Intel Corporation' + +# The short X.Y version +version = '' +# The full version, including alpha/beta/rc tags +release = '0.1' + + +# -- General configuration --------------------------------------------------- + +# If your documentation needs a minimal Sphinx version, state it here. +# +# needs_sphinx = '1.0' + +# Add any Sphinx extension module names here, as strings. They can be +# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom +# ones. +extensions = [ + 'sphinx.ext.autodoc', + 'sphinx.ext.doctest', + 'sphinx.ext.intersphinx', + 'sphinx.ext.todo', +] + +# Add any paths that contain templates here, relative to this directory. +templates_path = ['_templates'] + +# The suffix(es) of source filenames. +# You can specify multiple suffix as a list of string: +# +# source_suffix = ['.rst', '.md'] +source_suffix = '.rst' + +# The master toctree document. +master_doc = 'index' + +# The language for content autogenerated by Sphinx. Refer to documentation +# for a list of supported languages. +# +# This is also used if you do content translation via gettext catalogs. +# Usually you set "language" from the command line for these cases. +language = None + +# List of patterns, relative to source directory, that match files and +# directories to ignore when looking for source files. +# This pattern also affects html_static_path and html_extra_path. +exclude_patterns = [] + +# The name of the Pygments (syntax highlighting) style to use. +pygments_style = None + + +# -- Options for HTML output ------------------------------------------------- + +# The theme to use for HTML and HTML Help pages. See the documentation for +# a list of builtin themes. +# +html_theme = 'alabaster' + +# Theme options are theme-specific and customize the look and feel of a theme +# further. For a list of options available for each theme, see the +# documentation. +# +# html_theme_options = {} + +# Add any paths that contain custom static files (such as style sheets) here, +# relative to this directory. They are copied after the builtin static files, +# so a file named "default.css" will overwrite the builtin "default.css". +html_static_path = ['_static'] + +# Custom sidebar templates, must be a dictionary that maps document names +# to template names. +# +# The default sidebars (for documents that don't match any pattern) are +# defined by theme itself. Builtin themes are using these templates by +# default: ``['localtoc.html', 'relations.html', 'sourcelink.html', +# 'searchbox.html']``. +# +# html_sidebars = {} + + +# -- Options for HTMLHelp output --------------------------------------------- + +# Output file base name for HTML help builder. +htmlhelp_basename = 'reSTTemplatedoc' + + +# -- Options for LaTeX output ------------------------------------------------ + +latex_elements = { + # The paper size ('letterpaper' or 'a4paper'). + # + # 'papersize': 'letterpaper', + + # The font size ('10pt', '11pt' or '12pt'). + # + # 'pointsize': '10pt', + + # Additional stuff for the LaTeX preamble. + # + # 'preamble': '', + + # Latex figure (float) alignment + # + # 'figure_align': 'htbp', +} + +# Grouping the document tree into LaTeX files. List of tuples +# (source start file, target name, title, +# author, documentclass [howto, manual, or own class]). +latex_documents = [ + (master_doc, 'reSTTemplate.tex', 'reST Template Documentation', + 'Ben Fitch', 'manual'), +] + + +# -- Options for manual page output ------------------------------------------ + +# One entry per manual page. List of tuples +# (source start file, name, description, authors, manual section). +man_pages = [ + (master_doc, 'resttemplate', 'reST Template Documentation', + [author], 1) +] + + +# -- Options for Texinfo output ---------------------------------------------- + +# Grouping the document tree into Texinfo files. List of tuples +# (source start file, target name, title, author, +# dir menu entry, description, category) +texinfo_documents = [ + (master_doc, 'reSTTemplate', 'reST Template Documentation', + author, 'reSTTemplate', 'One line description of project.', + 'Miscellaneous'), +] + + +# -- Options for Epub output ------------------------------------------------- + +# Bibliographic Dublin Core info. +epub_title = project + +# The unique identifier of the text. This can be a ISBN number +# or the project homepage. +# +# epub_identifier = '' + +# A unique identification for the text. +# +# epub_uid = '' + +# A list of files that should not be packed into the epub file. +epub_exclude_files = ['search.html'] + + +# -- Extension configuration ------------------------------------------------- + +# -- Options for intersphinx extension --------------------------------------- + +# Example configuration for intersphinx: refer to the Python standard library. +intersphinx_mapping = {'https://docs.python.org/': None} + +# -- Options for todo extension ---------------------------------------------- + +# If true, `todo` and `todoList` produce output, else they produce nothing. +todo_include_todos = True diff --git a/docs/domains/blas/asum.rst b/docs/domains/blas/asum.rst new file mode 100644 index 000000000..8e7f53092 --- /dev/null +++ b/docs/domains/blas/asum.rst @@ -0,0 +1,121 @@ +.. _asum: + +asum +==== + + +.. container:: + + + Computes the sum of magnitudes of the vector elements. + + + .. container:: section + :name: GUID-C135E117-8018-473E-BE83-8833C95BB3B5 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void asum(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &result) + + ``asum`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + - T_res + * - ``float`` + - ``float`` + * - ``double`` + - ``double`` + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-6AFCECB5-6614-46AC-B921-AB5DED0D22B2 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The asum routine computes the sum of the magnitudes of elements of a + real vector, or the sum of magnitudes of the real and imaginary parts + of elements of a complex vector: + + + |image0| + + + where ``x`` is a vector with ``n`` elements. + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + +.. container:: section + :name: GUID-2B160DEB-ADBB-4044-8078-4B613A0DA4E1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + Buffer where the scalar result is stored (the sum of magnitudes of + the real and imaginary parts of all elements of the vector). + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-684BB993-83CA-4605-BD49-E493806C1ee1.png + :class: img-middle + diff --git a/docs/domains/blas/axpy.rst b/docs/domains/blas/axpy.rst new file mode 100644 index 000000000..50f2ed986 --- /dev/null +++ b/docs/domains/blas/axpy.rst @@ -0,0 +1,135 @@ +.. _axpy: + +axpy +==== + + +.. container:: + + + Computes a vector-scalar product and adds the result to a vector. + + + .. container:: section + :name: GUID-17ADB23B-C9B0-44B4-89F9-B7199DA9E872 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void axpy(queue &exec_queue, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy) + + ``axpy`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-4BC6BF9A-BAB9-4078-A6B5-9C7ECB9D4821 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The axpy routines compute a scalar-vector product and add the result + to a vector: + + + + + + y <- alpha*x+y + + + where: + + + ``x`` and ``y`` are vectors of ``n`` elements, + + + ``alpha`` is a scalar. + + +.. container:: section + :name: GUID-6F86EF6A-8FFE-4C6A-8B71-23B95C1F1365 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector x. + + + alpha + Specifies the scalar alpha. + + + x + Buffer holding input vector x. The buffer must be of size at least + ``(1 + (n – 1)*abs(incx))``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector y. The buffer must be of size at least + ``(1 + (n – 1)*abs(incy))``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + +.. container:: section + :name: GUID-A0926D96-B673-48A4-986A-033719589288 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector y. + + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/blas-level-1-routines.inc.rst b/docs/domains/blas/blas-level-1-routines.inc.rst new file mode 100644 index 000000000..7798d13a4 --- /dev/null +++ b/docs/domains/blas/blas-level-1-routines.inc.rst @@ -0,0 +1,92 @@ +.. _blas-level-1-routines: + +BLAS Level 1 Routines +===================== + + +.. container:: + + + BLAS Level 1 includes routines and functions, which perform + vector-vector operations. The following table lists the BLAS Level 1 + routine and function groups and the data types associated with them. + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Routine or Function Group with SYCL Buffer + - Data Types + - Description + * - \ `asum `__\ + - float, double, mixed float and std::complex, mixed double and std::complex + - Sum of vector magnitudes (functions) + * - \ `axpy `__\ + - float, double, std::complex, std::complex + - Scalar-vector product (routines) + * - \ `copy `__\ + - float, double, std::complex, std::complex + - Copy vector (routines) + * - \ `dot `__\ + - float, double, mixed float and double + - Dot product (functions) + * - \ `sdsdot `__\ + - mixed float and double + - Dot product with double precision (functions) + * - \ `dotc `__\ + - std::complex, std::complex + - Dot product conjugated (functions) + * - \ `dotu `__\ + - std::complex, std::complex + - Dot product unconjugated (functions) + * - \ `nrm2 `__\ + - float, double, mixed float and std::complex, mixed double and std::complex + - Vector 2-norm (Euclidean norm) (functions) + * - \ `rot `__\ + - float, double, mixed float and std::complex, mixed double and std::complex + - Plane rotation of points (routines) + * - \ `rotg `__\ + - float, double, std::complex, std::complex + - Generate Givens rotation of points (routines) + * - \ `rotm `__\ + - float, double + - Modified Givens plane rotation of points (routines) + * - \ `rotmg `__\ + - float, double + - Generate modified Givens plane rotation of points (routines) + * - \ `scal `__\ + - float, double, std::complex, std::complex, mixed float and std::complex, mixed double and std::complex + - Vector-scalar product (routines) + * - \ `swap `__\ + - float, double, std::complex, std::complex + - Vector-vector swap (routines) + * - \ `iamax `__\ + - float, double, std::complex, std::complex + - Index of the maximum absolute value element of a vector (functions) + * - \ `iamin `__\ + - float, double, std::complex, std::complex + - Index of the minimum absolute value element of a vector (functions) + +.. toctree:: + :hidden: + + asum + axpy + copy + dot + dotc + dotu + iamax + iamin + nrm2 + rot + rotg + rotm + rotmg + scal + sdsdot + swap + diff --git a/docs/domains/blas/blas-level-2-routines.inc.rst b/docs/domains/blas/blas-level-2-routines.inc.rst new file mode 100644 index 000000000..dbe97bbae --- /dev/null +++ b/docs/domains/blas/blas-level-2-routines.inc.rst @@ -0,0 +1,130 @@ +.. _blas-level-2-routines: + +BLAS Level 2 Routines +===================== + + +.. container:: + + + This section describes BLAS Level 2 routines, which perform + matrix-vector operations. The following table lists the BLAS Level 2 + routine groups and the data types associated with them. + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Routine or Function Group with SYCL Buffer + - Data Types + - Description + * - \ `gbmv `__\ + - float, double, std::complex, std::complex + - Matrix-vector product using a general band matrix + * - \ `gemv `__\ + - float, double, std::complex, std::complex + - Matrix-vector product using a general matrix + * - \ `ger `__\ + - float, double + - Rank-1 update of a general matrix + * - \ `gerc `__\ + - std::complex, std::complex + - Rank-1 update of a conjugated general matrix + * - \ `geru `__\ + - std::complex, std::complex + - Rank-1 update of a general matrix, unconjugated + * - \ `hbmv `__\ + - std::complex, std::complex + - Matrix-vector product using a Hermitian band matrix + * - \ `hemv `__\ + - std::complex, std::complex + - Matrix-vector product using a Hermitian matrix + * - \ `her `__\ + - std::complex, std::complex + - Rank-1 update of a Hermitian matrix + * - \ `her2 `__\ + - std::complex, std::complex + - Rank-2 update of a Hermitian matrix + * - \ `hpmv `__\ + - std::complex, std::complex + - Matrix-vector product using a Hermitian packed matrix + * - \ `hpr `__\ + - std::complex, std::complex + - Rank-1 update of a Hermitian packed matrix + * - \ `hpr2 `__\ + - std::complex, std::complex + - Rank-2 update of a Hermitian packed matrix + * - \ `sbmv `__\ + - float, double + - Matrix-vector product using symmetric band matrix + * - \ `spmv `__\ + - float, double + - Matrix-vector product using a symmetric packed matrix + * - \ `spr `__\ + - float, double + - Rank-1 update of a symmetric packed matrix + * - \ `spr2 `__\ + - float, double + - Rank-2 update of a symmetric packed matrix + * - \ `symv `__\ + - float, double + - Matrix-vector product using a symmetric matrix + * - \ `syr `__\ + - float, double + - Rank-1 update of a symmetric matrix + * - \ `syr2 `__\ + - float, double + - Rank-2 update of a symmetric matrix + * - \ `tbmv `__\ + - float, double, std::complex, std::complex + - Matrix-vector product using a triangular band matrix + * - \ `tbsv `__\ + - float, double, std::complex, std::complex + - Solution of a linear system of equations with a triangular band matrix + * - \ `tpmv `__\ + - float, double, std::complex, std::complex + - Matrix-vector product using a triangular packed matrix + * - \ `tpsv `__\ + - float, double, std::complex, std::complex + - Solution of a linear system of equations with a triangular packed matrix + * - \ `trmv `__\ + - float, double, std::complex, std::complex + - Matrix-vector product using a triangular matrix + * - \ `trsv `__\ + - float, double, std::complex, std::complex + - Solution of a linear system of equations with a triangular matrix + + + + +.. toctree:: + :hidden: + + gbmv + gemv + ger + gerc + geru + hbmv + hemv + her + her2 + hpmv + hpr + hpr2 + sbmv + spmv + spr + spr2 + symv + syr + syr2 + tbmv + tbsv + tpmv + tpsv + trmv + trsv diff --git a/docs/domains/blas/blas-level-3-routines.inc.rst b/docs/domains/blas/blas-level-3-routines.inc.rst new file mode 100644 index 000000000..a80c18bfc --- /dev/null +++ b/docs/domains/blas/blas-level-3-routines.inc.rst @@ -0,0 +1,90 @@ +.. _blas-level-3-routines: + +BLAS Level 3 Routines +===================== + + +.. container:: + + + BLAS Level 3 routines perform matrix-matrix operations. The following + table lists the BLAS Level 3 routine groups and the data types + associated with them. + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Routine or Function Group with SYCL Buffer + - Data Types + - Description + * - \ `gemm `__\ + - float, double, std::complex, std::complex + - Computes a matrix-matrix product with general matrices. + * - \ `hemm `__\ + - std::complex, std::complex + - Computes a matrix-matrix product where one input matrix is Hermitian and one is general. + * - \ `herk `__\ + - std::complex, std::complex + - Performs a Hermitian rank-k update. + * - \ `her2k `__\ + - std::complex, std::complex + - Performs a Hermitian rank-2k update. + * - \ `symm `__\ + - float, double, std::complex, std::complex + - Computes a matrix-matrix product where one input matrix is symmetric and one matrix is general. + * - \ `syrk `__\ + - float, double, std::complex, std::complex + - Performs a symmetric rank-k update. + * - \ `syr2k `__\ + - float, double, std::complex, std::complex + - Performs a symmetric rank-2k update. + * - \ `trmm `__\ + - float, double, std::complex, std::complex + - Computes a matrix-matrix product where one input matrix is triangular and one input matrix is general. + * - \ `trsm `__\ + - float, double, std::complex, std::complex + - Solves a triangular matrix equation (forward or backward solve). + + + + + - + + + .. container:: + :name: LI_21BA86AC0A4942A79BA0C7DC4ABC50C4 + + + The BLAS functions are blocked where possible to restructure + the code in a way that increases the localization of data + reference, enhances cache memory use, and reduces the + dependency on the memory bus. + + + - + + + .. container:: + :name: LI_9D82DEDFA672416D9B3EA8C9C2B6F0A3 + + + The code is distributed across the processors to maximize + parallelism. + + +.. toctree:: + :hidden: + + gemm + hemm + her2k + herk + symm + syr2k + syrk + trmm + trsm diff --git a/docs/domains/blas/blas.rst b/docs/domains/blas/blas.rst new file mode 100644 index 000000000..313673bf4 --- /dev/null +++ b/docs/domains/blas/blas.rst @@ -0,0 +1,12 @@ +.. _onemkl_blas: + +BLAS Routines ++++++++++++++ + +oneMKL provides a DPC++ interface to the Basic Linear Algebra Subprograms (BLAS) routines. + +.. include:: blas-level-1-routines.inc.rst +.. include:: blas-level-2-routines.inc.rst +.. include:: blas-level-3-routines.inc.rst + +**Parent topic:** :ref:`onemkl` diff --git a/docs/domains/blas/copy.rst b/docs/domains/blas/copy.rst new file mode 100644 index 000000000..df47419aa --- /dev/null +++ b/docs/domains/blas/copy.rst @@ -0,0 +1,116 @@ +.. _copy: + +copy +==== + + +.. container:: + + + Copies a vector to another vector. + + + .. container:: section + :name: GUID-D6B6C72E-9516-40C9-B034-9F344C41AAF3 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void copy(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy) + + ``copy`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-5E0A9C5F-BDD5-41E6-97CD-4316FD58C347 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The copy routines copy one vector to another: + + + + + + y ←x + + + where x and y are vectors of n elements. + + +.. container:: section + :name: GUID-6F86EF6A-8FFE-4C6A-8B71-23B95C1F1365 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector x. + + + x + Buffer holding input vector x. The buffer must be of size at least + ``(1 + (n – 1)*abs(incx))``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + incy + Stride of vector y. + + +.. container:: section + :name: GUID-4ABB603B-835C-428B-B880-2F088BAB5456 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector y. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/dot.rst b/docs/domains/blas/dot.rst new file mode 100644 index 000000000..7388eed8c --- /dev/null +++ b/docs/domains/blas/dot.rst @@ -0,0 +1,136 @@ +.. _dot: + +dot +=== + + +.. container:: + + + Computes the dot product of two real vectors. + + + .. container:: section + :name: GUID-13355B56-0278-45E5-B310-3B0AC541C675 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void dot(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &result) + + ``dot`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + - T_res + * - ``float`` + - ``float`` + * - ``double`` + - ``double`` + * - ``float`` + - ``double`` + + + + +.. container:: section + :name: GUID-4BC6BF9A-BAB9-4078-A6B5-9C7ECB9D4821 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The dot routines perform a dot product between two vectors: + + + |image0| + + + .. container:: Note + + + .. rubric:: Note + :name: note + :class: NoteTipHead + + + For the mixed precision version (inputs are float while result is + double), the dot product is computed with double precision. + + +.. container:: section + :name: GUID-6F86EF6A-8FFE-4C6A-8B71-23B95C1F1365 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vectors x and y. + + + x + Buffer holding input vector x. The buffer must be of size at least + ``(1 + (n – 1)*abs(incx))``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector y. The buffer must be of size at least + ``(1 + (n – 1)*abs(incy))``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + +.. container:: section + :name: GUID-CAAFE234-AF82-4B61-8406-D57EC527BED5 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + Buffer where the result (a scalar) will be stored. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-93DA36DC-40CA-4C01-B883-DABAB0D37ee1.png + :class: img-middle + diff --git a/docs/domains/blas/dotc.rst b/docs/domains/blas/dotc.rst new file mode 100644 index 000000000..08e07d1d3 --- /dev/null +++ b/docs/domains/blas/dotc.rst @@ -0,0 +1,121 @@ +.. _dotc: + +dotc +==== + + +.. container:: + + + Computes the dot product of two complex vectors, conjugating the + first vector. + + + .. container:: section + :name: GUID-9D36611B-564D-475B-8D98-5F53A4F698F5 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void dotc(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &result) + + ``dotc`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-3E4588D2-5FDE-43F1-955E-85173AE62252 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The ``dotc`` routines perform a dot product between two complex + vectors, conjugating the first of them: + + + |image0| + + +.. container:: section + :name: GUID-38675523-DEDD-4314-8486-7C66614ED2C7 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + The number of elements in vectors ``x`` and ``y``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + The stride of vector ``x``. + + + y + Buffer holding input vector ``y``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details.. + + + incy + The stride of vector ``y``. + + +.. container:: section + :name: GUID-B84A5D05-6B61-4D13-8185-2A349C41CE46 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + The buffer where the result (a scalar) is stored. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-AED001B6-9056-491F-ACBE-E06C82D17ee1.png + :class: img-middle + diff --git a/docs/domains/blas/dotu.rst b/docs/domains/blas/dotu.rst new file mode 100644 index 000000000..15cb71e7d --- /dev/null +++ b/docs/domains/blas/dotu.rst @@ -0,0 +1,119 @@ +.. _dotu: + +dotu +==== + + +.. container:: + + + Computes the dot product of two complex vectors. + + + .. container:: section + :name: GUID-27A695AE-7ED5-4CFF-9783-0E50D111BED2 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void dotu(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &result) + + ``dotu`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-7E67CFC6-917F-41A3-A664-F99EE4E04E43 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The dotu routines perform a dot product between two complex vectors: + + + |image0| + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vectors ``x`` and ``y``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector ``y``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + +.. container:: section + :name: GUID-2B160DEB-ADBB-4044-8078-4B613A0DA4E1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + Buffer where the result (a scalar) is stored. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-3605ACD9-02D1-46D7-B791-F2F76F0D9ee1.png + :class: img-middle + diff --git a/docs/domains/blas/gbmv.rst b/docs/domains/blas/gbmv.rst new file mode 100644 index 000000000..524d52972 --- /dev/null +++ b/docs/domains/blas/gbmv.rst @@ -0,0 +1,185 @@ +.. _gbmv: + +gbmv +==== + + +.. container:: + + + Computes a matrix-vector product with a general band matrix. + + + .. container:: section + :name: GUID-870EA7B0-09B5-43FF-90A4-6378B5D94B55 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void gbmv(queue &exec_queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, std::int64_t ku, T alpha, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``gbmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-71614419-BC91-4A1A-B743-FE52767C4926 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The ``gbmv`` routines compute a scalar-matrix-vector product and add + the result to a scalar-vector product, with a general band matrix. + The operation is defined as + + + + + + y <- alpha*op(A)*x + beta*y + + + where: + + + - op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + - ``alpha`` and ``beta`` are scalars, + + + - ``A`` is an ``m``-by-``n`` matrix with ``kl`` sub-diagonals and + ``ku`` super-diagonals, + + + - ``x`` and ``y`` are vectors. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + m + Number of rows of ``A``. Must be at least zero. + + + n + Number of columns of ``A``. Must be at least zero. + + + kl + Number of sub-diagonals of the matrix ``A``. Must be at least + zero. + + + ku + Number of super-diagonals of the matrix ``A``. Must be at least + zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least (``kl`` + + ``ku`` + 1), and positive. + + + x + Buffer holding input vector ``x``. The length ``len`` of vector + ``x`` is ``n`` if ``A`` is not transposed, and ``m`` if ``A`` is + transposed. The buffer must be of size at least (1 + (``len`` - + 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + beta + Scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The length ``len`` of + vector ``y`` is ``m``, if ``A`` is not transposed, and ``n`` if + ``A`` is transposed. The buffer must be of size at least (1 + + (``len`` - 1)*abs(``incy``)) where ``len`` is this length. See + `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-4B31584D-BC63-4032-A4A7-61BF3F163165 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/gemm.rst b/docs/domains/blas/gemm.rst new file mode 100644 index 000000000..8e529b1e8 --- /dev/null +++ b/docs/domains/blas/gemm.rst @@ -0,0 +1,216 @@ +.. _gemm: + +gemm +==== + + +.. container:: + + + Computes a matrix-matrix product with general matrices. + + + .. container:: section + :name: GUID-7885D940-FAC1-4F37-9E1C-A022DED99EBD + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void gemm(queue &exec_queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, T beta, buffer &c, std::int64_t ldc) + + ``gemm`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``half`` + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-14237C95-6322-47A4-BC11-D3CDD2118C42 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The gemm routines compute a scalar-matrix-matrix product and add the + result to a scalar-matrix product, with general matrices. The + operation is defined as + + + + + + C <- alpha*op(A)*op(B) + beta*C + + + where: + + + ``op(X)`` is one of ``op(X) = X``, or ``op(X) = XT``, or + ``op(X) = XH``, + + + ``alpha`` and ``beta`` are scalars, + + + ``A``, ``B`` and ``C`` are matrices: + + + ``op(A)`` is an ``m``-by-``k`` matrix, + + + ``op(B)`` is a ``k``-by-``n`` matrix, + + + ``C`` is an ``m``-by-``n`` matrix. + + +.. container:: section + :name: GUID-D89C4959-F0C2-4E91-8853-9225F0772DF0 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + transa + Specifies the form of ``op(A)``, the transposition operation + applied to ``A``. See + :ref:`onemkl_datatypes` + for more details. + + + transb + Specifies the form of ``op(B)``, the transposition operation + applied to ``B``. See + :ref:`onemkl_datatypes` + for more details. + + + m + Specifies the number of rows of the matrix ``op(A)`` and of the + matrix ``C``. The value of m must be at least zero. + + + n + Specifies the number of columns of the matrix ``op(B)`` and the + number of columns of the matrix ``B``. The value of n must be at + least zero. + + + k + Specifies the number of columns of the matrix ``op(A)`` and the + number of rows of the matrix ``op(B)``. The value of k must be at + least zero. + + + alpha + Scaling factor for the matrix-matrix product. + + + a + The buffer holding the input matrix ``A``. If ``A`` is not + transposed, ``A`` is an ``m``-by-``k`` matrix so the array ``a`` + must have size at least ``lda``\ \*\ ``k``. If ``A`` is + transposed, ``A`` is an ``k``-by-``m`` matrix so the array ``a`` + must have size at least ``lda``\ \*\ ``m``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + The leading dimension of ``A``. Must be at least m if ``A`` is not + transposed, and at least k if ``A`` is transposed. It must be + positive. + + + b + The buffer holding the input matrix ``B``. If ``B`` is not + transposed, ``B`` is an ``k``-by-``n`` matrix so the array ``b`` + must have size at least ``ldb``\ \*\ ``n``. If ``B`` is + transposed, ``B`` is an ``n``-by-``k`` matrix so the array ``b`` + must have size at least ``ldb``\ \*\ ``k``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + The leading dimension of ``B``. Must be at least k if ``B`` is not + transposed, and at least n if ``B`` is transposed. It must be + positive. + + + beta + Scaling factor for matrix ``C``. + + + c + The buffer holding the input/output matrix ``C``. It must have a + size of at least ldc\*n. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + The leading dimension of ``C``. It must be positive and at least + the size of m. + + +.. container:: section + :name: GUID-EEF5C7D0-D206-4961-809F-55DCA3E93F68 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + The buffer, which is overwritten by + ``alpha*op(A)*op(B) + beta*C``. + + +.. container:: section + :name: GUID-AC72653A-4AC8-4B9D-B7A9-13A725AA19BF + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``beta`` = 0, matrix ``C`` does not need to be initialized before + calling ``gemm``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/gemm_batch.rst b/docs/domains/blas/gemm_batch.rst new file mode 100644 index 000000000..11034ab68 --- /dev/null +++ b/docs/domains/blas/gemm_batch.rst @@ -0,0 +1,440 @@ +.. _gemm_batch: + +gemm_batch +========== + + +.. container:: + + + Computes groups of matrix-matrix product with general matrices. + + + .. container:: section + :name: GUID-7885D940-FAC1-4F37-9E1C-A022DED99EBD + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + **Group API** + + + .. cpp:function:: void gemm_batch(queue &exec_queue, buffer &transa_array, buffer &transb_array, buffer &m_array, buffer &n_array, buffer &k_array, buffer alpha_array, buffer &a_array, buffer &lda_array, buffer &b_array, buffer ldb_array, buffer &beta_array, buffer &c, buffer &ldc_array, std::int64_t group_count, buffer &group_size_array) + + **Strided API** + + + .. cpp:function:: void gemm_batch(queue &exec_queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t &lda, std::int64_t stridea, buffer &b, std::int64_t ldb, std::int64_t strideb, T beta, buffer &c, std::int64_t ldc, std::int64_t stridec, std::int64_t batch_size) + + ``gemm_batch`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-14237C95-6322-47A4-BC11-D3CDD2118C42 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The gemm_batch routines perform a series of matrix-matrix operations + with general matrices. They are similar to the gemm routine + counterparts, but the gemm_batch routines perform matrix-matrix + operations with groups of matrices. The groups contain matrices with + the same parameters. + + + For the group API, the operation is defined as + + + :: + + + offa = 0, offb = 0, offc = 0 + for i = 0 … group_count – 1 + transa, transb, m, n, k, lda, ldb, ldc, alpha, beta and group_size at position i in transa_array, transb_array, m_array, n_array, k_array, lda_array, ldb_array, ldc_array, alpha_array, beta_array and group_size_array + sizea = transa == onemkl::transpose::N ? lda * k : lda * m; + sizeb = transb == onemkl::transpose::N ? ldb * n : ldb * k; + sizec = ldc * n; + for j = 0 … group_size – 1 + A, B, and C are matrices of size sizea, sizeb and sizec at offset offa, offb and offc in a, b and c. + C := alpha * op(A) * op(B) + beta * C + offa += sizea, offb += sizeb, offc += sizec + end for + end for + + + For the strided API, the operation is defined as + + + :: + + + for i = 0 … batch_size – 1 + A, B and C are matrices at offset i * stridea, i * strideb, i * stridec in a, b and c. + C = alpha * op(A) * op(B) + beta * C + end for + + + where: + + + - op(X) is one of op(X) = X, or op(X) = X\ :sup:`T`, or op(X) = + X\ :sup:`H` + + + - ``alpha`` and ``beta`` are scalars + + + - ``A``, ``B``, and ``C`` are matrices + + + - The a, b and c buffers contains all the input matrices. The stride + between matrices is either given by the exact size of the matrix + (for the group API) or by the stride parameter. The total number + of matrices in a, b and c buffers is given by the + + |image0| + + for the + group API or by the ``batch_size`` parameter for the strided API. + + + Here, op(``A``) is ``m``\ ``x``\ ``k``, op(``B``) is + ``k``\ ``x``\ ``n``, and ``C`` is ``m``\ ``x``\ ``n``. + + +.. container:: section + :name: GUID-863264A0-4CE9-495F-A617-102E46D7A41A + + + .. rubric:: Input Parameters - Group API + :name: input-parameters---group-api + :class: sectiontitle + + + transa_array + Buffer holding ``group_count onemkl::transpose`` value. + + + For the group ``i``, ``transa`` is the ``i``\ th element in the + transa_array buffer and specifies the form of ``op(A)`` used in + the matrix multiplication. See + :ref:`onemkl_datatypes` for more + details. + + + transb_array + Buffer holding ``group_count onemkl::transpose`` value. + + + For the group ``i``, ``transb`` is the ``i``\ th element in the + transb_array buffer and specifies the form of ``op(B)`` used in + the matrix multiplication. See + :ref:`onemkl_datatypes` for more + details. + + + m_array + Buffer holding ``group_count`` integer. For the group ``i``, ``m`` + is the ``i``\ th element in the m_array buffer and specifies the + number of rows of ``op(A)`` and ``C``. Must be at least zero. + + + n_array + Buffer holding ``group_count`` integer. For the group ``i``, ``n`` + is the ``i``\ th element in the n_array buffer and specifies the + number of columns of ``op(B)`` and ``C``. Must be at least zero. + + + k_array + Buffer holding ``group_count`` integer. For the group ``i``, ``k`` + is the ``i``\ th element in the k_array buffer and specifies the + number of columns of ``op(A)`` and rows of ``op(B)``. Must be at + least zero. + + + alpha_array + Buffer holding ``group_count`` scalar element. For the group + ``i``, ``alpha`` is the ``i``\ th element in the alpha_array + buffer and specifies the scaling factor for the matrix-matrix + product. + + + a + Buffer holding the input matrices ``A``. The total size of the + buffer ``a`` must be at least the sum of the sizes of all the + matricies ``A``. That is, + + + |image1| + + + where + ``sizeai = lda_array[i] * (transa == onemkl::transpose::N ? k : m)`` + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + lda_array + Buffer holding ``group_count`` integer. For the group ``i``, + ``lda`` is the ``i``\ th element in the lda_array buffer and + specifies the leading dimension of ``A``. Must be at least ``m`` + if ``A`` is not transposed, and at least ``k`` if ``A`` is + transposed. Must be positive. + + + b + Buffer holding the input matrices ``B``. The total size of the + buffer ``b`` must be at least the sum of the sizes of all the + matricies ``B``. That is, + + + |image2| + + + where + ``sizebi = ldb_array[i] * (transb == onemkl::transpose::N ? n : k)`` + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldb_array + Buffer holding ``group_count`` integer. For the group ``i``, + ``ldb`` is the ``i``\ th element in the ldb_array buffer and + specifies the leading dimension of ``B``. Must be at least ``k`` + if ``B`` is not transposed, and at least ``n`` if ``B`` is + transposed. Must be positive. + + + beta_array + Buffer holding ``group_count`` scalar element. For the group + ``i``, ``beta`` is the ``i``\ th element in the beta_array buffer + and specifies the scaling factor for matrix C. + + + c + Buffer holding the input/output matrices ``C``. The total size of + the buffer ``c`` must be at least the sum of the sizes of all the + matricies ``C``. That is, + + + |image3| + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldc_array + Buffer holding ``group_count`` integer. For the group ``i``, + ``ldc`` is the ``i``\ th element in the ldc_array buffer and + specifies the leading dimension of ``C``. Must be positive and at + least ``m``. + + + group_count + Specifies the number of groups. Must be at least 0. + + + group_size_array + Buffer holding ``group_count`` integer. For the group ``i``, the + ``i``\ th element in the group_size_array buffer specifies the + number of matrix multiply operations in group ``i``. Each element + in ``group_size_array`` must be at least 0. + + +.. container:: section + :name: GUID-1E4953E6-F7B1-4FEE-BA5A-8C4BD51DC700 + + + .. rubric:: Output Parameters - Group API + :name: output-parameters---group-api + :class: sectiontitle + + + c + Overwritten by the ``m``\ :sub:`i`-by-``n``\ :sub:`i` matrix + ``(alphai*op(A)*op(B) + betai*C)`` for group ``i``. + + +.. container:: section + :name: GUID-D067773A-45A3-4D24-B10A-46E27834947E + + + .. rubric:: Input Parameters - Strided API + :name: input-parameters---strided-api + :class: sectiontitle + + + transa + Specifies ``op(A)`` the transposition operation applied to the + matrices A. See + :ref:`onemkl_datatypes` for more + details. + + + transb + Specifies ``op(B)`` the transposition operation applied to the + matrices B. See + :ref:`onemkl_datatypes` for more + details. + + + m + Number of rows of ``op(A)`` and ``C``. Must be at least zero. + + + n + Number of columns of ``op(B)`` and ``C``. Must be at least zero. + + + k + Number of columns of ``op(A)`` and rows of ``op(B)``. Must be at + least zero. + + + alpha + Scaling factor for the matrix-matrix products. + + + a + Buffer holding the input matrices ``A``. Must have size at least + ``stridea*batch_size``. + + + lda + Leading dimension of the matrices ``A``. Must be at least ``m`` if + the matrices ``A`` are not transposed, and at least ``k`` if the + matrices ``A`` are transposed. Must be positive. + + + stridea + Stride between the different ``A`` matrices. + + + If ``A`` are not transposed, the matrices ``A`` are ``m``-by-``k`` + matrices so stridea must be at least ``lda*k``. + + + If ``A`` are transposed, the matrices ``A`` are ``k``-by-``m`` + matrices so stridea must be at least ``lda*m``. + + + b + Buffer holding the input matrices ``B``. Must have size at least + ``strideb*batch_size``. + + + ldb + Leading dimension of the matrices ``B``. Must be at least ``k`` if + the matrices ``B`` are not transposed, and at least ``n`` if the + matrices ``B`` are transposed. Must be positive. + + + strideb + Stride between the different ``B`` matrices. + + + If ``B`` are not transposed, the matrices ``B`` are ``k``-by-``n`` + matrices so strideb must be at least ``ldb*n``. + + + If ``B`` are transposed, the matrices ``B`` are ``n``-by-``k`` + matrices so strideb must be at least ``ldb*k``. + + + beta + Scaling factor for the matrices ``C``. + + + c + Buffer holding input/output matrices ``C``. Must have size at + least ``stridec*batch_size``. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``m``. + + + stridec + Stride between the different ``C`` matrices. Must be at least + ``ldc*n``. + + + batch_size + Specifies the number of matrix multiply operations to perform. + + +.. container:: section + :name: GUID-98C3DE17-4F5F-41A1-B431-48148153ABBA + + + .. rubric:: Output Parameters - Strided API + :name: output-parameters---strided-api + :class: sectiontitle + + + c + Output buffer, overwritten by ``batch_size`` matrix multiply + operations of the form\ ``alpha*op(A)*op(B) + beta*C``. + + +.. container:: section + :name: GUID-AC72653A-4AC8-4B9D-B7A9-13A725AA19BF + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``beta`` = 0, matrix ``C`` does not need to be initialized before + calling gemm_batch. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-like-extensions` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee1.png + :class: img-middle +.. |image1| image:: ../equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee2.png + :class: img-middle +.. |image2| image:: ../equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee3.png + :class: img-middle +.. |image3| image:: ../equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee4.png + :class: img-middle + diff --git a/docs/domains/blas/gemm_ext.rst b/docs/domains/blas/gemm_ext.rst new file mode 100644 index 000000000..a49fabe8d --- /dev/null +++ b/docs/domains/blas/gemm_ext.rst @@ -0,0 +1,335 @@ +.. _gemm_ext: + +gemm_ext +======== + + +.. container:: + + + Computes a matrix-matrix product with general matrices. + + + .. container:: section + :name: GUID-7885D940-FAC1-4F37-9E1C-A022DED99EBD + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + **Standard API** + + + .. container:: dlsyntaxpara + + + .. cpp:function:: void gemm_ext(queue &exec_queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, Ts alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, Ts beta, buffer &c, std::int64_t ldc) + + ``gemm_ext`` supports the following precisions and devices. + + + .. list-table:: + :header-rows: 1 + + * - Ts + - Ta + - Tb + - Tc + * - ``float`` + - ``half`` + - ``half`` + - ``float`` + * - ``half`` + - ``half`` + - ``half`` + - ``half`` + * - ``float`` + - ``float`` + - ``float`` + - ``float`` + * - ``double`` + - ``double`` + - ``double`` + - ``double`` + * - ``std::complex`` + - ``std::complex`` + - ``std::complex`` + - ``std::complex`` + * - ``std::complex`` + - ``std::complex`` + - ``std::complex`` + - ``std::complex`` + + + + + **Offset API** + + + .. container:: dlsyntaxpara + + + .. cpp:function:: void gemm_ext(queue &exec_queue, transpose transa, transpose transb, offset offset_type, std::int64_t m, std::int64_t n, std::int64_t k, Ts alpha, buffer &a, std::int64_t lda, Ta ao, buffer &b, std::int64_t ldb, Tb bo, Ts beta, buffer &c, std::int64_t ldc, buffer &co) + + ``gemm_ext`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - Ts + - Ta + - Tb + - Tc + * - ``float`` + - ``int8_t`` + - ``uint8_t`` + - ``int32_t`` + + + + + .. container:: section + :name: GUID-14237C95-6322-47A4-BC11-D3CDD2118C42 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The gemm_ext routines compute a scalar-matrix-matrix product and + add the result to a scalar-matrix product, with general matrices. + The operation is defined as: + + + :: + + + C ← alpha*op(A)*op(B) + beta*C + + + for the standard API and + :: + + + C ← alpha*(op(A) - A_offset)*(op(B) - B_offset) + beta*C + C_offset + + + for the offset API + where: + + + - op(X) is one of op(X) = X, or op(X) = X\ :sup:`T`, or op(X) = + X\ :sup:`H` + + + - ``alpha`` and ``beta`` are scalars + + + - ``A_offset`` is an ``m``-by-``k`` matrix with every element + equal to the value ao + + + - ``B_offset`` is a ``k``-by-``n`` matrix with every element + equal to the value bo + + + - ``C_offset`` is an ``m``-by-``n`` matrix defined by the co + buffer as described in + :ref:`onemkl_datatypes` + + + - ``A``, ``B``, and ``C`` are matrices + + + Here, op(``A``) is ``m`` x ``k``, op(``B``) is ``k`` x ``n``, and + ``C`` is ``m`` x ``n``. + + + .. container:: section + :name: GUID-863264A0-4CE9-495F-A617-102E46D7A41A + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + transa + Specifies op(``A``), the transposition operation applied to + ``A``. See + :ref:`onemkl_datatypes` for + more details. + + + transb + Specifies op(``B``), the transposition operation applied to + ``B``. See + :ref:`onemkl_datatypes` for + more details. + + + offset_type (offset API only) + Specifies the form of ``C_offset`` used in the matrix + multiplication. See + :ref:`onemkl_datatypes` for + more details. + + + m + Number of rows of op(``A``) and ``C``. Must be at least zero. + + + n + Number of columns of op(``B``) and ``C``. Must be at least + zero. + + + k + Number of columns of op(``A``) and rows of op(``B``). Must be + at least zero. + + + alpha + Scaling factor for the matrix-matrix product. + + + a + Buffer holding the input matrix ``A``. + + + If ``A`` is not transposed, ``A`` is an ``m``-by-``k`` matrix + so the array ``a`` must have size at least ``lda``\ \*\ ``k``. + + + If ``A`` is transposed, ``A`` is a ``k``-by-``m`` matrix so the + array ``a`` must have size at least ``lda``\ \*\ ``m``. + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``m`` if ``A`` is + not transposed, and at least ``k`` if ``A`` is transposed. Must + be positive. + + + ao (offset API only) + Specifies the scalar offset value for matrix ``A``. + + + b + Buffer holding the input matrix ``B``. + + + If ``B`` is not transposed, ``B`` is a ``k``-by-``n`` matrix so + the array ``b`` must have size at least ``ldb``\ \*\ ``n``. + + + If ``B`` is transposed, ``B`` is an ``n``-by-``k`` matrix so + the array ``b`` must have size at least ``ldb``\ \*\ ``k``. + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be at least ``k`` if ``B`` is + not transposed, and at least ``n`` if ``B`` is transposed. Must + be positive. + + + bo (offset API only) + Specifies the scalar offset value for matrix ``B``. + + + beta + Scaling factor for matrix ``C``. + + + c + Buffer holding the input matrix ``C``. Must have size at least + ``ldc`` \* ``n``. See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least + ``m``. + + + co (offset API only) + Buffer holding the offset values for matrix ``C``. + + + If ``offset_type = offset::fix``, the ``co`` array must have + size at least 1. + + + If ``offset_type = offset::col``, the ``co`` array must have + size at least ``max(1,m)``. + + + If ``offset_type = offset::row``, the ``co`` array must have + size at least ``max(1,n)``. + + + See + :ref:`onemkl_datatypes` for + more details. + + + .. container:: section + :name: GUID-1E4953E6-F7B1-4FEE-BA5A-8C4BD51DC700 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + Output buffer, overwritten by alpha\*op(``A``)*op(``B``) + + beta\*\ ``C`` for the standard API and alpha\*(op(``A``) - + ``A_offset``)*(op(``B``) - ``B_offset``) + beta\*\ ``C`` + + ``C_offset`` for the offset API. + + + .. container:: section + :name: GUID-AC72653A-4AC8-4B9D-B7A9-13A725AA19BF + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``beta`` = 0, matrix ``C`` does not need to be initialized + before calling gemm_ext. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-like-extensions` + + + +.. container:: + diff --git a/docs/domains/blas/gemmt.rst b/docs/domains/blas/gemmt.rst new file mode 100644 index 000000000..e2ceb077d --- /dev/null +++ b/docs/domains/blas/gemmt.rst @@ -0,0 +1,228 @@ +.. _gemmt: + +gemmt +===== + + +.. container:: + + + Computes a matrix-matrix product with general matrices, but updates + only the upper or lower triangular part of the result matrix. + + + .. container:: section + :name: GUID-7885D940-FAC1-4F37-9E1C-A022DED99EBD + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. container:: dlsyntaxpara + + + .. cpp:function:: void gemmt(queue &exec_queue, uplo upper_lower, transpose transa, transpose transb, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, T beta, buffer &c, std::int64_t ldc) + + ``gemmt`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + + .. container:: section + :name: GUID-14237C95-6322-47A4-BC11-D3CDD2118C42 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The gemmt routines compute a scalar-matrix-matrix product and add + the result to the upper or lower part of a scalar-matrix product, + with general matrices. The operation is defined as: + + + :: + + + C <- alpha*op(A)*op(B) + beta*C + + + where: + + + - op(X) is one of op(X) = X, or op(X) = X\ :sup:`T`, or op(X) = + X\ :sup:`H` + + + - ``alpha`` and ``beta`` are scalars + + + - ``A``, ``B``, and ``C`` are matrices + + + Here, op(``A``) is ``n`` x ``k``, op(``B``) is ``k`` x ``n``, and + ``C`` is ``n`` x ``n``. + + + .. container:: section + :name: GUID-863264A0-4CE9-495F-A617-102E46D7A41A + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``C``\ ’s data is stored in its upper or + lower triangle. See + :ref:`onemkl_datatypes` for + more details. + + + transa + Specifies op(``A``), the transposition operation applied to + ``A``. See + :ref:`onemkl_datatypes` for + more details. + + + transb + Specifies op(``B``), the transposition operation applied to + ``B``. See + :ref:`onemkl_datatypes` for + more details. + + + n + Number of columns of op(``A``), columns of op(``B``), and + columns of\ ``C``. Must be at least zero. + + + k + Number of columns of op(``A``) and rows of op(``B``). Must be + at least zero. + + + alpha + Scaling factor for the matrix-matrix product. + + + a + Buffer holding the input matrix ``A``. + + + If ``A`` is not transposed, ``A`` is an ``n``-by-``k`` matrix + so the array ``a`` must have size at least ``lda``\ \*\ ``k``. + + + If ``A`` is transposed, ``A`` is a ``k``-by-``n`` matrix so the + array ``a`` must have size at least ``lda``\ \*\ ``n``. + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``n`` if ``A`` is + not transposed, and at least ``k`` if ``A`` is transposed. Must + be positive. + + + b + Buffer holding the input matrix ``B``. + + + If ``B`` is not transposed, ``B`` is a ``k``-by-``n`` matrix so + the array ``b`` must have size at least ``ldb``\ \*\ ``n``. + + + If ``B`` is transposed, ``B`` is an ``n``-by-``k`` matrix so + the array ``b`` must have size at least ``ldb``\ \*\ ``k``. + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be at least ``k`` if ``B`` is + not transposed, and at least ``n`` if ``B`` is transposed. Must + be positive. + + + beta + Scaling factor for matrix ``C``. + + + c + Buffer holding the input/output matrix ``C``. Must have size at + least ``ldc`` \* ``n``. See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least + ``m``. + + + .. container:: section + :name: GUID-1E4953E6-F7B1-4FEE-BA5A-8C4BD51DC700 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + Output buffer, overwritten by the upper or lower triangular + part ofalpha\*op(``A``)*op(``B``) + beta\*\ ``C``. + + + .. container:: section + :name: GUID-AC72653A-4AC8-4B9D-B7A9-13A725AA19BF + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``beta`` = 0, matrix ``C`` does not need to be initialized + before calling gemmt. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-like-extensions` + + + +.. container:: + diff --git a/docs/domains/blas/gemv.rst b/docs/domains/blas/gemv.rst new file mode 100644 index 000000000..1345bcdf0 --- /dev/null +++ b/docs/domains/blas/gemv.rst @@ -0,0 +1,174 @@ +.. _gemv: + +gemv +==== + + +.. container:: + + + Computes a matrix-vector product using a general matrix. + + + .. container:: section + :name: GUID-EA8D6705-E7C2-42E2-BE80-D9AD83645FCC + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void gemv(queue &exec_queue, transpose trans, std::int64_t m, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + gemv supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-AE220EED-6066-4881-8B3C-35207BAB0105 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The gemv routines compute a scalar-matrix-vector product and add the + result to a scalar-vector product, with a general matrix. The + operation is defined as + + + + + + y <- alpha*op(A)*x + beta*y + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``m``-by-``n`` matrix, and ``x``, ``y`` are vectors. + + +.. container:: section + :name: GUID-F3E8F201-6033-45A1-A326-CA4CFB631C3A + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + trans + Specifies ``op(A)``, the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + m + Specifies the number of rows of the matrix ``A``. The value of + ``m`` must be at least zero. + + + n + Specifies the number of columns of the matrix ``A``. The value of + ``n`` must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + The buffer holding the input matrix ``A``. Must have a size of at + least ``lda``\ \*n. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + The leading dimension of matrix ``A``. It must be at least m, and + positive. + + + x + Buffer holding input vector ``x``. The length ``len`` of vector + ``x`` is ``n`` if ``A`` is not transposed, and ``m`` if ``A`` is + transposed. The buffer must be of size at least (1 + (``len`` - + 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + The stride of vector ``x``. + + + beta + The scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The length ``len`` of + vector ``y`` is ``m``, if ``A`` is not transposed, and ``n`` if + ``A`` is transposed. The buffer must be of size at least (1 + + (``len`` - 1)*abs(``incy``)) where ``len`` is this length. See + `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + The stride of vector ``y``. + + +.. container:: section + :name: GUID-1533BCA6-E652-4A08-A82D-162F3CEBDD29 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + The buffer holding updated vector ``y``. + + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/ger.rst b/docs/domains/blas/ger.rst new file mode 100644 index 000000000..1a122ce84 --- /dev/null +++ b/docs/domains/blas/ger.rst @@ -0,0 +1,154 @@ +.. _ger: + +ger +=== + + +.. container:: + + + Computes a rank-1 update of a general matrix. + + + .. container:: section + :name: GUID-0DA23698-EB19-4AAF-A5FD-9BB530A9EFE0 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void ger(queue &exec_queue, std::int64_t m, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a, std::int64_t lda) + + ``ger`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-72E035B0-E1C2-442B-AE9D-2CB873E90FAF + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The ger routines compute a scalar-vector-vector product and add the + result to a general matrix. The operation is defined as + + + + + + A <- alpha*x*y :sup:`T` + A + + + where: + + + ``alpha`` is scalar, + + + ``A`` is an ``m``-by-``n`` matrix, + + + ``x`` is a vector length ``m``, + + + ``y`` is a vector length ``n``. + + +.. container:: section + :name: GUID-6953A2E5-0065-425C-986B-15966C793067 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + m + Number of rows of ``A``. Must be at least zero. + + + n + Number of columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``m`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``m``, and + positive. + + +.. container:: section + :name: GUID-E2A13688-1D12-4DD0-9752-3557E980ACC0 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated matrix ``A``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/gerc.rst b/docs/domains/blas/gerc.rst new file mode 100644 index 000000000..5a9c772ac --- /dev/null +++ b/docs/domains/blas/gerc.rst @@ -0,0 +1,154 @@ +.. _gerc: + +gerc +==== + + +.. container:: + + + Computes a rank-1 update (conjugated) of a general complex matrix. + + + .. container:: section + :name: GUID-5A1B0292-28F6-45EB-95C4-FDA03D8D5062 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void gerc(queue &exec_queue, std::int64_t m, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a, std::int64_t lda) + + ``gerc`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-6CB627E5-A9C7-488D-8366-E7944A5C889E + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The gerc routines compute a scalar-vector-vector product and add the + result to a general matrix. The operation is defined as + + + + + + A <- alpha*x*y :sup:`H` + A + + + where: + + + ``alpha`` is a scalar, + + + ``A`` is an ``m``-by-``n`` matrix, + + + ``x`` is a vector of length ``m``, + + + ``y`` is vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + m + Number of rows of ``A``. Must be at least zero. + + + n + Number of columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``m`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``m``, and + positive. + + +.. container:: section + :name: GUID-48944ED2-C10F-4B64-A91A-C9050AD24A92 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated matrix *A*. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/geru.rst b/docs/domains/blas/geru.rst new file mode 100644 index 000000000..121e2d13c --- /dev/null +++ b/docs/domains/blas/geru.rst @@ -0,0 +1,154 @@ +.. _geru: + +geru +==== + + +.. container:: + + + Computes a rank-1 update (unconjugated) of a general complex matrix. + + + .. container:: section + :name: GUID-5942D28E-EDD6-4759-B19E-FBB51F35125B + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void geru(queue &exec_queue, std::int64_t m, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a, std::int64_t lda) + + ``geru`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-75ECE219-BA77-48E8-B13B-FB504DD60CD4 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The geru routines routines compute a scalar-vector-vector product and + add the result to a general matrix. The operation is defined as + + + + + + A <- alpha*x*y :sup:`T` + A + + + where: + + + ``alpha`` is a scalar, + + + ``A`` is an ``m``-by-``n`` matrix, + + + ``x`` is a vector of length ``m``, + + + ``y`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + m + Number of rows of ``A``. Must be at least zero. + + + n + Number of columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``m`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``m``, and + positive. + + +.. container:: section + :name: GUID-6E9315E9-DDCF-485D-8BDF-AB4BF8448BE1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated matrix ``A``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/hbmv.rst b/docs/domains/blas/hbmv.rst new file mode 100644 index 000000000..e23481a2a --- /dev/null +++ b/docs/domains/blas/hbmv.rst @@ -0,0 +1,164 @@ +.. _hbmv: + +hbmv +==== + + +.. container:: + + + Computes a matrix-vector product using a Hermitian band matrix. + + + .. container:: section + :name: GUID-F5FF420B-922B-4552-8F55-6EBCA7177881 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void hbmv(queue &exec_queue, uplo upper_lower, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``hbmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-8AB4BAC9-8124-4B52-8C15-1BC673820EB9 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The hbmv routines compute a scalar-matrix-vector product and add the + result to a scalar-vector product, with a Hermitian band matrix. The + operation is defined as + + + + + + y <- alpha*A*x + beta*y + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``n``-by-``n`` Hermitian band matrix, with ``k`` + super-diagonals, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + k + Number of super-diagonals of the matrix ``A``. Must be at least + zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least (``k`` + 1), + and positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``m`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + beta + Scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-7261182A-450B-46F5-8C61-7133597D3530 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/hemm.rst b/docs/domains/blas/hemm.rst new file mode 100644 index 000000000..bc597e228 --- /dev/null +++ b/docs/domains/blas/hemm.rst @@ -0,0 +1,214 @@ +.. _hemm: + +hemm +==== + + +.. container:: + + + Computes a matrix-matrix product where one input matrix is Hermitian + and one is general. + + + .. container:: section + :name: GUID-F06C86BA-4F57-4608-B0D7-F7B920F867D7 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void hemm(queue &exec_queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, T beta, buffer &c, std::int64_t ldc) + + hemm supports the following precisions: + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-835E7F58-406E-444F-9DFD-121B84C22284 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The hemm routines compute a scalar-matrix-matrix product and add the + result to a scalar-matrix product, where one of the matrices in the + multiplication is Hermitian. The argument ``left_right`` determines + if the Hermitian matrix, ``A``, is on the left of the multiplication + (``left_right`` = ``side::left``) or on the right (``left_right`` = + ``side::right``). Depending on ``left_right``, the operation is + defined as + + + + + + C <- alpha*A*B + beta*C + + + or + + + + + + C <- alpha*B*A + beta*C + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is a Hermitian matrix, either ``m``-by-``m`` or ``n``-by-``n`` + matrices, + + + ``B`` and ``C`` are ``m``-by-``n`` matrices. + + +.. container:: section + :name: GUID-922C5F92-38B2-457B-B6C7-3CDD0531F97D + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + left_right + Specifies whether ``A`` is on the left side of the multiplication + (``side::left``) or on the right side (``side::right``). See + :ref:`onemkl_datatypes` for more + details. + + + uplo + Specifies whether ``A``'s data is stored in its upper or lower + triangle. See + :ref:`onemkl_datatypes` for more + details. + + + m + Specifies the number of rows of the matrix ``B`` and ``C``. + + + The value of ``m`` must be at least zero. + + + n + Specifies the number of columns of the matrix ``B`` and ``C``. + + + The value of ``n`` must be at least zero. + + + alpha + Scaling factor for the matrix-matrix product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``m`` if ``A`` is on the left of the multiplication, + or ``lda``\ \*\ ``n`` if ``A`` is on the right. See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + lda + Leading dimension of ``A``. Must be at least ``m`` if ``A`` is on + the left of the multiplication, or at least ``n`` if ``A`` is on + the right. Must be positive. + + + b + Buffer holding input matrix ``B``. Must have size at least + ``ldb``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be positive and at least ``m``. + + + beta + Scaling factor for matrix ``C``. + + + c + Buffer holding input/output matrix ``C``. Must have size at least + ``ldc``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``m``. + + +.. container:: section + :name: GUID-94385C78-968D-4C03-AA5C-7379D5607800 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + +   + + + + c + Output buffer, overwritten by ``alpha``\ \*\ ``A``\ \*\ ``B`` + + ``beta``\ \*\ ``C`` (``left_right`` = ``side::left``) or + ``alpha``\ \*\ ``B``\ \*\ ``A`` + ``beta``\ \*\ ``C`` + (``left_right`` = ``side::right``). + + +.. container:: section + :name: EXAMPLE_5EF48B8A07D849EA84A74FE22F0D5B24 + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``beta`` = 0, matrix ``C`` does not need to be initialized before + calling ``hemm``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/hemv.rst b/docs/domains/blas/hemv.rst new file mode 100644 index 000000000..289cdc0a8 --- /dev/null +++ b/docs/domains/blas/hemv.rst @@ -0,0 +1,158 @@ +.. _hemv: + +hemv +==== + + +.. container:: + + + Computes a matrix-vector product using a Hermitian matrix. + + + .. container:: section + :name: GUID-152B72DC-F67F-4D7D-96DA-67AE6AD41718 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void hemv(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``hemv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-0E4AE01A-4FE8-42AC-B236-409F4DD48F88 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The hemv routines compute a scalar-matrix-vector product and add the + result to a scalar-vector product, with a Hermitian matrix. The + operation is defined as + + + + + + y <- alpha*A*x + beta*y + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``n``-by-``n`` Hermitian matrix, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether *A* is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``m``, and + positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + beta + Scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-66566E59-9A52-4207-B123-AF45FA3A0FBC + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/her.rst b/docs/domains/blas/her.rst new file mode 100644 index 000000000..a13196fb6 --- /dev/null +++ b/docs/domains/blas/her.rst @@ -0,0 +1,148 @@ +.. _her: + +her +=== + + +.. container:: + + + Computes a rank-1 update of a Hermitian matrix. + + + .. container:: section + :name: GUID-252B1D4A-30C7-4678-9793-6A0C90DEB04A + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void her(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &a, std::int64_t lda) + + ``her`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-A06B7C00-CFD6-4A01-9739-19093823B58E + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The her routines compute a scalar-vector-vector product and add the + result to a Hermitian matrix. The operation is defined as + + + + + + A <- alpha*x*x :sup:`H` + A + + + where: + + + ``alpha`` is scalar, + + + ``A`` is an ``n``-by-``n`` Hermitian matrix, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether *A* is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``n``, and + positive. + + +.. container:: section + :name: GUID-89A60481-0763-4608-B346-3CC746467F28 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated upper triangular part of theHermitian + matrix ``A`` if ``upper_lower = upper`` or the updated + lowertriangular part of the Hermitian matrix ``A`` if + ``upper_lower = lower``. + + + The imaginary parts of the diagonal elementsare set to zero. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/her2.rst b/docs/domains/blas/her2.rst new file mode 100644 index 000000000..0c100195c --- /dev/null +++ b/docs/domains/blas/her2.rst @@ -0,0 +1,159 @@ +.. _her2: + +her2 +==== + + +.. container:: + + + Computes a rank-2 update of a Hermitian matrix. + + + .. container:: section + :name: GUID-4BED3537-E900-4260-A6EB-2F42CB1D3AFB + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void her2(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a, std::int64_t lda) + + ``her2`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-2B939041-9BCC-4AE8-A31D-2CFCA67B9B6A + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The her2 routines compute two scalar-vector-vector products and add + them to a Hermitian matrix. The operation is defined as + + + + + + A <- alpha*x*y :sup:`H` + conjg(alpha)*y*x :sup:`H` + A + + + where: + + + ``alpha`` is a scalar, + + + ``A`` is an ``n``-by-``n`` Hermitian matrix. + + + ``x`` and ``y`` are vectors or length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether *A* is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``n``, and + positive. + + +.. container:: section + :name: GUID-34B3837B-4980-458B-AC3A-EEE5F635834C + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated upper triangular part of theHermitian + matrix ``A`` if ``upper_lower = upper``, or the updated + lowertriangular part of the Hermitian matrix ``A`` if + ``upper_lower = lower``. + + + The imaginary parts of the diagonal elementsare set to zero. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/her2k.rst b/docs/domains/blas/her2k.rst new file mode 100644 index 000000000..98e9cac4f --- /dev/null +++ b/docs/domains/blas/her2k.rst @@ -0,0 +1,198 @@ +.. _her2k: + +her2k +===== + + +.. container:: + + + Performs a Hermitian rank-2k update. + + + .. container:: section + :name: GUID-1839F1B0-EFE0-40A4-901E-53E7F9B395C2 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void her2k(queue &exec_queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, T_real beta, buffer &c, std::int64_t ldc) + + her2k supports the following precisions: + + + .. list-table:: + :header-rows: 1 + + * - T + - T_real + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-6DDD93FE-028E-400C-BBD0-CA13132FAC35 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The her2k routines perform a rank-2k update of an ``n`` x ``n`` + Hermitian matrix ``C`` by general matrices ``A`` and ``B``. If + ``trans`` = ``transpose::nontrans``. The operation is defined as + + + + + + C <- alpha*A*B :sup:`H` + conjg(alpha)*B*A :sup:`H` + beta*C + + + where ``A`` is ``n`` x ``k`` and ``B`` is ``k`` x ``n``. + + + If ``trans`` = ``transpose::conjtrans``, the operation is defined as: + + + + + + C <- alpha*B*A :sup:`H` + conjg(alpha)*A*B :sup:`H` + beta*C + + + where ``A`` is ``k`` x ``n`` and ``B`` is ``n`` x ``k``. + + + In both cases: + + + ``alpha`` is a complex scalar and ``beta`` is a real scalar. + + + ``C`` is a Hermitian matrix and ``A, B`` are general matrices. + + + The inner dimension of both matrix multiplications is ``k``. + + +.. container:: section + :name: GUID-54538396-B04D-4A2A-8A7D-E503A6F815AD + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A``'s data is stored in its upper or lower + triangle. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies the operation to apply, as described above. Supported + operations are ``transpose::nontrans`` and + ``transpose::conjtrans``. + + + n + The number of rows and columns in ``C``. The value of ``n`` must + be at least zero. + + + k + The inner dimension of matrix multiplications. The value of ``k`` + must be at least equal to zero. + + + alpha + Complex scaling factor for the rank-2\ ``k`` update. + + + a + Buffer holding input matrix ``A``. If ``trans`` = + ``transpose::nontrans``, ``A`` is an ``n``-by-``k`` matrix so the + array ``a`` must have size at least ``lda``\ \*\ ``k``. Otherwise, + ``A`` is an ``k``-by-``n`` matrix so the array ``a`` must have + size at least ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``n`` if ``trans`` = + ``transpose::nontrans``, and at least ``k`` otherwise. Must be + positive. + + + beta + Real scaling factor for matrix ``C``. + + + b + Buffer holding input matrix ``B``. If ``trans`` = + ``transpose::nontrans``, ``B`` is an ``k``-by-``n`` matrix so the + array ``b`` must have size at least ``ldb``\ \*\ ``n``. Otherwise, + ``B`` is an ``n``-by-``k`` matrix so the array ``b`` must have + size at least ``ldb``\ \*\ ``k``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be at least ``k`` if ``trans`` = + ``transpose::nontrans``, and at least ``n`` otherwise. Must be + positive. + + + c + Buffer holding input/output matrix ``C``. Must have size at least + ``ldc``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``n``. + + +.. container:: section + :name: GUID-48D39D42-B29F-4428-A588-9058570B5D5E + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + Output buffer, overwritten by the updated ``C`` matrix. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/herk.rst b/docs/domains/blas/herk.rst new file mode 100644 index 000000000..03b510a1f --- /dev/null +++ b/docs/domains/blas/herk.rst @@ -0,0 +1,175 @@ +.. _herk: + +herk +==== + + +.. container:: + + + Performs a Hermitian rank-k update. + + + .. container:: section + :name: GUID-407B8203-A28D-468B-BA79-87FA865E75A2 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void herk(queue &exec_queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, T_real alpha, buffer &a, std::int64_t lda, T_real beta, buffer &c, std::int64_t ldc) + + herk supports the following precisions: + + + .. list-table:: + :header-rows: 1 + + * - T + - T_real + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-539B4E63-9CDF-4834-999A-4133CE5DE1E5 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The herk routines compute a rank-``k`` update of a Hermitian matrix + *C* by a general matrix ``A``. The operation is defined as: + + + + + + C <- alpha*op(A)*op(A) :sup:`H` + beta*C + + + where: + + + op(``X``) is one of op(``X``) = ``X`` or op(``X``) = ``X``\ :sup:`H`, + + + ``alpha`` and ``beta`` are real scalars, + + + ``C`` is a Hermitian matrix and ``A`` is a general matrix. + + + Here op(``A``) is ``n`` x ``k``, and ``C`` is ``n`` x ``n``. + + +.. container:: section + :name: GUID-7B880A06-4E53-4DE9-B0E6-D70673CF2638 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A``'s data is stored in its upper or lower + triangle. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. Supported operations are ``transpose::nontrans`` and + ``transpose::conjtrans``. + + + n + The number of rows and columns in ``C``.The value of ``n`` must be + at least zero. + + + k + Number of columns in op(``A``). + + + The value of ``k`` must be at least zero. + + + alpha + Real scaling factor for the rank-``k`` update. + + + a + Buffer holding input matrix ``A``. If ``trans`` = + ``transpose::nontrans``, ``A`` is an ``n``-by-``k`` matrix so the + array ``a`` must have size at least ``lda``\ \*\ ``k``. Otherwise, + ``A`` is an ``k``-by-``n`` matrix so the array ``a`` must have + size at least ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``n`` if ``A`` is not + transposed, and at least ``k`` if ``A`` is transposed. Must be + positive. + + + beta + Real scaling factor for matrix ``C``. + + + c + Buffer holding input/output matrix ``C``. Must have size at least + ``ldc``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``n``. + + +.. container:: section + :name: GUID-05309970-DEC8-4D87-90AA-958FC101E119 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + The output buffer, overwritten by + ``alpha``\ \*op(``A``)*op(``A``)\ :sup:`T` + ``beta``\ \*\ ``C``. + The imaginary parts of the diagonal elements are set to zero. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/hpmv.rst b/docs/domains/blas/hpmv.rst new file mode 100644 index 000000000..f71e4b392 --- /dev/null +++ b/docs/domains/blas/hpmv.rst @@ -0,0 +1,157 @@ +.. _hpmv: + +hpmv +==== + + +.. container:: + + + Computes a matrix-vector product using a Hermitian packed matrix. + + + .. container:: section + :name: GUID-C6E4A4A7-5CBE-46ED-A021-8FEAABAA2E93 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void hpmv(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &a, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``hpmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-A95C32C5-0371-429B-847C-4EE29FD9C480 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The hpmv routines compute a scalar-matrix-vector product and add the + result to a scalar-vector product, with a Hermitian packed matrix. + The operation is defined as + + + + + + y <- alpha*A*x + beta*y + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``n``-by-``n`` Hermitian matrix supplied in packed form, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether *A* is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``\ +1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + The imaginary parts of the diagonal elements need not be set and + are assumed to be zero. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + beta + Scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-416B82CD-C5B8-472A-8347-04997EA6D6E6 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/hpr.rst b/docs/domains/blas/hpr.rst new file mode 100644 index 000000000..1c0f38a0c --- /dev/null +++ b/docs/domains/blas/hpr.rst @@ -0,0 +1,147 @@ +.. _hpr: + +hpr +=== + + +.. container:: + + + Computes a rank-1 update of a Hermitian packed matrix. + + + .. container:: section + :name: GUID-61DC4DBA-9357-4129-B8A3-931E2E7335D4 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void hpr(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &a) + + ``hpr`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-02B8128C-02CE-4D5C-BE5D-DFD088C90475 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The hpr routines compute a scalar-vector-vector product and add the + result to a Hermitian packed matrix. The operation is defined as + + + + + + A <- alpha*x*x :sup:`H` + A + + + where: + + + ``alpha`` is scalar, + + + ``A`` is an ``n``-by-``n`` Hermitian matrix, supplied in packed form, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``-1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + The imaginary part of the diagonal elements need not be set and + are assumed to be zero + + +.. container:: section + :name: GUID-7261182A-450B-46F5-8C61-7133597D3530 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated upper triangularpart of the Hermitian + matrix ``A`` if ``upper_lower =upper``, or the updated lower + triangular part of theHermitian matrix ``A`` if + ``upper_lower =lower``. + + + The imaginary parts of the diagonal elements are set tozero. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/hpr2.rst b/docs/domains/blas/hpr2.rst new file mode 100644 index 000000000..bfe83d4b6 --- /dev/null +++ b/docs/domains/blas/hpr2.rst @@ -0,0 +1,163 @@ +.. _hpr2: + +hpr2 +==== + + +.. container:: + + + Performs a rank-2 update of a Hermitian packed matrix. + + + .. container:: section + :name: GUID-9F8EB534-6520-4470-85AC-6AD8F2467AD4 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void hpr2(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a) + + ``hpr2`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-16FE1EDC-1A72-4BAB-8AFF-C316C4CE5838 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The hpr2 routines compute two scalar-vector-vector products and add + them to a Hermitian packed matrix. The operation is defined as + + + + + + A <- alpha*x*y :sup:`H` + conjg(alpha)*y*x :sup:`H` + A + + + where: + + + ``alpha`` is a scalar, + + + ``A`` is an ``n``-by-``n`` Hermitian matrix, supplied in packed form, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether *A* is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``-1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + The imaginary parts of the diagonal elements need not be set and + are assumed to be zero. + + +.. container:: section + :name: GUID-9A77A2E0-F610-44EE-A3EE-81327B90A3FD + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + **sycl:** +   + + + + a + Buffer holding the updated upper triangularpart of the Hermitian + matrix ``A`` if ``upper_lower =upper``, or the updated lower + triangular part of theHermitian matrix ``A`` if + ``upper_lower =lower``. + + + The imaginary parts of the diagonal elements are set tozero. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/iamax.rst b/docs/domains/blas/iamax.rst new file mode 100644 index 000000000..1678b7279 --- /dev/null +++ b/docs/domains/blas/iamax.rst @@ -0,0 +1,132 @@ +.. _iamax: + +iamax +===== + + +.. container:: + + + Finds the index of the element with the largest absolute value in a + vector. + + + .. container:: section + :name: GUID-D1ABF76D-DB39-4C23-A217-EA2C7C6D1325 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void iamax(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &result) + + iamax supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std:complex`` + + + + +.. container:: section + :name: GUID-822D7950-256E-406D-9305-61F761080E69 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The iamax routines return an index ``i``\ such that ``x``\ [``i``] + has the maximum absolute value of all elements in vector ``x`` (real + variants), or such that ``|Re(x[i])| + |Im(x[i])|`` is maximal + (complex variants). + + + .. container:: Note + + + .. rubric:: Note + :name: note + :class: NoteTipHead + + + The index is zero-based. + + + If either ``n`` or ``incx`` are not positive, the routine returns + ``0``. + + + If more than one vector element is found with the same largest + absolute value, the index of the first one encountered is returned. + + + If the vector contains ``NaN`` values, then the routine returns the + index of the first ``NaN``. + + +.. container:: section + :name: GUID-CE43FE84-2066-4095-BB7E-0691CD045443 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + The number of elements in vector ``x``. + + + x + The buffer that holds the input vector ``x``. The buffer must be + of size at least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incx + The stride of vector ``x``. + + +.. container:: section + :name: ARGUMENTS_EC9F05BE9B09443F8BC59207D5EA40F1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + The buffer where the zero-based index ``i`` of the maximal element + is stored. + + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/iamin.rst b/docs/domains/blas/iamin.rst new file mode 100644 index 000000000..ca5ea696f --- /dev/null +++ b/docs/domains/blas/iamin.rst @@ -0,0 +1,130 @@ +.. _iamin: + +iamin +===== + + +.. container:: + + + Finds the index of the element with the smallest absolute value. + + + .. container:: section + :name: GUID-5D077B60-17B5-4961-AFF7-20D78BFB2A07 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void iamin(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &result) + + ``iamin`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-A820CE7B-E983-4D8F-A73A-753FD95BD507 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The iamin routines return an index ``i`` such that ``x``\ [``i``] has + the minimum absolute value of all elements in vector ``x`` (real + variants), or such that \|Re(``x``\ [``i``])\| + + \|Im(``x``\ [``i``])\| is maximal (complex variants). + + + .. container:: Note + + + .. rubric:: Note + :name: note + :class: NoteTipHead + + + The index is zero-based. + + + If either ``n`` or ``incx`` are not positive, the routine returns + ``0``. + + + If more than one vector element is found with the same smallest + absolute value, the index of the first one encountered is returned. + + + If the vector contains ``NaN`` values, then the routine returns the + index of the first ``NaN``. + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + +.. container:: section + :name: GUID-2B160DEB-ADBB-4044-8078-4B613A0DA4E1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + Buffer where the zero-based index ``i`` of the minimum element + will be stored. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/nrm2.rst b/docs/domains/blas/nrm2.rst new file mode 100644 index 000000000..dfbf2265c --- /dev/null +++ b/docs/domains/blas/nrm2.rst @@ -0,0 +1,121 @@ +.. _nrm2: + +nrm2 +==== + + +.. container:: + + + Computes the Euclidean norm of a vector. + + + .. container:: section + :name: GUID-F55A15D5-CCDA-4C44-B86F-C9A5FB36725E + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void nrm2(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &result) + + ``nrm2`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + - T_res + * - ``float`` + - ``float`` + * - ``double`` + - ``double`` + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-2BF2C965-5A8C-47F1-9C73-FB0E485CE32A + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The nrm2 routines computes Euclidean norm of a vector + + + + + + result = ||x||, + + + where: + + + ``x`` is a vector of ``n`` elements. + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + +.. container:: section + :name: GUID-2B160DEB-ADBB-4044-8078-4B613A0DA4E1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + Buffer where the Euclidean norm of the vector ``x`` will be + stored. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/rot.rst b/docs/domains/blas/rot.rst new file mode 100644 index 000000000..2dc20bdce --- /dev/null +++ b/docs/domains/blas/rot.rst @@ -0,0 +1,139 @@ +.. _rot: + +rot +=== + + +.. container:: + + + Performs rotation of points in the plane. + + + .. container:: section + :name: GUID-9DD44991-6A55-49EE-BD0C-F13406FFBE52 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void rot(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, T_scalar c, T_scalar s) + + ``rot`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + - T_scalar + * - ``float`` + - ``float`` + * - ``double`` + - ``double`` + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-8B7F46D1-5047-4D4C-AF66-F0A3E4AC2BA5 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + Given two vectors ``x`` and ``y`` of ``n`` elements, the rot routines + compute four scalar-vector products and update the input vectors with + the sum of two of these scalar-vector products as follow: + + + x <- c*x + s*y + + y <- c*y - s*x + + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector ``y``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + + c + Scaling factor. + + + s + Scaling factor. + + +.. container:: section + :name: GUID-2B160DEB-ADBB-4044-8078-4B613A0DA4E1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding updated buffer ``x``. + + + y + Buffer holding updated buffer ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/rotg.rst b/docs/domains/blas/rotg.rst new file mode 100644 index 000000000..3110a60a8 --- /dev/null +++ b/docs/domains/blas/rotg.rst @@ -0,0 +1,125 @@ +.. _rotg: + +rotg +==== + + +.. container:: + + + Computes the parameters for a Givens rotation. + + + .. container:: section + :name: GUID-E4B6E693-AC8C-4BB3-A197-3EB9E905B925 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void rotg(queue &exec_queue, buffer &a, buffer &b, buffer &c, buffer &s) + + ``rotg`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + - T_res + * - ``float`` + - ``float`` + * - ``double`` + - ``double`` + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-5614B81D-C736-4714-88AB-29B38F9B3589 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + Given the Cartesian coordinates ``(a, b)`` of a point, the rotg + routines return the parameters ``c``, ``s``, ``r``, and ``z`` + associated with the Givens rotation. The parameters ``c`` and ``s`` + define a unitary matrix such that: + + + The parameter ``z`` is defined such that if \|\ ``a``\ \| > + \|\ ``b``\ \|, ``z`` is ``s``; otherwise if ``c`` is not 0 ``z`` is + 1/``c``; otherwise ``z`` is 1. + + +.. container:: section + :name: GUID-C2003328-15AA-4DF0-A417-40BECCA7DEA3 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed + + + a + Buffer holding the ``x``-coordinate of the point. + + + b + Buffer holding the ``y``-coordinate of the point. + + +.. container:: section + :name: GUID-3B7937E3-2DF7-49A3-8F1E-2C9406BB4E88 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the parameter ``r`` associated with the Givens + rotation. + + + b + Buffer holding the parameter ``z`` associated with the Givens + rotation. + + + c + Buffer holding the parameter ``c`` associated with the Givens + rotation. + + + s + Buffer holding the parameter ``s`` associated with the Givens + rotation. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/rotm.rst b/docs/domains/blas/rotm.rst new file mode 100644 index 000000000..4d025c4ea --- /dev/null +++ b/docs/domains/blas/rotm.rst @@ -0,0 +1,170 @@ +.. _rotm: + +rotm +==== + + +.. container:: + + + Performs modified Givens rotation of points in the plane. + + + .. container:: section + :name: GUID-F8F2E2EB-1704-454D-BE45-C055D6F4E7D6 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void rotm(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer ¶m) + + ``rotm`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-856650C6-2998-4452-A34A-DF6CB801087D + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + Given two vectors ``x`` and ``y``, each vector element of these + vectors is replaced as follows: + + + | + | |image0| + + + for ``i`` from 1 to ``n``, where ``H`` is a modified Givens + transformation matrix. + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + + param + Buffer holding an array of size 5. The elements of the ``param`` + array are: + + + ``param``\ [0] contains a switch, ``flag``, + + + ``param``\ [1-4] contain *h\ 11*,\ *h\ 21*, *h\ 12*,\ *h\ 22* + respectively, the components ofthe modified Givens transformation + matrix ``H``. + + + Depending on the values of ``flag``, thecomponents of ``H`` are + set as follows: + + + | ``flag =``\ ``-1.0``: + | |image1| + + + | ``flag =``\ ``0.0``: + | |image2| + + + | ``flag =``\ ``1.0``: + | |image3| + + + | ``flag =``\ ``-2.0``: + | |image4| + + + In the last three cases, the matrix entries of 1.0, -1.0, 0.0 are + assumed based on the value of ``flag`` and are not required to be + set in the ``param`` vector. + + +.. container:: section + :name: GUID-062D805B-68FF-41F6-8D9A-329C92A77EA3 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding updated buffer ``x``. + + + y + Buffer holding updated buffer ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee1.png +.. |image1| image:: ../equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee2.png +.. |image2| image:: ../equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee3.png +.. |image3| image:: ../equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee4.png +.. |image4| image:: ../equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee5.png + diff --git a/docs/domains/blas/rotmg.rst b/docs/domains/blas/rotmg.rst new file mode 100644 index 000000000..64d6543ea --- /dev/null +++ b/docs/domains/blas/rotmg.rst @@ -0,0 +1,165 @@ +.. _rotmg: + +rotmg +===== + + +.. container:: + + + Computes the parameters for a modified Givens rotation. + + + .. container:: section + :name: GUID-DF41021D-C145-495B-A717-45FB5F36E676 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void rotmg(queue &exec_queue, buffer &d1, buffer &d2, buffer &x1, buffer &y1, buffer ¶m) + + ``rotmg`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-5525F11C-A739-487E-A7CC-6886A088035D + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + Given Cartesian coordinates (``x``\ :sub:`1`, ``y``\ :sub:`1`) of an + input vector, the rotmg routines compute the components of a modified + Givens transformation matrix ``H`` that zeros the ``y``-component of + the resulting vector: + + + | + | |image0| + + +.. container:: section + :name: GUID-21946B3A-A859-4293-8EE7-965328AA6717 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + d1 + Buffer holding the scaling factor for the ``x``-coordinate of the + input vector. + + + d2 + Buffer holding the scaling factor for the ``y``-coordinate of the + input vector. + + + x1 + Buffer holding the ``x``-coordinate of the input vector. + + + y1 + Scalar specifying the ``y``-coordinate of the input vector. + + +.. container:: section + :name: GUID-1C0481DB-BB35-4DB7-941F-649EDAA77C6F + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + d1 + Buffer holding the first diagonal element of the updated matrix. + + + d2 + Buffer holding the second diagonal element of the updated matrix. + + + x1 + Buffer holding the *x*-coordinate of the rotated vector before + scaling + + + param + Buffer holding an array of size 5. + + + The elements of the ``param`` array are: + + + ``param[0]`` contains a switch, ``flag``. the other array elements + ``param[1-4]`` contain the components of the array ``H``: + ``h``\ :sub:`11`, ``h``\ :sub:`21`, ``h``\ :sub:`12`, and + ``h``\ :sub:`22`, respectively. + + + Depending on the values of ``flag``, the components of ``H`` are + set as follows: + + + | ``flag =``\ ``-1.0``: + | |image1| + + + | ``flag =``\ ``0.0``: + | |image2| + + + | ``flag =``\ ``1.0``: + | |image3| + + + | ``flag =``\ ``-2.0``: + | |image4| + + + In the last three cases, the matrix entries of 1.0, -1.0, and 0.0 + are assumed based on the value of ``flag`` and are not required to + be set in the ``param`` vector. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee1.png +.. |image1| image:: ../equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee2.png +.. |image2| image:: ../equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee3.png +.. |image3| image:: ../equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee4.png +.. |image4| image:: ../equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee5.png + diff --git a/docs/domains/blas/sbmv.rst b/docs/domains/blas/sbmv.rst new file mode 100644 index 000000000..b28b9b027 --- /dev/null +++ b/docs/domains/blas/sbmv.rst @@ -0,0 +1,164 @@ +.. _sbmv: + +sbmv +==== + + +.. container:: + + + Computes a matrix-vector product with a symmetric band matrix. + + + .. container:: section + :name: GUID-BEDE7E82-C168-498D-BF65-085BBCEF9A27 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void sbmv(queue &exec_queue, uplo upper_lower, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``sbmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-4F227157-1724-4D1F-AFAB-58C722CA8D08 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The sbmv routines compute a scalar-matrix-vector product and add the + result to a scalar-vector product, with a symmetric band matrix. The + operation is defined as + + + + + + y <- alpha*A*x + beta*y + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``n``-by-``n`` symmetric matrix with ``k`` + super-diagonals, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + k + Number of super-diagonals of the matrix ``A``. Must be at least + zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least (``k`` + 1), + and positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + beta + Scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-ABBEA4DA-7B4C-489A-8063-BDC09FBB1ADD + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/scal.rst b/docs/domains/blas/scal.rst new file mode 100644 index 000000000..97075eac6 --- /dev/null +++ b/docs/domains/blas/scal.rst @@ -0,0 +1,131 @@ +.. _scal: + +scal +==== + + +.. container:: + + + Computes the product of a vector by a scalar. + + + .. container:: section + :name: GUID-178A4C6A-3BA5-40F7-A3D6-4B6590B75EB4 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void scal(queue &exec_queue, std::int64_t n, T_scalar alpha, buffer &x, std::int64_t incx) + + ``scal`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + - T_scalar + * - ``float`` + - ``float`` + * - ``double`` + - ``double`` + * - ``std::complex`` + - ``std::complex`` + * - ``std::complex`` + - ``std::complex`` + * - ``std::complex`` + - ``float`` + * - ``std::complex`` + - ``double`` + + + + +.. container:: section + :name: GUID-8DDCA613-2750-43D0-A89B-13866F2DDE8C + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The scal routines computes a scalar-vector product: + + + + + + x <- alpha*x + + + where: + + + ``x`` is a vector of ``n`` elements, + + + ``alpha`` is a scalar. + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + alpha + Specifies the scalar ``alpha``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + +.. container:: section + :name: GUID-B36EBB3E-C79B-49F8-9F47-7B19BD6BE105 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding updated buffer ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/sdsdot.rst b/docs/domains/blas/sdsdot.rst new file mode 100644 index 000000000..11414fd5b --- /dev/null +++ b/docs/domains/blas/sdsdot.rst @@ -0,0 +1,93 @@ +.. _sdsdot: + +sdsdot +====== + + +.. container:: + + + Computes a vector-vector dot product with double precision. + + + .. container:: section + :name: GUID-2DDFDC38-65FA-40F5-AACB-8E383623EF4A + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void sdsdot(queue &exec_queue, std::int64_t n, float sb, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &result) + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The sdsdot routines perform a dot product between two vectors with + double precision: + + + |image0| + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vectors ``x`` and ``y``. + + + sb + Single precision scalar to be added to the dot product. + + + x + Buffer holding input vector ``x``. The buffer must be of size + at least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and + Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector ``y``. The buffer must be of size + at least (1 + (``n`` - 1)*abs(``incxy``)). See `Matrix and + Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + result + Buffer where the result (a scalar) will be stored. If ``n`` < 0 + the result is ``sb``. + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. |image0| image:: ../equations/GUID-9DB212E1-03E2-430C-8B1F-8F5CBD4F2ee1.png + :class: img-middle + diff --git a/docs/domains/blas/spmv.rst b/docs/domains/blas/spmv.rst new file mode 100644 index 000000000..0b1690df5 --- /dev/null +++ b/docs/domains/blas/spmv.rst @@ -0,0 +1,153 @@ +.. _spmv: + +spmv +==== + + +.. container:: + + + Computes a matrix-vector product with a symmetric packed matrix. + + + .. container:: section + :name: GUID-BCC82B03-92EB-4D73-B69C-8AE8646FBEAC + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void spmv(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &a, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``spmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-D27BBFFF-79F4-4236-96A6-B305FA1858B0 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The spmv routines compute a scalar-matrix-vector product and add the + result to a scalar-vector product, with a symmetric packed matrix. + The operation is defined as + + + + + + y <- alpha*A*x + beta*y + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``n``-by-``n`` symmetric matrix, supplied in packed form. + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``\ +1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + beta + Scaling factor for vector ``y``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-23FF1F5C-5560-40B6-807D-B6352FA320D6 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/spr.rst b/docs/domains/blas/spr.rst new file mode 100644 index 000000000..0112706b6 --- /dev/null +++ b/docs/domains/blas/spr.rst @@ -0,0 +1,145 @@ +.. _spr: + +spr +=== + + +.. container:: + + + Performs a rank-1 update of a symmetric packed matrix. + + + .. container:: section + :name: GUID-34904813-AFD9-4349-9DAC-A7221FBE9F97 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void spr(queue &exec_queue, uplo upper_lower, std::std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &a) + + ``spr`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-E387B33A-CA59-45D8-BB01-31DF76C82A0D + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The spr routines compute a scalar-vector-vector product and add the + result to a symmetric packed matrix. The operation is defined as + + + + + + A <- alpha*x*x :sup:`T` + A + + + where: + + + ``alpha`` is scalar, + + + ``A`` is an ``n``-by-``n`` symmetric matrix, supplied in packed form, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``-n))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + +.. container:: section + :name: GUID-9FBC2F3B-EB8F-4733-ABBA-08D5685A761B + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + **sycl:** +   + + + + a + Buffer holding the updated upper triangularpart of the symmetric + matrix ``A`` if ``upper_lower =upper``, or the updated lower + triangular part of thesymmetric matrix ``A`` if + ``upper_lower =lower``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/spr2.rst b/docs/domains/blas/spr2.rst new file mode 100644 index 000000000..ca78f30ac --- /dev/null +++ b/docs/domains/blas/spr2.rst @@ -0,0 +1,156 @@ +.. _spr2: + +spr2 +==== + + +.. container:: + + + Computes a rank-2 update of a symmetric packed matrix. + + + .. container:: section + :name: GUID-44B72132-1EC0-41FA-9189-4596CFD651B0 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void spr2(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a) + + ``spr`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-3AF7EB4D-B3FE-4C0A-B7A0-6E286D4C642F + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The spr2 routines compute two scalar-vector-vector products and add + them to a symmetric packed matrix. The operation is defined as + + + + + + A <- alpha*x*y :sup:`T` + alpha*y*x :sup:`T` + A + + + where: + + + ``alpha`` is scalar, + + + ``A`` is an ``n``-by-``n`` symmetric matrix, supplied in packed form, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``-1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + +.. container:: section + :name: GUID-9796BA93-31FB-40B9-B139-219905913736 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + **sycl:** +   + + + + a + Buffer holding the updated upper triangularpart of the symmetric + matrix ``A`` if ``upper_lower =upper`` or the updated lower + triangular part of thesymmetric matrix ``A`` if + ``upper_lower =lower``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/swap.rst b/docs/domains/blas/swap.rst new file mode 100644 index 000000000..3d4542779 --- /dev/null +++ b/docs/domains/blas/swap.rst @@ -0,0 +1,128 @@ +.. _swap: + +swap +==== + + +.. container:: + + + Swaps a vector with another vector. + + + .. container:: section + :name: GUID-F0DF0055-DF25-4EC7-8FF2-48D4FA91E42E + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void swap(queue &exec_queue, std::int64_t n, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy) + + swap supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-FE88C4B7-4C74-41F8-94DE-E62888DD3BA4 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + Given two vectors of ``n`` elements, ``x`` and ``y``, the swap + routines return vectors ``y`` and ``x`` swapped, each replacing the + other. + + + + + + y <- x, x <- y + + +.. container:: section + :name: GUID-A615800D-734E-4997-BB91-1C76AEEE9EC2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + n + Number of elements in vector ``x``. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector x. + + + y + Buffer holding input vector ``y``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incy + Stride of vector y. + + +.. container:: section + :name: GUID-106AC665-DCBA-40ED-8779-0D9017064855 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding updated buffer ``x``, that is, the input vector + ``y``. + + + y + Buffer holding updated buffer ``y``, that is, the input vector + ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-1-routines` + + + +.. container:: + diff --git a/docs/domains/blas/symm.rst b/docs/domains/blas/symm.rst new file mode 100644 index 000000000..c14d9d2bf --- /dev/null +++ b/docs/domains/blas/symm.rst @@ -0,0 +1,206 @@ +.. _symm: + +symm +==== + + +.. container:: + + + Computes a matrix-matrix product where one input matrix is symmetric + and one matrix is general. + + + .. container:: section + :name: GUID-BFE36A6B-941E-4B49-AB0E-CFB687B1AD64 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void symm(queue &exec_queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, T beta, buffer &c, std::int64_t ldc) + + symm supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-E8FE37B0-C527-4AA6-B57F-AE3F4843F23A + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The symm routines compute a scalar-matrix-matrix product and add the + result to a scalar-matrix product, where one of the matrices in the + multiplication is symmetric. The argument ``left_right`` determines + if the symmetric matrix, ``A``, is on the left of the multiplication + (``left_right`` = ``side::left``) or on the right (``left_right`` = + ``side::right``). Depending on ``left_right``, the operation is + defined as + + + + + + C <- alpha*A*B + beta*C, + + + or + + + + + + C <- alpha*B*A + beta*C, + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is a symmetric matrix, either ``m``-by-``m`` or ``n``-by-``n``, + + + ``B`` and ``C`` are ``m``-by-``n`` matrices. + + +.. container:: section + :name: GUID-70716375-C54E-4AA6-94DC-65AF79D46BB2 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + left_right + Specifies whether ``A`` is on the left side of the multiplication + (``side::left``) or on the right side (``side::right``). See + :ref:`onemkl_datatypes` for more + details. + + + upper_lower + Specifies whether *A*'s data is stored in its upper or lower + triangle. See + :ref:`onemkl_datatypes` for more + details. + + + m + Number of rows of ``B`` and ``C``. The value of ``m`` must be at + least zero. + + + n + Number of columns of ``B`` and ``C``. The value of ``n`` must be + at least zero. + + + alpha + Scaling factor for the matrix-matrix product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``m`` if ``A`` is on the left of the multiplication, + or ``lda``\ \*\ ``n`` if ``A`` is on the right. See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + lda + Leading dimension of ``A``. Must be at least ``m`` if ``A`` is on + the left of the multiplication, or at least ``n`` if ``A`` is on + the right. Must be positive. + + + b + Buffer holding input matrix ``B``. Must have size at least + ``ldb``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be positive and at least ``m``. + + + beta + Scaling factor for matrix ``C``. + + + c + Buffer holding input/output matrix ``C``. Must have size at least + ``ldc``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``m``. + + +.. container:: section + :name: GUID-DD569858-5D3C-4565-8BAB-FE548427DCF2 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + Output buffer, overwritten by ``alpha``\ \*\ ``A``\ \*\ ``B`` + + ``beta``\ \*\ ``C`` (``left_right`` = ``side::left``) or + ``alpha``\ \*\ ``B``\ \*\ ``A`` + ``beta``\ \*\ ``C`` + (``left_right`` = ``side::right``). + + +.. container:: section + :name: EXAMPLE_5EF48B8A07D849EA84A74FE22F0D5B24 + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``beta`` = 0, matrix ``C`` does not need to be initialized before + calling ``symm``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/symv.rst b/docs/domains/blas/symv.rst new file mode 100644 index 000000000..8d59ed90a --- /dev/null +++ b/docs/domains/blas/symv.rst @@ -0,0 +1,154 @@ +.. _symv: + +symv +==== + + +.. container:: + + + Computes a matrix-vector product for a symmetric matrix. + + + .. container:: section + :name: GUID-1E9C9EA9-0366-420E-A704-AB605C8ED92A + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void symv(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx, T beta, buffer &y, std::int64_t incy) + + ``symv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-DE8D8321-D53D-4226-A940-CDE0E720EC95 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The symv routines routines compute a scalar-matrix-vector product and + add the result to a scalar-vector product, with a symmetric matrix. + The operation is defined as + + + + + + y <- alpha*A*x + beta*y + + + where: + + + ``alpha`` and ``beta`` are scalars, + + + ``A`` is an ``n``-by-``n`` symmetric matrix, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``m``, and + positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + +.. container:: section + :name: GUID-E16C8443-A2A4-483C-9D46-FF428E80FEB0 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + y + Buffer holding the updated vector ``y``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/syr.rst b/docs/domains/blas/syr.rst new file mode 100644 index 000000000..e3ff12e6b --- /dev/null +++ b/docs/domains/blas/syr.rst @@ -0,0 +1,146 @@ +.. _syr: + +syr +=== + + +.. container:: + + + Computes a rank-1 update of a symmetric matrix. + + + .. container:: section + :name: GUID-E620D36F-6B4E-40A6-8BDA-3D625DEF55A8 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void syr(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &a, std::int64_t lda) + + ``syr`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-E154DE4B-4559-4471-B92B-46AF8777AC97 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The syr routines compute a scalar-vector-vector product add them and + add the result to a matrix, with a symmetric matrix. The operation is + defined as + + + + + + A <- alpha*x*x :sup:`T` + A + + + where: + + + ``alpha`` is scalar, + + + ``A`` is an ``n``-by-``n`` symmetric matrix, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``n``, and + positive. + + +.. container:: section + :name: GUID-C03D1215-FD77-4AD8-8FA2-C48A5D8B938C + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated upper triangularpart of the symmetric + matrix ``A`` if ``upper_lower =upper`` or the updated lower + triangular part of thesymmetric matrix ``A`` if + ``upper_lower =lower``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/syr2.rst b/docs/domains/blas/syr2.rst new file mode 100644 index 000000000..6459801cf --- /dev/null +++ b/docs/domains/blas/syr2.rst @@ -0,0 +1,157 @@ +.. _syr2: + +syr2 +==== + + +.. container:: + + + Computes a rank-2 update of a symmetric matrix. + + + .. container:: section + :name: GUID-580F2222-D47E-43A3-B9A2-037F353825D5 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void syr2(queue &exec_queue, uplo upper_lower, std::int64_t n, T alpha, buffer &x, std::int64_t incx, buffer &y, std::int64_t incy, buffer &a, std::int64_t lda) + + ``syr2`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + + + + +.. container:: section + :name: GUID-CDA05459-F2FE-4933-A552-D6E52EC46D13 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The syr2 routines compute two scalar-vector-vector product add them + and add the result to a matrix, with a symmetric matrix. The + operation is defined as + + + + + + A <- alpha*x*y :sup:`T` + alpha*y*x :sup:`T` + A + + + where: + + + ``alpha`` is a scalar, + + + ``A`` is an ``n``-by-``n`` symmetric matrix, + + + ``x`` and ``y`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + n + Number of columns of ``A``. Must be at least zero. + + + alpha + Scaling factor for the matrix-vector product. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + + y + Buffer holding input/output vector ``y``. The buffer must be of + size at least (1 + (``n`` - 1)*abs(``incy``)). See `Matrix and + Vector Storage <../matrix-storage.html>`__ + for more details. + + + incy + Stride of vector ``y``. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``n``, and + positive. + + +.. container:: section + :name: GUID-6992A39F-8AB7-42D9-B126-4F8ECF9C1ECE + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + a + Buffer holding the updated upper triangularpart of the symmetric + matrix ``A`` if ``upper_lower =upper``, or the updated lower + triangular part of thesymmetric matrix ``A`` if + ``upper_lower =lower``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/syr2k.rst b/docs/domains/blas/syr2k.rst new file mode 100644 index 000000000..e5687c856 --- /dev/null +++ b/docs/domains/blas/syr2k.rst @@ -0,0 +1,196 @@ +.. _syr2k: + +syr2k +===== + + +.. container:: + + + Performs a symmetric rank-2k update. + + + .. container:: section + :name: GUID-EED2648B-6435-4DD1-AC36-21039DFC61DD + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void syr2k(queue &exec_queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb, T beta, buffer &c, std::int64_t ldc) + + syr2k supports the following precisions: + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-1FB46B8F-1B13-4A6B-A3A5-0A5B34049068 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The syr2k routines perform a rank-2k update of an ``n`` x ``n`` + symmetric matrix ``C`` by general matrices ``A`` and ``B``. If + ``trans`` = ``transpose::nontrans``, the operation is defined as: + + + + + + C <- alpha*(A*B :sup:`T` + B*A :sup:`T`) + beta*C + + + where ``A`` is ``n`` x ``k`` and ``B`` is ``k`` x ``n``. + + + If ``trans`` = ``transpose::trans``, the operationis defined as: + + + + + + C <- alpha*(A :sup:`T`*B + B :sup:`T`*A) + beta*C + + + where ``A`` is ``k`` x ``n`` and ``B`` is ``n`` x ``k``. + + + In both cases: + + + ``alpha`` and ``beta`` are scalars, + + + ``C`` is a symmetric matrix and ``A``,\ ``B`` are general matrices, + + + The inner dimension of both matrix multiplications is ``k``. + + +.. container:: section + :name: GUID-3EBEFBDD-93AF-4376-9BA2-A7042179BF13 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A``'s data is stored in its upper or lower + triangle. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies the operation to apply, as described above. Conjugation + is never performed, even if ``trans`` = ``transpose::conjtrans``. + + + n + Number of rows and columns in ``C``.The value of ``n`` must be at + least zero. + + + k + Inner dimension of matrix multiplications.The value of ``k`` must + be at least zero. + + + alpha + Scaling factor for the rank-2\ ``k`` update. + + + a + Buffer holding input matrix ``A``. If ``A`` is not transposed, + ``A`` is an ``m``-by-``k`` matrix so the array ``a`` must have + size at least ``lda``\ \*\ ``k``. If ``A`` is transposed, ``A`` is + an ``k``-by-``m`` matrix so the array ``a`` must have size at + least ``lda``\ \*\ ``m``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``n`` if ``trans`` = + ``transpose::nontrans``, and at least ``k`` otherwise. Must be + positive. + + + b + Buffer holding input matrix ``B``. If ``trans`` = + ``transpose::nontrans``, ``B`` is an ``k``-by-``n`` matrix so the + array ``b`` must have size at least ``ldb``\ \*\ ``n``. Otherwise, + ``B`` is an ``n``-by-``k`` matrix so the array ``b`` must have + size at least ``ldb``\ \*\ ``k``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be at least ``k`` if ``trans`` = + ``transpose::nontrans``, and at least ``n`` otherwise. Must be + positive. + + + beta + Scaling factor for matrix ``C``. + + + c + Buffer holding input/output matrix ``C``. Must have size at least + ``ldc``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details + + + ldc + Leading dimension of ``C``. Must be positive and at least ``n``. + + +.. container:: section + :name: GUID-5779F783-54BC-4887-9CBB-96B8EC9F00E9 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + Output buffer, overwritten by the updated C matrix. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/syrk.rst b/docs/domains/blas/syrk.rst new file mode 100644 index 000000000..d097db2dd --- /dev/null +++ b/docs/domains/blas/syrk.rst @@ -0,0 +1,172 @@ +.. _syrk: + +syrk +==== + + +.. container:: + + + Performs a symmetric rank-k update. + + + .. container:: section + :name: GUID-F8123F9B-A182-4BDB-A1A3-90FEC4F56231 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void syrk(queue &exec_queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, T alpha, buffer &a, std::int64_t lda, T beta, buffer &c, std::int64_t ldc) + + syrk supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-8E133139-EE58-44B8-A507-2263BDD1399B + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The syrk routines perform a rank-k update of a symmetric matrix ``C`` + by a general matrix ``A``. The operation is defined as: + + + + + + C <- alpha*op(A)*op(A)T + beta*C + + + where: + + + op(``X``) is one of op(``X``) = ``X`` or op(``X``) = ``X``\ :sup:`T` + , + + + ``alpha`` and ``beta`` are scalars, + + + ``C`` is a symmetric matrix and ``A``\ is a general matrix. + + + Here op(``A``) is ``n``-by-``k``, and ``C`` is ``n``-by-``n``. + + +.. container:: section + :name: GUID-96D007CC-23F0-46FA-9085-6DBFC5BB30E6 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A``'s data is stored in its upper or lower + triangle. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A`` + (See + :ref:`onemkl_datatypes` for more + details). Conjugation is never performed, even if ``trans`` = + ``transpose::conjtrans``. + + + n + Number of rows and columns in ``C``. The value of ``n`` must be at + least zero. + + + k + Number of columns in op(``A``).The value of ``k`` must be at least + zero. + + + alpha + Scaling factor for the rank-``k`` update. + + + a + Buffer holding input matrix ``A``. If ``trans`` = + ``transpose::nontrans``, ``A`` is an ``n``-by-``k`` matrix so the + array ``a`` must have size at least ``lda``\ \*\ ``k``. Otherwise, + ``A`` is an ``k``-by-``n`` matrix so the array ``a`` must have + size at least ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``n`` if ``A`` is not + transposed, and at least ``k`` if ``A`` is transposed. Must be + positive. + + + beta + Scaling factor for matrix ``C``. + + + c + Buffer holding input/output matrix ``C``. Must have size at least + ``ldc``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``n``. + + +.. container:: section + :name: GUID-E14CE68E-2E28-48BB-8FD7-B84A21563BDA + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + c + Output buffer, overwritten by + ``alpha``\ \*op(``A``)*op(``A``)\ :sup:`T` + ``beta``\ \*\ ``C``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/tbmv.rst b/docs/domains/blas/tbmv.rst new file mode 100644 index 000000000..eb5f7acf0 --- /dev/null +++ b/docs/domains/blas/tbmv.rst @@ -0,0 +1,160 @@ +.. _tbmv: + +tbmv +==== + + +.. container:: + + + Computes a matrix-vector product using a triangular band matrix. + + + .. container:: section + :name: GUID-BAC06253-0516-4F7F-97E6-C4CBA2DBB1A2 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void tbmv(queue &exec_queue, uplo upper_lower, transpose trans, diag unit_nonunit, std::int64_t n, std::int64_t k, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx) + + ``tbmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-4279E883-09A1-48F0-B9DA-8A1E86886B17 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The tbmv routines compute a matrix-vector product with a triangular + band matrix. The operation is defined as + + + + + + x <- op(A)*x + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``A`` is an ``n``-by-``n`` unit or non-unit, upper or lower + triangular band matrix, with (``k`` + 1) diagonals, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_nonunit + Specifies whether the matrix ``A`` is unit triangular or not. See + :ref:`onemkl_datatypes` + for more details. + + + n + Numbers of rows and columns of ``A``. Must be at least zero. + + + k + Number of sub/super-diagonals of the matrix ``A``. Must be at + least zero. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least (``k`` + 1), + and positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + +.. container:: section + :name: GUID-0B96A584-2EC7-484C-9FB0-C632053F0461 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding the updated vector ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/tbsv.rst b/docs/domains/blas/tbsv.rst new file mode 100644 index 000000000..73aab67bd --- /dev/null +++ b/docs/domains/blas/tbsv.rst @@ -0,0 +1,162 @@ +.. _tbsv: + +tbsv +==== + + +.. container:: + + + Solves a system of linear equations whose coefficients are in a + triangular band matrix. + + + .. container:: section + :name: GUID-4AC7186F-2D61-44C2-95BC-5981E750A021 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void tbsv(queue &exec_queue, uplo upper_lower, transpose trans, diag unit_nonunit, std::int64_t n, std::int64_t k, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx) + + ``tbsv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-5AF4221C-AB14-4F9B-97A8-CAA78DF05E36 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The tbsv routines solve a system of linear equations whose + coefficients are in a triangular band matrix. The operation is + defined as + + + + + + op(A)*x = b + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``A`` is an ``n``-by-``n`` unit or non-unit, upper or lower + triangular band matrix, with (``k`` + 1) diagonals, + + + ``b`` and ``x`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_nonunit + Specifies whether the matrix ``A`` is unit triangular or not. See + :ref:`onemkl_datatypes` + for more details. + + + n + Number of rows and columns of ``A``. Must be at least zero. + + + k + Number of sub/super-diagonals of the matrix ``A``. Must be at + least zero. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least (``k`` + 1), + and positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + +.. container:: section + :name: GUID-24B3C6B8-7FBD-4B24-84F2-242635B3026E + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding the solution vector ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/tpmv.rst b/docs/domains/blas/tpmv.rst new file mode 100644 index 000000000..f8bd2b136 --- /dev/null +++ b/docs/domains/blas/tpmv.rst @@ -0,0 +1,150 @@ +.. _tpmv: + +tpmv +==== + + +.. container:: + + + Computes a matrix-vector product using a triangular packed matrix. + + + .. container:: section + :name: GUID-5785B6D6-DB9C-43FA-B98A-009D5E077A9D + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void tpmv(queue &exec_queue, uplo upper_lower, transpose trans, diag unit_nonunit, std::int64_t n, buffer &a, buffer &x, std::int64_t incx) + + ``tpmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-A045480A-2EC1-4C73-A836-468324FCC85A + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The tpmv routines compute a matrix-vector product with a triangular + packed matrix. The operation is defined as + + + + + + x <- op(A)*x + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``A`` is an ``n``-by-``n`` unit or non-unit, upper or lower + triangular band matrix, supplied in packed form, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_nonunit + Specifies whether the matrix ``A`` is unit triangular or not. See + :ref:`onemkl_datatypes` + for more details. + + + n + Numbers of rows and columns of ``A``. Must be at least zero. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``\ +1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + +.. container:: section + :name: GUID-180038D9-902F-4B20-AB6B-E38F2A6C83E4 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding the updated vector ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/tpsv.rst b/docs/domains/blas/tpsv.rst new file mode 100644 index 000000000..0ec419bd0 --- /dev/null +++ b/docs/domains/blas/tpsv.rst @@ -0,0 +1,153 @@ +.. _tpsv: + +tpsv +==== + + +.. container:: + + + Solves a system of linear equations whose coefficients are in a + triangular packed matrix. + + + .. container:: section + :name: GUID-230CF8CA-B38D-4CB6-9917-029FEF53EBED + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void tpsv(queue &exec_queue, uplo upper_lower, transpose trans, diag unit_nonunit, std::int64_t n, std::int64_t k, buffer &a, buffer &x, std::int64_t incx) + + ``tpsv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-7AD9F8E2-1343-4A6D-8C6A-F68D934292B7 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The tpsv routines solve a system of linear equations whose + coefficients are in a triangular packed matrix. The operation is + defined as + + + + + + op(A)*x = b + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``A`` is an ``n``-by-``n`` unit or non-unit, upper or lower + triangular band matrix, supplied in packed form, + + + ``b`` and ``x`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_nonunit + Specifies whether the matrix ``A`` is unit triangular or not. See + :ref:`onemkl_datatypes` + for more details. + + + n + Numbers of rows and columns of ``A``. Must be at least zero. + + + a + Buffer holding input matrix ``A``. Must have size at least + (``n``\ \*(``n``\ +1))/2. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + x + Buffer holding the ``n``-element right-hand side vector ``b``. The + buffer must be of size at least (1 + (``n`` - 1)*abs(``incx``)). + See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + +.. container:: section + :name: GUID-F515C77C-1E84-424B-A00A-874ACBEFBF9E + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding the solution vector ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/trmm.rst b/docs/domains/blas/trmm.rst new file mode 100644 index 000000000..2dbbc85b0 --- /dev/null +++ b/docs/domains/blas/trmm.rst @@ -0,0 +1,208 @@ +.. _trmm: + +trmm +==== + + +.. container:: + + + Computes a matrix-matrix product where one input matrix is triangular + and one input matrix is general. + + + .. container:: section + :name: GUID-15B16EFC-8B31-4459-88DC-A8C5EF6C9932 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void trmm(queue &exec_queue, uplo upper_lower, transpose transa, diag unit_diag, std::int64_t m, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb) + + trmm supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-E1AAECF3-E29D-411F-B052-2F2E8080F3A1 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The trmm routines compute a scalar-matrix-matrix product where one of + the matrices in the multiplication is triangular. The argument + ``left_right`` determines if the triangular matrix, ``A``, is on the + left of the multiplication (``left_right`` = ``side::left``) or on + the right (``left_right`` = ``side::right``). Depending on + ``left_right``. The operation is defined as + + + + + + B <- alpha*op(A)*B + + + or + + + + + + B <- alpha*B*op(A) + + + where: + + + op(``A``) is one of op(``A``) = *A*, or op(``A``) = ``A``\ :sup:`T`, + or op(``A``) = ``A``\ :sup:`H`, + + + ``alpha`` is a scalar, + + + ``A`` is a triangular matrix, and ``B`` is a general matrix. + + + Here ``B`` is ``m`` x ``n`` and ``A`` is either ``m`` x ``m`` or + ``n`` x ``n``, depending on ``left_right``. + + +.. container:: section + :name: GUID-DE8B0FD7-11E3-42BC-99ED-3A07040FA6CB + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + left_right + Specifies whether ``A`` is on the left side of the multiplication + (``side::left``) or on the right side (``side::right``). See + :ref:`onemkl_datatypes` for more + details. + + + uplo + Specifies whether the matrix ``A`` is upper or lower triangular. + See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_diag + Specifies whether ``A`` is assumed to be unit triangular (all + diagonal elements are 1). See + :ref:`onemkl_datatypes` for more + details. + + + m + Specifies the number of rows of ``B``. The value of ``m`` must be + at least zero. + + + n + Specifies the number of columns of ``B``. The value of ``n`` must + be at least zero. + + + alpha + Scaling factor for the matrix-matrix product. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``m`` if ``left_right`` = ``side::left``, or + ``lda``\ \*\ ``n`` if ``left_right`` = ``side::right``. See + `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``m`` if + ``left_right`` = ``side::left``, and at least ``n`` if + ``left_right`` = ``side::right``. Must be positive. + + + b + Buffer holding input/output matrix ``B``. Must have size at least + ``ldb``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be at least ``m`` and positive. + + +.. container:: section + :name: GUID-1F1FF9D8-3833-4C9E-9CAC-53BA1791DCF1 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + b + Output buffer, overwritten by ``alpha``\ \*op(``A``)\*\ ``B`` or + ``alpha``\ \*\ ``B``\ \*op(``A``). + + +.. container:: section + :name: EXAMPLE_5EF48B8A07D849EA84A74FE22F0D5B24 + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``alpha`` = 0, matrix ``B`` is set to zero, and ``A`` and ``B`` do + not need to be initialized at entry. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/trmv.rst b/docs/domains/blas/trmv.rst new file mode 100644 index 000000000..14476e1e8 --- /dev/null +++ b/docs/domains/blas/trmv.rst @@ -0,0 +1,155 @@ +.. _trmv: + +trmv +==== + + +.. container:: + + + Computes a matrix-vector product using a triangular matrix. + + + .. container:: section + :name: GUID-15041079-C2F5-4D3C-85C2-262E184F7FFE + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void trmv(queue &exec_queue, uplo upper_lower, transpose trans, diag unit_nonunit, std::int64_t n, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx) + + ``trmv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-420DC613-E11B-48A8-B73F-55B55EBFC3B7 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The trmv routines compute a matrix-vector product with a triangular + matrix. The operation is defined + + + + + + x <- op(A)*x + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``A`` is an ``n``-by-``n`` unit or non-unit, upper or lower + triangular band matrix, + + + ``x`` is a vector of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_nonunit + Specifies whether the matrix ``A`` is unit triangular or not. See + :ref:`onemkl_datatypes` + for more details. + + + n + Numbers of rows and columns of ``A``. Must be at least zero. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``n``, and + positive. + + + x + Buffer holding input vector ``x``. The buffer must be of size at + least (1 + (``n`` - 1)*abs(``incx``)). See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + +.. container:: section + :name: GUID-7BF1D5C9-EB8C-4BD6-B0E7-A66DAC3221F9 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding the updated vector ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/blas/trsm.rst b/docs/domains/blas/trsm.rst new file mode 100644 index 000000000..958d231db --- /dev/null +++ b/docs/domains/blas/trsm.rst @@ -0,0 +1,205 @@ +.. _trsm: + +trsm +==== + + +.. container:: + + + Solves a triangular matrix equation (forward or backward solve). + + + .. container:: section + :name: GUID-6F8E0E22-B30A-4825-B508-CEDE0CAC8B90 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void trsm(queue &exec_queue, side left_right, uplo upper_lower, transpose transa, diag unit_diag, std::int64_t m, std::int64_t n, T alpha, buffer &a, std::int64_t lda, buffer &b, std::int64_t ldb) + + trsm supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-AE6CFEF4-4058-49C3-BABC-2B05D6594555 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The trsm routines solve one of the following matrix equations: + + + + + + op(A)*X = alpha*B, + + + or + + + + + + X*op(A) = alpha*B, + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``alpha`` is a scalar, + + + ``A`` is a triangular matrix, and + + + ``B`` and ``X`` are ``m`` x ``n`` general matrices. + + + ``A`` is either ``m`` x ``m`` or ``n`` x ``n``, depending on whether + it multiplies ``X`` on the left or right. On return, the matrix ``B`` + is overwritten by the solution matrix ``X``. + + +.. container:: section + :name: GUID-0BBDCB60-8CDE-4EBD-BDE5-F7688B4B29F4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + left_right + Specifies whether ``A`` multiplies ``X`` on the left + (``side::left``) or on the right (``side::right``). See + :ref:`onemkl_datatypes` for more + details. + + + uplo + Specifies whether the matrix ``A`` is upper or lower triangular. + See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_diag + Specifies whether ``A`` is assumed to be unit triangular (all + diagonal elements are 1). See + :ref:`onemkl_datatypes` for more + details. + + + m + Specifies the number of rows of ``B``. The value of ``m`` must be + at least zero. + + + n + Specifies the number of columns of ``B``. The value of ``n`` must + be at least zero. + + + alpha + Scaling factor for the solution. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``m`` if ``left_right`` = ``side::left``, or + ``lda``\ \*\ ``n`` if ``left_right`` = ``side::right``. See + `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of ``A``. Must be at least ``m`` if + ``left_right`` = ``side::left``, and at least ``n`` if + ``left_right`` = ``side::right``. Must be positive. + + + b + Buffer holding input/output matrix ``B``. Must have size at least + ``ldb``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + ldb + Leading dimension of ``B``. Must be at least ``m`` and positive. + + +.. container:: section + :name: GUID-7AC6C3B9-7A31-4E0B-B770-FD607E7F9BE5 + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + b + Output buffer. Overwritten by the solution matrix ``X``. + + +.. container:: section + :name: EXAMPLE_5EF48B8A07D849EA84A74FE22F0D5B24 + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``alpha`` = 0, matrix ``B`` is set to zero, and ``A`` and ``B`` do + not need to be initialized at entry. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-3-routines` + + + +.. container:: + diff --git a/docs/domains/blas/trsm_batch.rst b/docs/domains/blas/trsm_batch.rst new file mode 100644 index 000000000..0cf82ce21 --- /dev/null +++ b/docs/domains/blas/trsm_batch.rst @@ -0,0 +1,436 @@ +.. _trsm_batch: + +trsm_batch +========== + + +.. container:: + + + Computes groups of matrix-matrix product with general matrices. + + + .. container:: section + :name: GUID-6F8E0E22-B30A-4825-B508-CEDE0CAC8B90 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + **Group API** + + + .. cpp:function:: void trsm_batch(queue &exec_queue, buffer &left_right_array, buffer &upper_lower_array, buffer &trans_array, buffer &unit_diag_array, buffer &m_array, buffer &n_array, buffer &alpha_array, buffer &a_array, buffer &lda_array, buffer &b_array, buffer ldb_array, std::int64_t group_count, buffer &group_size_array) + + **Strided API** + + + .. cpp:function:: void trsm_batch(queue &exec_queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, T alpha, buffer &a, std::int64_t lda, std::int64_t stridea, buffer &b, std::int64_t ldb, std::int64_t strideb, std::int64_t batch_size) + + ``trsm_batch`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-AE6CFEF4-4058-49C3-BABC-2B05D6594555 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The trsm_batch routines solve a series of equations of the form op(A) + \* X = alpha \* B or X \* op(A) = alpha \* B. They are similar to the + trsm routine counterparts, but the trsm_batch routines solve linear + equations with groups of matrices. The groups contain matrices with + the same parameters. + + + For the group API, the operation is defined as + + + :: + + + offa = 0, offb = 0 + for i = 0 … group_count – 1 + left_right, uplo, trans, unit_diag, m, n, lda, ldb, alpha and group_size at position i in left_right_array, uplo_array, trans_array, unit_diag_array, m_array, n_array, lda_array, ldb_array, alpha_array and group_size_array + sizea = left_right == onemkl::side::L ? lda * m : lda * n; + sizeb = ldb * n; + for j = 0 … group_size – 1 + A and B are matrices of size sizea and sizeb at offset offa and offb in a and b. + if (left_right == onemkl::side::L) then + computes X such that op(A) * X = alpha * B + else + computes X such that X * op(A) = alpha * B + end if + B := X + offa += sizea, offb += sizeb + end for + end for + + + For the strided API, the operation is defined as + + + :: + + + for i = 0 … batch_size – 1 + A and B are matrices at offset i * stridea and i * strideb in a and b. + if (left_right == onemkl::side::L) then + computes X such that op(A) * X = alpha * B + else + computes X such that X * op(A) = alpha * B + end if + B := X + end for + + + where: + + + - op(``A``) is one of op(``A``) = ``A``, or op(A) = ``A``\ :sup:`T`, + or op(``A``) = ``A``\ :sup:`H` + + + - alpha is a scalar + + + - ``A`` is a triangular matrix + + + - ``B`` and ``X`` are ``m`` x ``n`` general matrices + + + - The a and b buffers contains all the input matrices. The stride + between matrices is either given by the exact size of the matrix + (for the group API) or by the stride parameter. The total number + of matrices in a and b is given by the + + |image0| + + for the strided + API. + + + ``A`` is either ``m`` x ``m`` or ``n`` x ``n``,depending on whether + it multiplies ``X`` on the leftor right. On return, the matrix ``B`` + is overwrittenby the solution matrix ``X``. + + +.. container:: section + :name: GUID-863264A0-4CE9-495F-A617-102E46D7A41A + + + .. rubric:: Input Parameters - Group API + :name: input-parameters---group-api + :class: sectiontitle + + + left_right_array + Buffer holding ``group_count onemkl::side`` value. + + + For the group ``i``, ``left_right`` is the ``i``\ th element in + the left_right_array buffer and specifies whether ``A`` multiplies + ``X`` on the left (``side::left``) or on the right + (``side::right``). See + :ref:`onemkl_datatypes` for more + details. + + + uplo_array + Buffer holding ``group_count onemkl::uplo`` value. + + + For the group ``i``, ``uplo`` is the ``i``\ th element in the + uplo_array buffer and specifies whether ``A`` is upper or lower + triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans_array + Buffer holding ``group_count onemkl::transpose`` value. + + + For the group ``i``, ``trans`` is the ``i``\ th element in the + trans_array buffer and specifies the form of ``op``\ (``A``) used + in the matrix multiplication. See + :ref:`onemkl_datatypes` for more + details. + + + unit_diag__array + Buffer holding ``group_count onemkl::diag`` value. + + + For the group ``i``, ``unit_diag`` is the ``i``\ th element in the + unit_diag_array buffer and specifies whether ``A`` is assumed to + be unit triangular (all diagonal elements are 1). See + :ref:`onemkl_datatypes` for more + details. + + + m_array + Buffer holding ``group_count`` integer. For the group ``i``, ``m`` + is the ``i``\ th element in the m_array buffer and specifies the + number of rows of ``B``. Must be at least zero. + + + n_array + Buffer holding ``group_count`` integer. For the group ``i``, ``n`` + is the ``i``\ th element in the n_array buffer and specifies the + number of columns of ``B``. Must be at least zero. + + + alpha_array + Buffer holding ``group_count`` scalar element. For the group + ``i``, ``alpha`` is the ``i``\ th element in the alpha_array + buffer and specifies the scaling factor for the matrix-matrix + product. + + + a + Buffer holding the input matrix ``A``. The total size of the + buffer ``a`` must be at least the sum of the sizes of all the + matricies ``A``. That is, + + + |image1| + + + where + ``sizeai = lda_array[i] * (left_right == onemkl::side::L ? m : n)`` + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + lda_array + Buffer holding ``group_count`` integer. For the group ``i``, + ``lda`` is the ``i``\ th element in the lda_array buffer and + specifies the leading dimension of ``A``. Must be at least ``m`` + if ``A`` is not transposed, and at least ``k`` if ``A`` is + transposed. Must be positive. + + + b + Buffer holding the input matrix ``B``. The total size of the + buffer ``b`` must be at least the sum of the sizes of all the + matricies ``B``. That is, + + + |image2| + + + See `Matrix + Storage <../matrix-storage.html>`__ for + more details. + + + ldb_array + Buffer holding ``group_count`` integer. For the group ``i``, + ``ldb`` is the ``i``\ th element in the ldb_array buffer and + specifies the leading dimension of ``B``. Must be at least ``n``. + Must be positive. + + + group_count + Specifies the number of groups. Must be at least 0. + + + group_size_array + Buffer holding the group_count integer. For the group ``i``, + ``ldb`` is the ``i``\ th element in the group_size_array buffer + specifies the number of matrix multiply operations in + group\ ``i``. Each element in group_size_array must be at least 0. + + +.. container:: section + :name: GUID-1E4953E6-F7B1-4FEE-BA5A-8C4BD51DC700 + + + .. rubric:: Output Parameters - Group API + :name: output-parameters---group-api + :class: sectiontitle + + + b + Output buffer, overwritten by the ``total_batch_count`` solution + matrices ``X``. + + +.. container:: section + :name: GUID-D067773A-45A3-4D24-B10A-46E27834947E + + + .. rubric:: Input Parameters - Strided API + :name: input-parameters---strided-api + :class: sectiontitle + + + left_right + Specifies whether the matrices ``A`` multiply ``X`` on the left + (``side::left``) or on the right (``side::right``). See + :ref:`onemkl_datatypes` for more + details. + + + uplo + Specifies whether the matrices ``A`` are upper or lower + triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies ``op(A)``, the transposition operation applied to the + matrices ``A``. See + :ref:`onemkl_datatypes` for more + details. + + + unit_diag + Specifies whether the matrices ``A`` are assumed to be unit + triangular (all diagonal elements are 1.). See + :ref:`onemkl_datatypes` for more + details. + + + m + Number of rows of the ``B`` matrices. Must be at least zero. + + + n + Number of columns of the ``B`` matrices. Must be at least zero. + + + alpha + Scaling factor for the solutions. + + + a + Buffer holding the input matrices ``A``. Must have size at least + ``stridea*batch_size``. + + + lda + Leading dimension of the matrices ``A``. Must be at least ``m`` if + left_right = ``side::left``, and at least ``n`` if left_right = + ``side::right``. Must be positive. + + + stridea + Stride between the different ``A`` matrices. + + + If left_right = ``side::left``, the matrices ``A`` are + ``m``-by-``m`` matrices, so stridea must be at least lda\*\ ``m``. + + + If left_right = ``side::right``, the matrices ``A`` are + ``n``-by-``n`` matrices, so stridea must be at least lda\*\ ``n``. + + + b + Buffer holding the input matrices ``B``. Must have size at least + ``strideb*batch_size``. + + + ldb + Leading dimension of the matrices ``B``. Must be at least ``m`` + and must be positive. + + + strideb + Stride between the different ``B`` matrices. Must be at least + ldb\*\ ``n``. + + + beta + Scaling factor for the matrices ``C``. + + + c + Buffer holding input/output matrices ``C``. Must have size at + least ``stridec*batch_size``. + + + ldc + Leading dimension of ``C``. Must be positive and at least ``m``. + + + stridec + Stride between the different ``C`` matrices. Must be at least + ``ldc*n``. + + + batch_size + Specifies the number of triangular linear systems to solve. + + +.. container:: section + :name: GUID-98C3DE17-4F5F-41A1-B431-48148153ABBA + + + .. rubric:: Output Parameters - Strided API + :name: output-parameters---strided-api + :class: sectiontitle + + + b + Output buffer, overwritten by ``batch_size`` solution matrices + ``X``. + + +.. container:: section + :name: GUID-AC72653A-4AC8-4B9D-B7A9-13A725AA19BF + + + .. rubric:: Notes + :name: notes + :class: sectiontitle + + + If ``alpha`` = 0, matrix ``B`` is set to zero, and the matrices ``A`` + and ``B`` do not need to be initialized before calling trsm_batch. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-like-extensions` + + + +.. container:: + + +.. |image0| image:: ../equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee1.png + :class: img-middle +.. |image1| image:: ../equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee2.png + :class: img-middle +.. |image2| image:: ../equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee3.png + :class: img-middle + diff --git a/docs/domains/blas/trsv.rst b/docs/domains/blas/trsv.rst new file mode 100644 index 000000000..e1dba6e43 --- /dev/null +++ b/docs/domains/blas/trsv.rst @@ -0,0 +1,157 @@ +.. _trsv: + +trsv +==== + + +.. container:: + + + Solves a system of linear equations whose coefficients are in a + triangular matrix. + + + .. container:: section + :name: GUID-9BA4C1B6-479B-41B1-BCA8-7826F40DA952 + + + .. rubric:: Syntax + :name: syntax + :class: sectiontitle + + + .. cpp:function:: void trsv(queue &exec_queue, uplo upper_lower, transpose trans, diag unit_nonunit, std::int64_t n, std::int64_t k, buffer &a, std::int64_t lda, buffer &x, std::int64_t incx) + + ``trsv`` supports the following precisions. + + + .. list-table:: + :header-rows: 1 + + * - T + * - ``float`` + * - ``double`` + * - ``std::complex`` + * - ``std::complex`` + + + + +.. container:: section + :name: GUID-D500B67B-5DD6-4471-B0BD-53FD9A3C7BF2 + + + .. rubric:: Description + :name: description + :class: sectiontitle + + + The trsv routines compute a matrix-vector product with a triangular + band matrix. The operation is defined as + + + + + + op(A)*x = b + + + where: + + + op(``A``) is one of op(``A``) = ``A``, or op(``A``) = + ``A``\ :sup:`T`, or op(``A``) = ``A``\ :sup:`H`, + + + ``A`` is an ``n``-by-``n`` unit or non-unit, upper or lower + triangular matrix, + + + ``b`` and ``x`` are vectors of length ``n``. + + +.. container:: section + :name: GUID-E1436726-01FE-4206-871E-B905F59A96B4 + + + .. rubric:: Input Parameters + :name: input-parameters + :class: sectiontitle + + + exec_queue + The queue where the routine should be executed. + + + upper_lower + Specifies whether ``A`` is upper or lower triangular. See + :ref:`onemkl_datatypes` for more + details. + + + trans + Specifies op(``A``), the transposition operation applied to ``A``. + See + :ref:`onemkl_datatypes` for more + details. + + + unit_nonunit + Specifies whether the matrix ``A`` is unit triangular or not. See + :ref:`onemkl_datatypes` + for more details. + + + n + Numbers of rows and columns of ``A``. Must be at least zero. + + + a + Buffer holding input matrix ``A``. Must have size at least + ``lda``\ \*\ ``n``. See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + lda + Leading dimension of matrix ``A``. Must be at least ``n``, and + positive. + + + x + Buffer holding the ``n``-element right-hand side vector ``b``. The + buffer must be of size at least (1 + (``n`` - 1)*abs(``incx``)). + See `Matrix and Vector + Storage <../matrix-storage.html>`__ for + more details. + + + incx + Stride of vector ``x``. + + +.. container:: section + :name: GUID-7E0AF44F-2D83-41A3-A58E-50400ECDBD9A + + + .. rubric:: Output Parameters + :name: output-parameters + :class: sectiontitle + + + x + Buffer holding the solution vector ``x``. + + +.. container:: familylinks + + + .. container:: parentlink + + + **Parent topic:** :ref:`blas-level-2-routines` + + + +.. container:: + diff --git a/docs/domains/equations/GUID-0026D841-74F3-43C0-8EB5-F9E4107EF95D-low.gif b/docs/domains/equations/GUID-0026D841-74F3-43C0-8EB5-F9E4107EF95D-low.gif new file mode 100644 index 000000000..8a6229952 Binary files /dev/null and b/docs/domains/equations/GUID-0026D841-74F3-43C0-8EB5-F9E4107EF95D-low.gif differ diff --git a/docs/domains/equations/GUID-02486559-84D2-4880-8EAB-2085A5BE2D10-low.jpg b/docs/domains/equations/GUID-02486559-84D2-4880-8EAB-2085A5BE2D10-low.jpg new file mode 100644 index 000000000..3dd71939b Binary files /dev/null and b/docs/domains/equations/GUID-02486559-84D2-4880-8EAB-2085A5BE2D10-low.jpg differ diff --git a/docs/domains/equations/GUID-02EEA5FC-8F46-4034-86D9-99900F93373C-low.gif b/docs/domains/equations/GUID-02EEA5FC-8F46-4034-86D9-99900F93373C-low.gif new file mode 100644 index 000000000..7220f68cf Binary files /dev/null and b/docs/domains/equations/GUID-02EEA5FC-8F46-4034-86D9-99900F93373C-low.gif differ diff --git a/docs/domains/equations/GUID-03C78F41-D16C-4BFE-B087-BE4F5A6A0ee1.png b/docs/domains/equations/GUID-03C78F41-D16C-4BFE-B087-BE4F5A6A0ee1.png new file mode 100644 index 000000000..64d43300e Binary files /dev/null and b/docs/domains/equations/GUID-03C78F41-D16C-4BFE-B087-BE4F5A6A0ee1.png differ diff --git a/docs/domains/equations/GUID-04615D14-A026-4BF0-ACD6-0FC822FEC64E-low.gif b/docs/domains/equations/GUID-04615D14-A026-4BF0-ACD6-0FC822FEC64E-low.gif new file mode 100644 index 000000000..952be1865 Binary files /dev/null and b/docs/domains/equations/GUID-04615D14-A026-4BF0-ACD6-0FC822FEC64E-low.gif differ diff --git a/docs/domains/equations/GUID-061AF9F8-B166-4154-9BF1-4E2C99F1CE1F-low.png b/docs/domains/equations/GUID-061AF9F8-B166-4154-9BF1-4E2C99F1CE1F-low.png new file mode 100644 index 000000000..52e9a914d Binary files /dev/null and b/docs/domains/equations/GUID-061AF9F8-B166-4154-9BF1-4E2C99F1CE1F-low.png differ diff --git a/docs/domains/equations/GUID-081A19C4-609F-4736-BCCF-D680013A2775-low.gif b/docs/domains/equations/GUID-081A19C4-609F-4736-BCCF-D680013A2775-low.gif new file mode 100644 index 000000000..410e26c9f Binary files /dev/null and b/docs/domains/equations/GUID-081A19C4-609F-4736-BCCF-D680013A2775-low.gif differ diff --git a/docs/domains/equations/GUID-09FC1496-B5B3-4DF6-A3EE-E6410BE1EFD2-low.jpg b/docs/domains/equations/GUID-09FC1496-B5B3-4DF6-A3EE-E6410BE1EFD2-low.jpg new file mode 100644 index 000000000..399b53dab Binary files /dev/null and b/docs/domains/equations/GUID-09FC1496-B5B3-4DF6-A3EE-E6410BE1EFD2-low.jpg differ diff --git a/docs/domains/equations/GUID-0A406EAC-6A1D-4D81-977C-08C018161E3F-low.jpg b/docs/domains/equations/GUID-0A406EAC-6A1D-4D81-977C-08C018161E3F-low.jpg new file mode 100644 index 000000000..8355ef081 Binary files /dev/null and b/docs/domains/equations/GUID-0A406EAC-6A1D-4D81-977C-08C018161E3F-low.jpg differ diff --git a/docs/domains/equations/GUID-0A8E6C61-9171-4584-927A-83AC482ADC4D-low.gif b/docs/domains/equations/GUID-0A8E6C61-9171-4584-927A-83AC482ADC4D-low.gif new file mode 100644 index 000000000..45a2a0446 Binary files /dev/null and b/docs/domains/equations/GUID-0A8E6C61-9171-4584-927A-83AC482ADC4D-low.gif differ diff --git a/docs/domains/equations/GUID-0B9A59CC-C914-429D-AF87-93B16DABD291-low.jpg b/docs/domains/equations/GUID-0B9A59CC-C914-429D-AF87-93B16DABD291-low.jpg new file mode 100644 index 000000000..bdd79355c Binary files /dev/null and b/docs/domains/equations/GUID-0B9A59CC-C914-429D-AF87-93B16DABD291-low.jpg differ diff --git a/docs/domains/equations/GUID-0C4BC54F-4C29-4577-80AA-BCBCD291582A-low.jpg b/docs/domains/equations/GUID-0C4BC54F-4C29-4577-80AA-BCBCD291582A-low.jpg new file mode 100644 index 000000000..4903f92db Binary files /dev/null and b/docs/domains/equations/GUID-0C4BC54F-4C29-4577-80AA-BCBCD291582A-low.jpg differ diff --git a/docs/domains/equations/GUID-0EDD2BB7-A284-495A-84F6-8A210AC499CD-low.gif b/docs/domains/equations/GUID-0EDD2BB7-A284-495A-84F6-8A210AC499CD-low.gif new file mode 100644 index 000000000..2164c6861 Binary files /dev/null and b/docs/domains/equations/GUID-0EDD2BB7-A284-495A-84F6-8A210AC499CD-low.gif differ diff --git a/docs/domains/equations/GUID-0F47CAD3-006C-4A78-B229-413313667ee1.png b/docs/domains/equations/GUID-0F47CAD3-006C-4A78-B229-413313667ee1.png new file mode 100644 index 000000000..6d5a0c873 Binary files /dev/null and b/docs/domains/equations/GUID-0F47CAD3-006C-4A78-B229-413313667ee1.png differ diff --git a/docs/domains/equations/GUID-11B96BA7-C321-446D-A7B6-8D84C8CBC076-low.png b/docs/domains/equations/GUID-11B96BA7-C321-446D-A7B6-8D84C8CBC076-low.png new file mode 100644 index 000000000..f0f8001cf Binary files /dev/null and b/docs/domains/equations/GUID-11B96BA7-C321-446D-A7B6-8D84C8CBC076-low.png differ diff --git a/docs/domains/equations/GUID-12315BD9-4F4D-42B0-A9B5-68A22D40756D-low.jpg b/docs/domains/equations/GUID-12315BD9-4F4D-42B0-A9B5-68A22D40756D-low.jpg new file mode 100644 index 000000000..7f89f612e Binary files /dev/null and b/docs/domains/equations/GUID-12315BD9-4F4D-42B0-A9B5-68A22D40756D-low.jpg differ diff --git a/docs/domains/equations/GUID-19F7C7EA-5657-4016-87A6-4E2721994C56-low.gif b/docs/domains/equations/GUID-19F7C7EA-5657-4016-87A6-4E2721994C56-low.gif new file mode 100644 index 000000000..c54c63d65 Binary files /dev/null and b/docs/domains/equations/GUID-19F7C7EA-5657-4016-87A6-4E2721994C56-low.gif differ diff --git a/docs/domains/equations/GUID-1D36B5CE-3BF3-4762-926B-05C5527FAE45-low.gif b/docs/domains/equations/GUID-1D36B5CE-3BF3-4762-926B-05C5527FAE45-low.gif new file mode 100644 index 000000000..847d0e1ad Binary files /dev/null and b/docs/domains/equations/GUID-1D36B5CE-3BF3-4762-926B-05C5527FAE45-low.gif differ diff --git a/docs/domains/equations/GUID-2293B947-42D6-4E5F-BBB3-9DC135AA724A-low.gif b/docs/domains/equations/GUID-2293B947-42D6-4E5F-BBB3-9DC135AA724A-low.gif new file mode 100644 index 000000000..d497fad9f Binary files /dev/null and b/docs/domains/equations/GUID-2293B947-42D6-4E5F-BBB3-9DC135AA724A-low.gif differ diff --git a/docs/domains/equations/GUID-281DBA27-691A-4B62-A255-FC33EA28D8D5-low.jpg b/docs/domains/equations/GUID-281DBA27-691A-4B62-A255-FC33EA28D8D5-low.jpg new file mode 100644 index 000000000..dd1a85df9 Binary files /dev/null and b/docs/domains/equations/GUID-281DBA27-691A-4B62-A255-FC33EA28D8D5-low.jpg differ diff --git a/docs/domains/equations/GUID-2BEFE049-EB3B-4FC9-AD75-ABA053617238-low.jpg b/docs/domains/equations/GUID-2BEFE049-EB3B-4FC9-AD75-ABA053617238-low.jpg new file mode 100644 index 000000000..3350502a9 Binary files /dev/null and b/docs/domains/equations/GUID-2BEFE049-EB3B-4FC9-AD75-ABA053617238-low.jpg differ diff --git a/docs/domains/equations/GUID-2D60A1A4-9522-40FC-AEEA-B64EB795144C-low.gif b/docs/domains/equations/GUID-2D60A1A4-9522-40FC-AEEA-B64EB795144C-low.gif new file mode 100644 index 000000000..689e680dc Binary files /dev/null and b/docs/domains/equations/GUID-2D60A1A4-9522-40FC-AEEA-B64EB795144C-low.gif differ diff --git a/docs/domains/equations/GUID-2F91B385-0AC2-41D3-AE61-48F63A7DBB02-low.png b/docs/domains/equations/GUID-2F91B385-0AC2-41D3-AE61-48F63A7DBB02-low.png new file mode 100644 index 000000000..f9f19a239 Binary files /dev/null and b/docs/domains/equations/GUID-2F91B385-0AC2-41D3-AE61-48F63A7DBB02-low.png differ diff --git a/docs/domains/equations/GUID-32A750B8-7BCC-409B-BD48-E88FBEF86D32-low.jpg b/docs/domains/equations/GUID-32A750B8-7BCC-409B-BD48-E88FBEF86D32-low.jpg new file mode 100644 index 000000000..c35a12613 Binary files /dev/null and b/docs/domains/equations/GUID-32A750B8-7BCC-409B-BD48-E88FBEF86D32-low.jpg differ diff --git a/docs/domains/equations/GUID-3605ACD9-02D1-46D7-B791-F2F76F0D9ee1.png b/docs/domains/equations/GUID-3605ACD9-02D1-46D7-B791-F2F76F0D9ee1.png new file mode 100644 index 000000000..b985bb006 Binary files /dev/null and b/docs/domains/equations/GUID-3605ACD9-02D1-46D7-B791-F2F76F0D9ee1.png differ diff --git a/docs/domains/equations/GUID-38C12345-5E6E-4D94-8072-460502CB52EC-low.gif b/docs/domains/equations/GUID-38C12345-5E6E-4D94-8072-460502CB52EC-low.gif new file mode 100644 index 000000000..341bd1074 Binary files /dev/null and b/docs/domains/equations/GUID-38C12345-5E6E-4D94-8072-460502CB52EC-low.gif differ diff --git a/docs/domains/equations/GUID-3A9C1154-2E42-416F-8865-06E7382A3AA7-low.jpg b/docs/domains/equations/GUID-3A9C1154-2E42-416F-8865-06E7382A3AA7-low.jpg new file mode 100644 index 000000000..70f0de3b8 Binary files /dev/null and b/docs/domains/equations/GUID-3A9C1154-2E42-416F-8865-06E7382A3AA7-low.jpg differ diff --git a/docs/domains/equations/GUID-41F8B94B-EEFA-4308-925A-D2DF486FBF8B-low.gif b/docs/domains/equations/GUID-41F8B94B-EEFA-4308-925A-D2DF486FBF8B-low.gif new file mode 100644 index 000000000..342dcbd49 Binary files /dev/null and b/docs/domains/equations/GUID-41F8B94B-EEFA-4308-925A-D2DF486FBF8B-low.gif differ diff --git a/docs/domains/equations/GUID-41FCF9BC-28B7-4030-B904-1DBA03DD328C-low.gif b/docs/domains/equations/GUID-41FCF9BC-28B7-4030-B904-1DBA03DD328C-low.gif new file mode 100644 index 000000000..8a6229952 Binary files /dev/null and b/docs/domains/equations/GUID-41FCF9BC-28B7-4030-B904-1DBA03DD328C-low.gif differ diff --git a/docs/domains/equations/GUID-428BFB7A-6E88-4D12-9707-885C02A93A8E-low.jpg b/docs/domains/equations/GUID-428BFB7A-6E88-4D12-9707-885C02A93A8E-low.jpg new file mode 100644 index 000000000..3f14817bb Binary files /dev/null and b/docs/domains/equations/GUID-428BFB7A-6E88-4D12-9707-885C02A93A8E-low.jpg differ diff --git a/docs/domains/equations/GUID-42AF2BFE-F8F1-4F96-A4E0-05D4FB5A7ee1.png b/docs/domains/equations/GUID-42AF2BFE-F8F1-4F96-A4E0-05D4FB5A7ee1.png new file mode 100644 index 000000000..b985bb006 Binary files /dev/null and b/docs/domains/equations/GUID-42AF2BFE-F8F1-4F96-A4E0-05D4FB5A7ee1.png differ diff --git a/docs/domains/equations/GUID-482EEED2-95DF-4AA3-A484-E2CC41F29ee1.png b/docs/domains/equations/GUID-482EEED2-95DF-4AA3-A484-E2CC41F29ee1.png new file mode 100644 index 000000000..d012e4480 Binary files /dev/null and b/docs/domains/equations/GUID-482EEED2-95DF-4AA3-A484-E2CC41F29ee1.png differ diff --git a/docs/domains/equations/GUID-482EEED2-95DF-4AA3-A484-E2CC41F29ee2.png b/docs/domains/equations/GUID-482EEED2-95DF-4AA3-A484-E2CC41F29ee2.png new file mode 100644 index 000000000..dd0491c68 Binary files /dev/null and b/docs/domains/equations/GUID-482EEED2-95DF-4AA3-A484-E2CC41F29ee2.png differ diff --git a/docs/domains/equations/GUID-4835D5B4-6232-45CD-9A49-0264F8B0DBF4-low.gif b/docs/domains/equations/GUID-4835D5B4-6232-45CD-9A49-0264F8B0DBF4-low.gif new file mode 100644 index 000000000..f4d54db71 Binary files /dev/null and b/docs/domains/equations/GUID-4835D5B4-6232-45CD-9A49-0264F8B0DBF4-low.gif differ diff --git a/docs/domains/equations/GUID-4D962DF4-16F2-438B-8866-4F105DC41242-low.jpg b/docs/domains/equations/GUID-4D962DF4-16F2-438B-8866-4F105DC41242-low.jpg new file mode 100644 index 000000000..291680b40 Binary files /dev/null and b/docs/domains/equations/GUID-4D962DF4-16F2-438B-8866-4F105DC41242-low.jpg differ diff --git a/docs/domains/equations/GUID-4F76F5A1-251F-4AC0-A2E0-A3B4B6F39ee1.png b/docs/domains/equations/GUID-4F76F5A1-251F-4AC0-A2E0-A3B4B6F39ee1.png new file mode 100644 index 000000000..0c745c1a2 Binary files /dev/null and b/docs/domains/equations/GUID-4F76F5A1-251F-4AC0-A2E0-A3B4B6F39ee1.png differ diff --git a/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee1.png b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee1.png new file mode 100644 index 000000000..159240a3d Binary files /dev/null and b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee1.png differ diff --git a/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee2.png b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee2.png new file mode 100644 index 000000000..abc61e806 Binary files /dev/null and b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee2.png differ diff --git a/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee3.png b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee3.png new file mode 100644 index 000000000..6ca5e88f7 Binary files /dev/null and b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee3.png differ diff --git a/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee4.png b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee4.png new file mode 100644 index 000000000..7b2111b11 Binary files /dev/null and b/docs/domains/equations/GUID-50960934-BF9F-4070-BC8E-AE05FD9AFee4.png differ diff --git a/docs/domains/equations/GUID-5159E8DD-25FF-473A-86AA-1E71FFCD018C-low.jpg b/docs/domains/equations/GUID-5159E8DD-25FF-473A-86AA-1E71FFCD018C-low.jpg new file mode 100644 index 000000000..b7f737c66 Binary files /dev/null and b/docs/domains/equations/GUID-5159E8DD-25FF-473A-86AA-1E71FFCD018C-low.jpg differ diff --git a/docs/domains/equations/GUID-5193801D-8E3B-43A2-989E-09A8431FD34E-low.png b/docs/domains/equations/GUID-5193801D-8E3B-43A2-989E-09A8431FD34E-low.png new file mode 100644 index 000000000..2a64154c2 Binary files /dev/null and b/docs/domains/equations/GUID-5193801D-8E3B-43A2-989E-09A8431FD34E-low.png differ diff --git a/docs/domains/equations/GUID-51FA6703-70D8-4D01-B7AB-A163F4CDFC94-low.jpg b/docs/domains/equations/GUID-51FA6703-70D8-4D01-B7AB-A163F4CDFC94-low.jpg new file mode 100644 index 000000000..32656a177 Binary files /dev/null and b/docs/domains/equations/GUID-51FA6703-70D8-4D01-B7AB-A163F4CDFC94-low.jpg differ diff --git a/docs/domains/equations/GUID-5F24679E-82BB-4ECB-AC87-02FC7CF3C77F-low.jpg b/docs/domains/equations/GUID-5F24679E-82BB-4ECB-AC87-02FC7CF3C77F-low.jpg new file mode 100644 index 000000000..d94da88ba Binary files /dev/null and b/docs/domains/equations/GUID-5F24679E-82BB-4ECB-AC87-02FC7CF3C77F-low.jpg differ diff --git a/docs/domains/equations/GUID-5FAAD02F-09F5-4B78-B404-384F1270FA1C-low.jpg b/docs/domains/equations/GUID-5FAAD02F-09F5-4B78-B404-384F1270FA1C-low.jpg new file mode 100644 index 000000000..7ad17fdd5 Binary files /dev/null and b/docs/domains/equations/GUID-5FAAD02F-09F5-4B78-B404-384F1270FA1C-low.jpg differ diff --git a/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee1.png b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee1.png new file mode 100644 index 000000000..96492de34 Binary files /dev/null and b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee1.png differ diff --git a/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee2.png b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee2.png new file mode 100644 index 000000000..0f9069655 Binary files /dev/null and b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee2.png differ diff --git a/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee3.png b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee3.png new file mode 100644 index 000000000..7e0976ad3 Binary files /dev/null and b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee3.png differ diff --git a/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee4.png b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee4.png new file mode 100644 index 000000000..00cf06979 Binary files /dev/null and b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee4.png differ diff --git a/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee5.png b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee5.png new file mode 100644 index 000000000..3c5f4e31b Binary files /dev/null and b/docs/domains/equations/GUID-608D9BA6-827F-48DE-A01F-0EE5991F7ee5.png differ diff --git a/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee1.png b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee1.png new file mode 100644 index 000000000..96492de34 Binary files /dev/null and b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee1.png differ diff --git a/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee2.png b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee2.png new file mode 100644 index 000000000..0f9069655 Binary files /dev/null and b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee2.png differ diff --git a/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee3.png b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee3.png new file mode 100644 index 000000000..7e0976ad3 Binary files /dev/null and b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee3.png differ diff --git a/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee4.png b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee4.png new file mode 100644 index 000000000..00cf06979 Binary files /dev/null and b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee4.png differ diff --git a/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee5.png b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee5.png new file mode 100644 index 000000000..3c5f4e31b Binary files /dev/null and b/docs/domains/equations/GUID-67FC4AB3-40CB-441F-BA9F-88BAAC78Cee5.png differ diff --git a/docs/domains/equations/GUID-684BB993-83CA-4605-BD49-E493806C1ee1.png b/docs/domains/equations/GUID-684BB993-83CA-4605-BD49-E493806C1ee1.png new file mode 100644 index 000000000..0c745c1a2 Binary files /dev/null and b/docs/domains/equations/GUID-684BB993-83CA-4605-BD49-E493806C1ee1.png differ diff --git a/docs/domains/equations/GUID-6ABD7CD8-8E05-409D-B84F-9B88E4CDE9DB-low.gif b/docs/domains/equations/GUID-6ABD7CD8-8E05-409D-B84F-9B88E4CDE9DB-low.gif new file mode 100644 index 000000000..386e8d1a4 Binary files /dev/null and b/docs/domains/equations/GUID-6ABD7CD8-8E05-409D-B84F-9B88E4CDE9DB-low.gif differ diff --git a/docs/domains/equations/GUID-6F53C93C-0634-4E53-8874-5ACBD4C9AA3E-low.gif b/docs/domains/equations/GUID-6F53C93C-0634-4E53-8874-5ACBD4C9AA3E-low.gif new file mode 100644 index 000000000..70bf0d733 Binary files /dev/null and b/docs/domains/equations/GUID-6F53C93C-0634-4E53-8874-5ACBD4C9AA3E-low.gif differ diff --git a/docs/domains/equations/GUID-6FB20CE7-1E2A-4340-995F-0E621DEF0E2D-low.jpg b/docs/domains/equations/GUID-6FB20CE7-1E2A-4340-995F-0E621DEF0E2D-low.jpg new file mode 100644 index 000000000..a70fd7088 Binary files /dev/null and b/docs/domains/equations/GUID-6FB20CE7-1E2A-4340-995F-0E621DEF0E2D-low.jpg differ diff --git a/docs/domains/equations/GUID-749B9421-ABAF-41EA-B8B9-3C9941EF5B72-low.png b/docs/domains/equations/GUID-749B9421-ABAF-41EA-B8B9-3C9941EF5B72-low.png new file mode 100644 index 000000000..d0e3b4927 Binary files /dev/null and b/docs/domains/equations/GUID-749B9421-ABAF-41EA-B8B9-3C9941EF5B72-low.png differ diff --git a/docs/domains/equations/GUID-75532DED-BE44-4D85-B9C0-99C825778ee1.png b/docs/domains/equations/GUID-75532DED-BE44-4D85-B9C0-99C825778ee1.png new file mode 100644 index 000000000..b985bb006 Binary files /dev/null and b/docs/domains/equations/GUID-75532DED-BE44-4D85-B9C0-99C825778ee1.png differ diff --git a/docs/domains/equations/GUID-788A8218-34E5-4625-8E51-A5D36A113D23-low.gif b/docs/domains/equations/GUID-788A8218-34E5-4625-8E51-A5D36A113D23-low.gif new file mode 100644 index 000000000..d61b37fe0 Binary files /dev/null and b/docs/domains/equations/GUID-788A8218-34E5-4625-8E51-A5D36A113D23-low.gif differ diff --git a/docs/domains/equations/GUID-7AE86F5B-8BE2-42D5-B6C7-AF9FF41CCE11-low.jpg b/docs/domains/equations/GUID-7AE86F5B-8BE2-42D5-B6C7-AF9FF41CCE11-low.jpg new file mode 100644 index 000000000..34c3e6996 Binary files /dev/null and b/docs/domains/equations/GUID-7AE86F5B-8BE2-42D5-B6C7-AF9FF41CCE11-low.jpg differ diff --git a/docs/domains/equations/GUID-7CA9B2F2-8D7C-4955-8EA8-D67616FB5B08-low.gif b/docs/domains/equations/GUID-7CA9B2F2-8D7C-4955-8EA8-D67616FB5B08-low.gif new file mode 100644 index 000000000..a43d0e718 Binary files /dev/null and b/docs/domains/equations/GUID-7CA9B2F2-8D7C-4955-8EA8-D67616FB5B08-low.gif differ diff --git a/docs/domains/equations/GUID-7F65198B-719A-44FB-8983-BBD3C196A663-low.jpg b/docs/domains/equations/GUID-7F65198B-719A-44FB-8983-BBD3C196A663-low.jpg new file mode 100644 index 000000000..4892b7a4f Binary files /dev/null and b/docs/domains/equations/GUID-7F65198B-719A-44FB-8983-BBD3C196A663-low.jpg differ diff --git a/docs/domains/equations/GUID-817D9F12-6BD9-4B74-BFA0-39A03D0660C3-low.jpg b/docs/domains/equations/GUID-817D9F12-6BD9-4B74-BFA0-39A03D0660C3-low.jpg new file mode 100644 index 000000000..eb320aa07 Binary files /dev/null and b/docs/domains/equations/GUID-817D9F12-6BD9-4B74-BFA0-39A03D0660C3-low.jpg differ diff --git a/docs/domains/equations/GUID-8742E2B1-94AF-4622-B964-181611E3D1F2-low.jpg b/docs/domains/equations/GUID-8742E2B1-94AF-4622-B964-181611E3D1F2-low.jpg new file mode 100644 index 000000000..7655db245 Binary files /dev/null and b/docs/domains/equations/GUID-8742E2B1-94AF-4622-B964-181611E3D1F2-low.jpg differ diff --git a/docs/domains/equations/GUID-895E6B37-AC54-40D4-B134-E2816B7F30D3-low.gif b/docs/domains/equations/GUID-895E6B37-AC54-40D4-B134-E2816B7F30D3-low.gif new file mode 100644 index 000000000..7220f68cf Binary files /dev/null and b/docs/domains/equations/GUID-895E6B37-AC54-40D4-B134-E2816B7F30D3-low.gif differ diff --git a/docs/domains/equations/GUID-8AD223ED-624A-4390-9514-D8EF20BD04EE-low.gif b/docs/domains/equations/GUID-8AD223ED-624A-4390-9514-D8EF20BD04EE-low.gif new file mode 100644 index 000000000..7d3a12f01 Binary files /dev/null and b/docs/domains/equations/GUID-8AD223ED-624A-4390-9514-D8EF20BD04EE-low.gif differ diff --git a/docs/domains/equations/GUID-8C1F2803-8F8F-4795-BF16-41856C6442CF-low.jpg b/docs/domains/equations/GUID-8C1F2803-8F8F-4795-BF16-41856C6442CF-low.jpg new file mode 100644 index 000000000..caa25cd88 Binary files /dev/null and b/docs/domains/equations/GUID-8C1F2803-8F8F-4795-BF16-41856C6442CF-low.jpg differ diff --git a/docs/domains/equations/GUID-8F2DCE6A-CB54-4CEA-A5EB-937893A3DB34-low.gif b/docs/domains/equations/GUID-8F2DCE6A-CB54-4CEA-A5EB-937893A3DB34-low.gif new file mode 100644 index 000000000..570584179 Binary files /dev/null and b/docs/domains/equations/GUID-8F2DCE6A-CB54-4CEA-A5EB-937893A3DB34-low.gif differ diff --git a/docs/domains/equations/GUID-8F8B1A27-FCBD-4E61-ACC0-459C9EBAE376-low.gif b/docs/domains/equations/GUID-8F8B1A27-FCBD-4E61-ACC0-459C9EBAE376-low.gif new file mode 100644 index 000000000..a3c443bdc Binary files /dev/null and b/docs/domains/equations/GUID-8F8B1A27-FCBD-4E61-ACC0-459C9EBAE376-low.gif differ diff --git a/docs/domains/equations/GUID-904ADCA4-1F33-4C30-90AD-128AA11689FF-low.png b/docs/domains/equations/GUID-904ADCA4-1F33-4C30-90AD-128AA11689FF-low.png new file mode 100644 index 000000000..53012958e Binary files /dev/null and b/docs/domains/equations/GUID-904ADCA4-1F33-4C30-90AD-128AA11689FF-low.png differ diff --git a/docs/domains/equations/GUID-9114D36E-F829-485D-BF04-8747E20120BD-low.gif b/docs/domains/equations/GUID-9114D36E-F829-485D-BF04-8747E20120BD-low.gif new file mode 100644 index 000000000..653aed125 Binary files /dev/null and b/docs/domains/equations/GUID-9114D36E-F829-485D-BF04-8747E20120BD-low.gif differ diff --git a/docs/domains/equations/GUID-93DA36DC-40CA-4C01-B883-DABAB0D37ee1.png b/docs/domains/equations/GUID-93DA36DC-40CA-4C01-B883-DABAB0D37ee1.png new file mode 100644 index 000000000..b985bb006 Binary files /dev/null and b/docs/domains/equations/GUID-93DA36DC-40CA-4C01-B883-DABAB0D37ee1.png differ diff --git a/docs/domains/equations/GUID-96C9ACB0-9A38-4682-85C6-4E71711C32C0-low.gif b/docs/domains/equations/GUID-96C9ACB0-9A38-4682-85C6-4E71711C32C0-low.gif new file mode 100644 index 000000000..b6791b9fe Binary files /dev/null and b/docs/domains/equations/GUID-96C9ACB0-9A38-4682-85C6-4E71711C32C0-low.gif differ diff --git a/docs/domains/equations/GUID-96DF4ACE-8587-423F-B50A-E9A58BE272F9-low.gif b/docs/domains/equations/GUID-96DF4ACE-8587-423F-B50A-E9A58BE272F9-low.gif new file mode 100644 index 000000000..8b9ca33a7 Binary files /dev/null and b/docs/domains/equations/GUID-96DF4ACE-8587-423F-B50A-E9A58BE272F9-low.gif differ diff --git a/docs/domains/equations/GUID-98EC239E-D5C9-4960-834B-827656CF3052-low.gif b/docs/domains/equations/GUID-98EC239E-D5C9-4960-834B-827656CF3052-low.gif new file mode 100644 index 000000000..e9730cf1a Binary files /dev/null and b/docs/domains/equations/GUID-98EC239E-D5C9-4960-834B-827656CF3052-low.gif differ diff --git a/docs/domains/equations/GUID-9932E129-264C-42F7-A75D-00E3705ABB80-low.png b/docs/domains/equations/GUID-9932E129-264C-42F7-A75D-00E3705ABB80-low.png new file mode 100644 index 000000000..1b75bd749 Binary files /dev/null and b/docs/domains/equations/GUID-9932E129-264C-42F7-A75D-00E3705ABB80-low.png differ diff --git a/docs/domains/equations/GUID-9A33FF12-964D-4450-949B-6AB7246435C7-low.gif b/docs/domains/equations/GUID-9A33FF12-964D-4450-949B-6AB7246435C7-low.gif new file mode 100644 index 000000000..0d67ba8b7 Binary files /dev/null and b/docs/domains/equations/GUID-9A33FF12-964D-4450-949B-6AB7246435C7-low.gif differ diff --git a/docs/domains/equations/GUID-9AB7A841-1EEC-49D6-BBF8-5B346FB32C1A-low.jpg b/docs/domains/equations/GUID-9AB7A841-1EEC-49D6-BBF8-5B346FB32C1A-low.jpg new file mode 100644 index 000000000..868120d4d Binary files /dev/null and b/docs/domains/equations/GUID-9AB7A841-1EEC-49D6-BBF8-5B346FB32C1A-low.jpg differ diff --git a/docs/domains/equations/GUID-9B0556B7-20F4-4EC9-875B-F6654CAC0C73-low.gif b/docs/domains/equations/GUID-9B0556B7-20F4-4EC9-875B-F6654CAC0C73-low.gif new file mode 100644 index 000000000..a2ec84826 Binary files /dev/null and b/docs/domains/equations/GUID-9B0556B7-20F4-4EC9-875B-F6654CAC0C73-low.gif differ diff --git a/docs/domains/equations/GUID-9B91DAAE-72DD-4799-9983-12B021993ee1.png b/docs/domains/equations/GUID-9B91DAAE-72DD-4799-9983-12B021993ee1.png new file mode 100644 index 000000000..417cb8172 Binary files /dev/null and b/docs/domains/equations/GUID-9B91DAAE-72DD-4799-9983-12B021993ee1.png differ diff --git a/docs/domains/equations/GUID-9BCB2B87-3DED-4764-A182-30A4FAA4A2E2-low.gif b/docs/domains/equations/GUID-9BCB2B87-3DED-4764-A182-30A4FAA4A2E2-low.gif new file mode 100644 index 000000000..4e0e25a5d Binary files /dev/null and b/docs/domains/equations/GUID-9BCB2B87-3DED-4764-A182-30A4FAA4A2E2-low.gif differ diff --git a/docs/domains/equations/GUID-9DB212E1-03E2-430C-8B1F-8F5CBD4F2ee1.png b/docs/domains/equations/GUID-9DB212E1-03E2-430C-8B1F-8F5CBD4F2ee1.png new file mode 100644 index 000000000..417cb8172 Binary files /dev/null and b/docs/domains/equations/GUID-9DB212E1-03E2-430C-8B1F-8F5CBD4F2ee1.png differ diff --git a/docs/domains/equations/GUID-A3054E0D-CFC1-44E8-89F7-B5A232903EE9-low.jpg b/docs/domains/equations/GUID-A3054E0D-CFC1-44E8-89F7-B5A232903EE9-low.jpg new file mode 100644 index 000000000..f1b5c59cd Binary files /dev/null and b/docs/domains/equations/GUID-A3054E0D-CFC1-44E8-89F7-B5A232903EE9-low.jpg differ diff --git a/docs/domains/equations/GUID-A3089763-5ACF-46DB-AFFF-197043DD5932-low.gif b/docs/domains/equations/GUID-A3089763-5ACF-46DB-AFFF-197043DD5932-low.gif new file mode 100644 index 000000000..0f0b88af2 Binary files /dev/null and b/docs/domains/equations/GUID-A3089763-5ACF-46DB-AFFF-197043DD5932-low.gif differ diff --git a/docs/domains/equations/GUID-A43FDBB0-21D0-4087-AAD9-4289019DD4C3-low.gif b/docs/domains/equations/GUID-A43FDBB0-21D0-4087-AAD9-4289019DD4C3-low.gif new file mode 100644 index 000000000..1cb89f8b8 Binary files /dev/null and b/docs/domains/equations/GUID-A43FDBB0-21D0-4087-AAD9-4289019DD4C3-low.gif differ diff --git a/docs/domains/equations/GUID-A5408434-7126-4EEC-8AD1-856204EBF263-low.jpg b/docs/domains/equations/GUID-A5408434-7126-4EEC-8AD1-856204EBF263-low.jpg new file mode 100644 index 000000000..78d8abfde Binary files /dev/null and b/docs/domains/equations/GUID-A5408434-7126-4EEC-8AD1-856204EBF263-low.jpg differ diff --git a/docs/domains/equations/GUID-A7CA23B7-756F-45C6-85B3-3A8924939D7D-low.jpg b/docs/domains/equations/GUID-A7CA23B7-756F-45C6-85B3-3A8924939D7D-low.jpg new file mode 100644 index 000000000..970b4843d Binary files /dev/null and b/docs/domains/equations/GUID-A7CA23B7-756F-45C6-85B3-3A8924939D7D-low.jpg differ diff --git a/docs/domains/equations/GUID-A984CAB6-AB6E-41AC-885E-DE4A33635480-low.jpg b/docs/domains/equations/GUID-A984CAB6-AB6E-41AC-885E-DE4A33635480-low.jpg new file mode 100644 index 000000000..b2a978d99 Binary files /dev/null and b/docs/domains/equations/GUID-A984CAB6-AB6E-41AC-885E-DE4A33635480-low.jpg differ diff --git a/docs/domains/equations/GUID-AAA6EA17-BA1C-4185-A5F4-8997B64E3BDD-low.gif b/docs/domains/equations/GUID-AAA6EA17-BA1C-4185-A5F4-8997B64E3BDD-low.gif new file mode 100644 index 000000000..14436e3a9 Binary files /dev/null and b/docs/domains/equations/GUID-AAA6EA17-BA1C-4185-A5F4-8997B64E3BDD-low.gif differ diff --git a/docs/domains/equations/GUID-ABAAF79E-E46B-4053-8A64-9CC8B9C84A3F-low.jpg b/docs/domains/equations/GUID-ABAAF79E-E46B-4053-8A64-9CC8B9C84A3F-low.jpg new file mode 100644 index 000000000..45811b95b Binary files /dev/null and b/docs/domains/equations/GUID-ABAAF79E-E46B-4053-8A64-9CC8B9C84A3F-low.jpg differ diff --git a/docs/domains/equations/GUID-ACF0A9F7-1A40-490B-BF70-EE6C63C21738-low.png b/docs/domains/equations/GUID-ACF0A9F7-1A40-490B-BF70-EE6C63C21738-low.png new file mode 100644 index 000000000..cf281cc7b Binary files /dev/null and b/docs/domains/equations/GUID-ACF0A9F7-1A40-490B-BF70-EE6C63C21738-low.png differ diff --git a/docs/domains/equations/GUID-AED001B6-9056-491F-ACBE-E06C82D17ee1.png b/docs/domains/equations/GUID-AED001B6-9056-491F-ACBE-E06C82D17ee1.png new file mode 100644 index 000000000..554076c87 Binary files /dev/null and b/docs/domains/equations/GUID-AED001B6-9056-491F-ACBE-E06C82D17ee1.png differ diff --git a/docs/domains/equations/GUID-B2211D34-A472-4FB8-9CFB-1E11AF4F0ee1.png b/docs/domains/equations/GUID-B2211D34-A472-4FB8-9CFB-1E11AF4F0ee1.png new file mode 100644 index 000000000..554076c87 Binary files /dev/null and b/docs/domains/equations/GUID-B2211D34-A472-4FB8-9CFB-1E11AF4F0ee1.png differ diff --git a/docs/domains/equations/GUID-B229F6A5-0619-4F06-994B-8A734C356ee1.png b/docs/domains/equations/GUID-B229F6A5-0619-4F06-994B-8A734C356ee1.png new file mode 100644 index 000000000..06b72f14d Binary files /dev/null and b/docs/domains/equations/GUID-B229F6A5-0619-4F06-994B-8A734C356ee1.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e10.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e10.png new file mode 100644 index 000000000..a9d77e072 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e10.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e11.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e11.png new file mode 100644 index 000000000..bc1ac8699 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e11.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee1.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee1.png new file mode 100644 index 000000000..131ff7207 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee1.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee2.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee2.png new file mode 100644 index 000000000..bb47a01f1 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee2.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee3.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee3.png new file mode 100644 index 000000000..b17af66f9 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee3.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee4.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee4.png new file mode 100644 index 000000000..159142641 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee4.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee5.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee5.png new file mode 100644 index 000000000..f60afb1ba Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee5.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee6.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee6.png new file mode 100644 index 000000000..b4b6c9476 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee6.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee7.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee7.png new file mode 100644 index 000000000..eda6084f6 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee7.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee8.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee8.png new file mode 100644 index 000000000..93f258312 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee8.png differ diff --git a/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee9.png b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee9.png new file mode 100644 index 000000000..b17af66f9 Binary files /dev/null and b/docs/domains/equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee9.png differ diff --git a/docs/domains/equations/GUID-BF3DF32F-5256-4DFD-9653-FAD2C740BCA5-low.gif b/docs/domains/equations/GUID-BF3DF32F-5256-4DFD-9653-FAD2C740BCA5-low.gif new file mode 100644 index 000000000..64a4e0349 Binary files /dev/null and b/docs/domains/equations/GUID-BF3DF32F-5256-4DFD-9653-FAD2C740BCA5-low.gif differ diff --git a/docs/domains/equations/GUID-CA113DF0-DE46-42A1-99AF-93F6F76E72EA-low.gif b/docs/domains/equations/GUID-CA113DF0-DE46-42A1-99AF-93F6F76E72EA-low.gif new file mode 100644 index 000000000..35553f044 Binary files /dev/null and b/docs/domains/equations/GUID-CA113DF0-DE46-42A1-99AF-93F6F76E72EA-low.gif differ diff --git a/docs/domains/equations/GUID-CD24FF51-197B-40A1-83A8-514788192ee1.png b/docs/domains/equations/GUID-CD24FF51-197B-40A1-83A8-514788192ee1.png new file mode 100644 index 000000000..23f8a8d14 Binary files /dev/null and b/docs/domains/equations/GUID-CD24FF51-197B-40A1-83A8-514788192ee1.png differ diff --git a/docs/domains/equations/GUID-CD24FF51-197B-40A1-83A8-514788192ee2.png b/docs/domains/equations/GUID-CD24FF51-197B-40A1-83A8-514788192ee2.png new file mode 100644 index 000000000..1854bac58 Binary files /dev/null and b/docs/domains/equations/GUID-CD24FF51-197B-40A1-83A8-514788192ee2.png differ diff --git a/docs/domains/equations/GUID-CF961E8B-3127-4493-839A-C045E325BC42-low.jpg b/docs/domains/equations/GUID-CF961E8B-3127-4493-839A-C045E325BC42-low.jpg new file mode 100644 index 000000000..36674edb4 Binary files /dev/null and b/docs/domains/equations/GUID-CF961E8B-3127-4493-839A-C045E325BC42-low.jpg differ diff --git a/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee1.png b/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee1.png new file mode 100644 index 000000000..45f229e16 Binary files /dev/null and b/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee1.png differ diff --git a/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee2.png b/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee2.png new file mode 100644 index 000000000..e4f4b4ba2 Binary files /dev/null and b/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee2.png differ diff --git a/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee3.png b/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee3.png new file mode 100644 index 000000000..ba307693f Binary files /dev/null and b/docs/domains/equations/GUID-D352DB8F-BC76-4A5E-A7CA-5B4CAAA90ee3.png differ diff --git a/docs/domains/equations/GUID-D4002137-8BA4-4D20-871B-550F2C6F9CE8-low.gif b/docs/domains/equations/GUID-D4002137-8BA4-4D20-871B-550F2C6F9CE8-low.gif new file mode 100644 index 000000000..b9a8985d1 Binary files /dev/null and b/docs/domains/equations/GUID-D4002137-8BA4-4D20-871B-550F2C6F9CE8-low.gif differ diff --git a/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee1.png b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee1.png new file mode 100644 index 000000000..a35f523e5 Binary files /dev/null and b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee1.png differ diff --git a/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee2.png b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee2.png new file mode 100644 index 000000000..0f9069655 Binary files /dev/null and b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee2.png differ diff --git a/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee3.png b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee3.png new file mode 100644 index 000000000..7e0976ad3 Binary files /dev/null and b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee3.png differ diff --git a/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee4.png b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee4.png new file mode 100644 index 000000000..00cf06979 Binary files /dev/null and b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee4.png differ diff --git a/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee5.png b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee5.png new file mode 100644 index 000000000..3c5f4e31b Binary files /dev/null and b/docs/domains/equations/GUID-D6A2FFBB-116D-4A37-A278-47F163915ee5.png differ diff --git a/docs/domains/equations/GUID-D703292D-2A37-42C6-B713-E38B801F0114-low.gif b/docs/domains/equations/GUID-D703292D-2A37-42C6-B713-E38B801F0114-low.gif new file mode 100644 index 000000000..6bb49e822 Binary files /dev/null and b/docs/domains/equations/GUID-D703292D-2A37-42C6-B713-E38B801F0114-low.gif differ diff --git a/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee1.png b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee1.png new file mode 100644 index 000000000..45f229e16 Binary files /dev/null and b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee1.png differ diff --git a/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee2.png b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee2.png new file mode 100644 index 000000000..e4f4b4ba2 Binary files /dev/null and b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee2.png differ diff --git a/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee3.png b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee3.png new file mode 100644 index 000000000..d8dd46dd0 Binary files /dev/null and b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee3.png differ diff --git a/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee4.png b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee4.png new file mode 100644 index 000000000..74c4fb66b Binary files /dev/null and b/docs/domains/equations/GUID-D797E8FA-B0CE-417C-98F1-896CDFB4Fee4.png differ diff --git a/docs/domains/equations/GUID-D90F2CB0-58B4-42F5-A1F9-FD1EA859DD44-low.png b/docs/domains/equations/GUID-D90F2CB0-58B4-42F5-A1F9-FD1EA859DD44-low.png new file mode 100644 index 000000000..d1a93ccf1 Binary files /dev/null and b/docs/domains/equations/GUID-D90F2CB0-58B4-42F5-A1F9-FD1EA859DD44-low.png differ diff --git a/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee1.png b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee1.png new file mode 100644 index 000000000..a35f523e5 Binary files /dev/null and b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee1.png differ diff --git a/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee2.png b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee2.png new file mode 100644 index 000000000..0f9069655 Binary files /dev/null and b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee2.png differ diff --git a/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee3.png b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee3.png new file mode 100644 index 000000000..7e0976ad3 Binary files /dev/null and b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee3.png differ diff --git a/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee4.png b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee4.png new file mode 100644 index 000000000..00cf06979 Binary files /dev/null and b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee4.png differ diff --git a/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee5.png b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee5.png new file mode 100644 index 000000000..3c5f4e31b Binary files /dev/null and b/docs/domains/equations/GUID-DA21ECDC-F63E-4971-BA3F-492E69335ee5.png differ diff --git a/docs/domains/equations/GUID-DF275A8A-05D4-49D9-9031-E4A9382C284C-low.gif b/docs/domains/equations/GUID-DF275A8A-05D4-49D9-9031-E4A9382C284C-low.gif new file mode 100644 index 000000000..829f8c827 Binary files /dev/null and b/docs/domains/equations/GUID-DF275A8A-05D4-49D9-9031-E4A9382C284C-low.gif differ diff --git a/docs/domains/equations/GUID-E3193631-248D-4D18-A094-30BB6FF50687-low.jpg b/docs/domains/equations/GUID-E3193631-248D-4D18-A094-30BB6FF50687-low.jpg new file mode 100644 index 000000000..08b88f54e Binary files /dev/null and b/docs/domains/equations/GUID-E3193631-248D-4D18-A094-30BB6FF50687-low.jpg differ diff --git a/docs/domains/equations/GUID-E48BA1ED-9ABF-487F-80F3-1FA1E0F6EABC-low.jpg b/docs/domains/equations/GUID-E48BA1ED-9ABF-487F-80F3-1FA1E0F6EABC-low.jpg new file mode 100644 index 000000000..4588f55a6 Binary files /dev/null and b/docs/domains/equations/GUID-E48BA1ED-9ABF-487F-80F3-1FA1E0F6EABC-low.jpg differ diff --git a/docs/domains/equations/GUID-E5BC391B-F8BC-45E0-9A58-84319AC0B246-low.gif b/docs/domains/equations/GUID-E5BC391B-F8BC-45E0-9A58-84319AC0B246-low.gif new file mode 100644 index 000000000..b85132a86 Binary files /dev/null and b/docs/domains/equations/GUID-E5BC391B-F8BC-45E0-9A58-84319AC0B246-low.gif differ diff --git a/docs/domains/equations/GUID-EA1939AE-5968-4E6A-8396-6F44E73939AF-low.png b/docs/domains/equations/GUID-EA1939AE-5968-4E6A-8396-6F44E73939AF-low.png new file mode 100644 index 000000000..d91ad19c3 Binary files /dev/null and b/docs/domains/equations/GUID-EA1939AE-5968-4E6A-8396-6F44E73939AF-low.png differ diff --git a/docs/domains/equations/GUID-EA77C856-3F93-40ED-AB2A-0F1BD8C4CE7A-low.jpg b/docs/domains/equations/GUID-EA77C856-3F93-40ED-AB2A-0F1BD8C4CE7A-low.jpg new file mode 100644 index 000000000..453a3f4ed Binary files /dev/null and b/docs/domains/equations/GUID-EA77C856-3F93-40ED-AB2A-0F1BD8C4CE7A-low.jpg differ diff --git a/docs/domains/equations/GUID-F01C1454-13EC-4D30-8E73-8E41755B8AF2-low.gif b/docs/domains/equations/GUID-F01C1454-13EC-4D30-8E73-8E41755B8AF2-low.gif new file mode 100644 index 000000000..8ae60b4fc Binary files /dev/null and b/docs/domains/equations/GUID-F01C1454-13EC-4D30-8E73-8E41755B8AF2-low.gif differ diff --git a/docs/domains/equations/GUID-F85E385E-ACAD-4DC6-95EC-7C8A85836AAD-low.gif b/docs/domains/equations/GUID-F85E385E-ACAD-4DC6-95EC-7C8A85836AAD-low.gif new file mode 100644 index 000000000..7522a48c4 Binary files /dev/null and b/docs/domains/equations/GUID-F85E385E-ACAD-4DC6-95EC-7C8A85836AAD-low.gif differ diff --git a/docs/domains/equations/GUID-F928F918-624A-444A-BB76-7D26D1E1BC62-low.gif b/docs/domains/equations/GUID-F928F918-624A-444A-BB76-7D26D1E1BC62-low.gif new file mode 100644 index 000000000..0d67ba8b7 Binary files /dev/null and b/docs/domains/equations/GUID-F928F918-624A-444A-BB76-7D26D1E1BC62-low.gif differ diff --git a/docs/domains/equations/GUID-FB387120-1A86-45B9-BE20-97247EF0ABB5-low.gif b/docs/domains/equations/GUID-FB387120-1A86-45B9-BE20-97247EF0ABB5-low.gif new file mode 100644 index 000000000..4e351aa1e Binary files /dev/null and b/docs/domains/equations/GUID-FB387120-1A86-45B9-BE20-97247EF0ABB5-low.gif differ diff --git a/docs/domains/matrix-storage.rst b/docs/domains/matrix-storage.rst new file mode 100644 index 000000000..3c8c7a388 --- /dev/null +++ b/docs/domains/matrix-storage.rst @@ -0,0 +1,373 @@ +.. _matrix-storage: + +Matrix Storage +============== + + +.. container:: + + + The oneMKL BLAS and LAPACK routines for DPC++ use several matrix and + vector storage formats. These are the same formats used in + traditional Fortran BLAS/LAPACK. + + + .. container:: section + :name: GUID-5DD12E9B-8079-4EFD-8B91-5CF134D55082 + + + .. rubric:: General Matrix + :name: general-matrix + :class: sectiontitle + + + A general matrix ``A`` of ``m`` rows and ``n`` columns with + leading dimension ``lda`` is represented as a one dimensional + array ``a`` of size of at least ``lda`` \* ``n``. Before entry in + any BLAS function using a general matrix, the leading ``m`` by + ``n`` part of the array ``a`` must contain the matrix ``A``. The + elements of each column are contiguous in memory while the + elements of each row are at distance ``lda`` from the element in + the same row and the previous column. + + + Visually, the matrix + + + |image0| + + + is stored in memory as an array + + + |image1| + + + .. container:: section + :name: GUID-150ACC6B-BB73-4E6E-A7F7-9CE07707FA0E + + + .. rubric:: Triangular Matrix + :name: triangular-matrix + :class: sectiontitle + + + A triangular matrix ``A`` of ``n`` rows and ``n`` columns with + leading dimension ``lda`` is represented as a one dimensional + array ``a``, of a size of at least ``lda`` \* ``n``. The elements + of each column are contiguous in memory while the elements of each + row are at distance ``lda`` from the element in the same row and + the previous column. + + + Before entry in any BLAS function using a triangular matrix, + + + - If ``upper_lower = uplo::upper``, the leading ``n`` by ``n`` + upper triangular part of the array ``a`` must contain the upper + triangular part of the matrix ``A``. The strictly lower + triangular part of the array ``a`` is not referenced. In other + words, the matrix + + + |image2| + + + is stored in memory as the array + + + |image3| + + + - If ``upper_lower = uplo::lower``, the leading ``n`` by ``n`` + lower triangular part of the array ``a`` must contain the lower + triangular part of the matrix ``A``. The strictly upper + triangular part of the array ``a`` is not referenced. That is, + the matrix + + + |image4| + + + is stored in memory as the array + + + |image5| + + + .. container:: section + :name: GUID-4A6389BD-0396-4C6D-8AA4-C59EDAC7A991 + + + .. rubric:: Band Matrix + :name: band-matrix + :class: sectiontitle + + + A general band matrix ``A`` of ``m`` rows and ``n`` columns with + ``kl`` sub-diagonals, ``ku`` super-diagonals, and leading + dimension ``lda`` is represented as a one dimensional array ``a`` + of a size of at least ``lda`` \* ``n``. + + + Before entry in any BLAS function using a general band matrix, the + leading (``kl`` + ``ku`` + 1\ ``)`` by ``n`` part of the array + ``a`` must contain the matrix ``A``. This matrix must be supplied + column-by-column, with the main diagonal of the matrix in row + ``ku`` of the array (0-based indexing), the first super-diagonal + starting at position 1 in row (``ku`` - 1), the first sub-diagonal + starting at position 0 in row (``ku`` + 1), and so on. Elements in + the array ``a`` that do not correspond to elements in the band + matrix (such as the top left ``ku`` by ``ku`` triangle) are not + referenced. + + + Visually, the matrix ``A`` = + + + |image6| + + + is stored in memory as an array + + + |image7| + The following program segment transfers a band matrix from + conventional full matrix storage (variable ``matrix``, with + leading dimension ``ldm``) to band storage (variable ``a``, with + leading dimension ``lda``): + + + :: + + + for (j = 0; j < n; j++) { + k = ku – j; + for (i = max(0, j – ku); i < min(m, j + kl + 1); i++) { + a[(k + i) + j * lda] = matrix[i + j * ldm]; + } + } + + + .. container:: section + :name: GUID-D85FAA87-6868-4DCA-BD38-9C4F4214BD52 + + + .. rubric:: Triangular Band Matrix + :name: triangular-band-matrix + :class: sectiontitle + + + A triangular band matrix ``A`` of ``n`` rows and ``n`` columns + with ``k`` sub/super-diagonals and leading dimension ``lda`` is + represented as a one dimensional array ``a`` of size at least + ``lda`` \* ``n``. + + + Before entry in any BLAS function using a triangular band matrix, + + + - If ``upper_lower = uplo::upper``, the leading (``k`` + 1) by + ``n`` part of the array ``a`` must contain the upper triangular + band part of the matrix ``A``. This matrix must be supplied + column-by-column with the main diagonal of the matrix in row + (``k``) of the array, the first super-diagonal starting at + position 1 in row (``k`` - 1), and so on. Elements in the array + ``a`` that do not correspond to elements in the triangular band + matrix (such as the top left ``k`` by ``k`` triangle) are not + referenced. + + + Visually, the matrix + + + |image8| + + + is stored as an array + + + .. container:: fignone + :name: GUID-CBD17940-8F30-4779-AEB3-C17E9ADB60EC + + + |image9| + + + The following program segment transfers a band matrix from + conventional full matrix storage (variable ``matrix``, with + leading dimension ``ldm``) to band storage (variable ``a``, + with leading dimension ``lda``): + + + :: + + + for (j = 0; j < n; j++) { + m = k – j; + for (i = max(0, j – k); i <= j; i++) { + a[(m + i) + j * lda] = matrix[i + j * ldm]; + } + } + + + - If ``upper_lower = uplo::lower``, the leading (``k`` + 1) by + ``n`` part of the array ``a`` must contain the upper triangular + band part of the matrix ``A``. This matrix must be supplied + column-by-column with the main diagonal of the matrix in row 0 + of the array, the first sub-diagonal starting at position 0 in + row 1, and so on. Elements in the array ``a`` that do not + correspond to elements in the triangular band matrix (such as + the bottom right ``k`` by ``k`` triangle) are not referenced. + + + That is, the matrix + + + |image10| + + + is stored as the array + + + .. container:: fignone + :name: GUID-D89A1D4C-831C-4D8E-AD9F-0DFB968841E1 + + + |image11| + + + The following program segment transfers a band matrix from + conventional full matrix storage (variable ``matrix``, with + leading dimension ``ldm``) to band storage (variable ``a``, + with leading dimension ``lda``): + + + :: + + + for (j = 0; j < n; j++) { + m = –j; + for (i = j; i < min(n, j + k + 1); i++) { + a[(m + i) + j * lda] = matrix[i + j * ldm]; + } + } + + + .. container:: section + :name: GUID-41C95365-5CE1-46F9-869C-27647E5ABE2B + + + .. rubric:: Packed Triangular Matrix + :name: packed-triangular-matrix + :class: sectiontitle + + + A triangular matrix ``A`` of ``n`` rows and ``n`` columns is + represented in packed format as a one dimensional array ``a`` of + size at least (``n``\ \*(``n`` + 1))/2. All elements in the upper + or lower part of the matrix ``A`` are stored contiguously in the + array ``a``. + + + Before entry in any BLAS function using a triangular packed + matrix, + + + - If ``upper_lower = uplo::upper``, the first (``n``\ \*(``n`` + + 1))/2 elements in the array ``a`` must contain the upper + triangular part of the matrix ``A`` packed sequentially, column + by column so that ``a``\ [0] contains ``A``\ :sub:`11`, + ``a``\ [1] and ``a``\ [2] contain ``A``\ :sub:`12` and + ``A``\ :sub:`22` respectively, and so on. Hence, the matrix + + + |image12| + + + is stored as the array + + + |image13| + + + - If ``upper_lower = uplo::lower``, the first (``n``\ \*(``n`` + + 1))/2 elements in the array ``a`` must contain the lower + triangular part of the matrix ``A`` packed sequentially, column + by column so that ``a``\ [0] contains ``A``\ :sub:`11`, + ``a``\ [1] and ``a``\ [2] contain ``A``\ :sub:`21` and + ``A``\ :sub:`31` respectively, and so on. The matrix + + + |image14| + + + is stored as the array + + + |image15| + + + .. container:: section + :name: GUID-44CE5D09-0117-46D1-B6EA-333D4550A457 + + + .. rubric:: Vector + :name: vector + :class: sectiontitle + + + A vector ``X`` of ``n`` elements with increment ``incx`` is + represented as a one dimensional array ``x`` of size at least (1 + + (``n`` - 1) \* abs(``incx``)). + + + Visually, the vector + + + |image16| + + + is stored in memory as an array + + + .. container:: fignone + :name: GUID-6929FFA1-5209-4D51-A2B8-CCA373841258 + + + |image17| + + + **Parent topic:** :ref:`onemkl` + + +.. |image0| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee1.png + :class: img-middle +.. |image1| image:: equations/GUID-9932E129-264C-42F7-A75D-00E3705ABB80-low.png +.. |image2| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee2.png + :class: img-middle +.. |image3| image:: equations/GUID-904ADCA4-1F33-4C30-90AD-128AA11689FF-low.png +.. |image4| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee3.png + :class: img-middle +.. |image5| image:: equations/GUID-2F91B385-0AC2-41D3-AE61-48F63A7DBB02-low.png +.. |image6| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee4.png + :class: img-middle +.. |image7| image:: equations/GUID-ACF0A9F7-1A40-490B-BF70-EE6C63C21738-low.png +.. |image8| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee5.png + :class: img-middle +.. |image9| image:: equations/GUID-5193801D-8E3B-43A2-989E-09A8431FD34E-low.png +.. |image10| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee6.png + :class: img-middle +.. |image11| image:: equations/GUID-11B96BA7-C321-446D-A7B6-8D84C8CBC076-low.png +.. |image12| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee7.png + :class: img-middle +.. |image13| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee8.png + :class: img-middle +.. |image14| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15ee9.png + :class: img-middle +.. |image15| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e10.png + :class: img-middle +.. |image16| image:: equations/GUID-B9AEF80A-AD5F-4B59-9F21-60672FB15e11.png + :class: img-middle +.. |image17| image:: equations/GUID-EA1939AE-5968-4E6A-8396-6F44E73939AF-low.png + diff --git a/docs/index.rst b/docs/index.rst new file mode 100644 index 000000000..aba71bd0d --- /dev/null +++ b/docs/index.rst @@ -0,0 +1,21 @@ +.. + Copyright 2020 Intel Corporation + +.. _onemkl: + +***************** +oneMKL Interfaces +***************** + +Introduction +============ + +oneMKL Interfaces is an open-source implementation of oneMKL Data Parallel C++ (DPC++) interfaces accordingly to `oneMKL specification `_ that can work with multiple devices (backends) using device specific libraries underneath + + +.. toctree:: + :maxdepth: 2 + + onemkl-datatypes.rst + domains/matrix-storage.rst + domains/blas/blas.rst diff --git a/docs/onemkl-datatypes.rst b/docs/onemkl-datatypes.rst new file mode 100755 index 000000000..d1b20714b --- /dev/null +++ b/docs/onemkl-datatypes.rst @@ -0,0 +1,195 @@ +.. _onemkl_datatypes: + +oneMKL defined datatypes +======================== + + +.. container:: + + + oneMKL BLAS and LAPACK for Data Parallel C++ (DPC++) introduces + several new enumeration data types, which are type-safe versions of + the traditional Fortran characters in BLAS and LAPACK. They are + declared in ``types.hpp``, which is included automatically when + you include ``onemkl.hpp``. Like + all oneMKL DPC++ functionality, they belong to the + namespace ``onemkl``. + + + Each enumeration value comes with two names: A single-character name + (the traditional BLAS/LAPACK character) and a longer, descriptive + name. The two names are exactly equivalent and may be used + interchangeably. + + + .. container:: section + :name: GUID-97715A19-7DDE-4738-9E7A-53554E5B702B + + + .. rubric:: transpose + :name: transpose + :class: sectiontitle + + + The ``transpose`` type specifies whether an input matrix should be + transposed and/or conjugated. It can take the following values: + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Short Name + - Long Name + - Description + * - ``transpose::N`` + - ``transpose::nontrans`` + - Do not transpose or conjugate the matrix. + * - ``transpose::T`` + - ``transpose::trans`` + - Transpose the matrix. + * - ``transpose::C`` + - ``transpose::conjtrans`` + - Perform Hermitian transpose (transpose and conjugate). Only applicable to complex matrices. + + + + + .. container:: section + :name: GUID-DD566CC1-62E5-4AF1-A407-FB4E979B753D + + + .. rubric:: uplo + :name: uplo + :class: sectiontitle + + + The ``uplo`` type specifies whether the lower or upper triangle of + a triangular, symmetric, or Hermitian matrix should be accessed. + It can take the following values: + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Short Name + - Long Name + - Description + * - ``uplo::U`` + - ``uplo::upper`` + - Access the upper triangle of the matrix. + * - ``uplo::L`` + - ``uplo::lower`` + - Access the lower triangle of the matrix. + + + + + In both cases, elements that are not in the selected triangle are + not accessed or updated. + + + .. container:: section + :name: GUID-419CF945-4868-400D-B05C-50ABABD73961 + + + .. rubric:: diag + :name: diag + :class: sectiontitle + + + The ``diag`` type specifies the values on the diagonal of a + triangular matrix. It can take the following values: + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Short Name + - Long Name + - Description + * - ``diag::N`` + - ``diag::nonunit`` + - The matrix is not unit triangular. The diagonal entries are stored with the matrix data. + * - ``diag::U`` + - ``diag::unit`` + - The matrix is unit triangular (the diagonal entries are all 1s). The diagonal entries in the matrix data are not accessed. + + + + + .. container:: section + :name: GUID-538307BC-A47D-4290-B5B4-CB54CFB25242 + + + .. rubric:: side + :name: side + :class: sectiontitle + + + The ``side`` type specifies the order of matrix multiplication + when one matrix has a special form (triangular, symmetric, or + Hermitian): + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Short Name + - Long Name + - Description + * - ``side::L`` + - ``side::left`` + - The special form matrix is on the left in the multiplication. + * - ``side::R`` + - ``side::right`` + - The special form matrix is on the right in the multiplication. + + + + + .. container:: section + :name: GUID-D25C1BB5-81B8-4591-A815-C881B59E7C5B + + + .. rubric:: offset + :name: offset + :class: sectiontitle + + + The ``offset`` type specifies whether the offset to apply to an + output matrix is a fix offset, column offset or row offset. It can + take the following values + + + .. container:: tablenoborder + + + .. list-table:: + :header-rows: 1 + + * - Short Name + - Long Name + - Description + * - ``offset::F`` + - ``offset::fix`` + - The offset to apply to the output matrix is fix, all the inputs in the ``C_offset`` matrix has the same value given by the first element in the ``co`` array. + * - ``offset::C`` + - ``offset::column`` + - The offset to apply to the output matrix is a column offset, that is to say all the columns in the ``C_offset`` matrix are the same and given by the elements in the ``co`` array. + * - ``offset::R`` + - ``offset::row`` + - The offset to apply to the output matrix is a row offset, that is to say all the rows in the ``C_offset`` matrix are the same and given by the elements in the ``co`` array. + + **Parent topic:** :ref:`onemkl` diff --git a/include/onemkl/blas/blas.hpp b/include/onemkl/blas/blas.hpp new file mode 100644 index 000000000..6dd01af6c --- /dev/null +++ b/include/onemkl/blas/blas.hpp @@ -0,0 +1,1423 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BLAS_HPP_ +#define _ONEMKL_BLAS_HPP_ + +#include +#include +#include + +#include "onemkl/types.hpp" + +#include "onemkl/detail/backends_selector.hpp" + +#include "onemkl/blas/predicates.hpp" + +#include "onemkl/blas/detail/blas_loader.hpp" +#include "onemkl/blas/detail/mklcpu/blas_ct.hpp" +#include "onemkl/blas/detail/mklgpu/blas_ct.hpp" + +namespace onemkl { +namespace blas { + +static inline void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + detail::asum(select_backend(queue), queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +static inline void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + detail::asum(select_backend(queue), queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +static inline void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + detail::asum(select_backend(queue), queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +static inline void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + detail::asum(select_backend(queue), queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + detail::axpy(select_backend(queue), queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + detail::axpy(select_backend(queue), queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + detail::axpy(select_backend(queue), queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + detail::axpy(select_backend(queue), queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +static inline void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + detail::copy(select_backend(queue), queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +static inline void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + detail::copy(select_backend(queue), queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +static inline void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + detail::copy(select_backend(queue), queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +static inline void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + detail::copy(select_backend(queue), queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + detail::dot(select_backend(queue), queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + detail::dot(select_backend(queue), queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + detail::dot(select_backend(queue), queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotc_precondition(queue, n, x, incx, y, incy, result); + detail::dotc(select_backend(queue), queue, n, x, incx, y, incy, result); + dotc_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotc_precondition(queue, n, x, incx, y, incy, result); + detail::dotc(select_backend(queue), queue, n, x, incx, y, incy, result); + dotc_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotu_precondition(queue, n, x, incx, y, incy, result); + detail::dotu(select_backend(queue), queue, n, x, incx, y, incy, result); + dotu_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotu_precondition(queue, n, x, incx, y, incy, result); + detail::dotu(select_backend(queue), queue, n, x, incx, y, incy, result); + dotu_postcondition(queue, n, x, incx, y, incy, result); +} + +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + detail::gbmv(select_backend(queue), queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + detail::gbmv(select_backend(queue), queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + detail::gbmv(select_backend(queue), queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + detail::gbmv(select_backend(queue), queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::gemm(select_backend(queue), queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::gemm(select_backend(queue), queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::gemm(select_backend(queue), queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::gemm(select_backend(queue), queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::gemm(select_backend(queue), queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + detail::gemv(select_backend(queue), queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + detail::gemv(select_backend(queue), queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + detail::gemv(select_backend(queue), queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + detail::gemv(select_backend(queue), queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + ger_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + detail::ger(select_backend(queue), queue, m, n, alpha, x, incx, y, incy, a, lda); + ger_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + ger_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + detail::ger(select_backend(queue), queue, m, n, alpha, x, incx, y, incy, a, lda); + ger_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda) { + gerc_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + detail::gerc(select_backend(queue), queue, m, n, alpha, x, incx, y, incy, a, lda); + gerc_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda) { + gerc_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + detail::gerc(select_backend(queue), queue, m, n, alpha, x, incx, y, incy, a, lda); + gerc_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda) { + geru_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + detail::geru(select_backend(queue), queue, m, n, alpha, x, incx, y, incy, a, lda); + geru_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda) { + geru_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + detail::geru(select_backend(queue), queue, m, n, alpha, x, incx, y, incy, a, lda); + geru_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + detail::hbmv(select_backend(queue), queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); + hbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + detail::hbmv(select_backend(queue), queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); + hbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + hemm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + detail::hemm(select_backend(queue), queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); + hemm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + hemm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + detail::hemm(select_backend(queue), queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); + hemm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hemv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + detail::hemv(select_backend(queue), queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); + hemv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hemv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + detail::hemv(select_backend(queue), queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); + hemv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + her_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + detail::her(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a, lda); + her_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +static inline void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + her_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + detail::her(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a, lda); + her_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +static inline void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda) { + her2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + detail::her2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + her2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda) { + her2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + detail::her2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + her2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + her2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::her2k(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); + her2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + her2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::her2k(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); + her2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + herk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + detail::herk(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); + herk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +static inline void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + herk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + detail::herk(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); + herk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +static inline void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + hpmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + detail::hpmv(select_backend(queue), queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + hpmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +static inline void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + hpmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + detail::hpmv(select_backend(queue), queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + hpmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +static inline void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { + hpr_precondition(queue, upper_lower, n, alpha, x, incx, a); + detail::hpr(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a); + hpr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +static inline void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { + hpr_precondition(queue, upper_lower, n, alpha, x, incx, a); + detail::hpr(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a); + hpr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +static inline void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a) { + hpr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + detail::hpr2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a); + hpr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +static inline void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a) { + hpr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + detail::hpr2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a); + hpr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + detail::iamax(select_backend(queue), queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + detail::iamax(select_backend(queue), queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + detail::iamax(select_backend(queue), queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + detail::iamax(select_backend(queue), queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + detail::iamin(select_backend(queue), queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + detail::iamin(select_backend(queue), queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + detail::iamin(select_backend(queue), queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + detail::iamin(select_backend(queue), queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + detail::nrm2(select_backend(queue), queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + detail::nrm2(select_backend(queue), queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + detail::nrm2(select_backend(queue), queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + detail::nrm2(select_backend(queue), queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +static inline void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + detail::rot(select_backend(queue), queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +static inline void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, double c, + double s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + detail::rot(select_backend(queue), queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +static inline void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, float c, + float s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + detail::rot(select_backend(queue), queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +static inline void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + double c, double s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + detail::rot(select_backend(queue), queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { + rotg_precondition(queue, a, b, c, s); + detail::rotg(select_backend(queue), queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { + rotg_precondition(queue, a, b, c, s); + detail::rotg(select_backend(queue), queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + rotg_precondition(queue, a, b, c, s); + detail::rotg(select_backend(queue), queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + rotg_precondition(queue, a, b, c, s); + detail::rotg(select_backend(queue), queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +static inline void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + rotm_precondition(queue, n, x, incx, y, incy, param); + detail::rotm(select_backend(queue), queue, n, x, incx, y, incy, param); + rotm_postcondition(queue, n, x, incx, y, incy, param); +} + +static inline void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + rotm_precondition(queue, n, x, incx, y, incy, param); + detail::rotm(select_backend(queue), queue, n, x, incx, y, incy, param); + rotm_postcondition(queue, n, x, incx, y, incy, param); +} + +static inline void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m) { + rotmg_precondition(queue, d1, d2, x1, y1, param); + detail::rotmg(select_backend(queue), queue, d1, d2, x1, y1, param); + rotmg_postcondition(queue, d1, d2, x1, y1, param); +} + +static inline void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + double y1, cl::sycl::buffer ¶m) { + rotmg_precondition(queue, d1, d2, x1, y1, param); + detail::rotmg(select_backend(queue), queue, d1, d2, x1, y1, param); + rotmg_postcondition(queue, d1, d2, x1, y1, param); +} + +static inline void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + sbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + detail::sbmv(select_backend(queue), queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); + sbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + sbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + detail::sbmv(select_backend(queue), queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); + sbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + detail::scal(select_backend(queue), queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +static inline void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + detail::scal(select_backend(queue), queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +static inline void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + detail::scal(select_backend(queue), queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +static inline void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + detail::scal(select_backend(queue), queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +static inline void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + detail::scal(select_backend(queue), queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +static inline void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + detail::scal(select_backend(queue), queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +static inline void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + sdsdot_precondition(queue, n, sb, x, incx, y, incy, result); + detail::sdsdot(select_backend(queue), queue, n, sb, x, incx, y, incy, result); + sdsdot_postcondition(queue, n, sb, x, incx, y, incy, result); +} + +static inline void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, + std::int64_t incy) { + spmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + detail::spmv(select_backend(queue), queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + spmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +static inline void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, + std::int64_t incy) { + spmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + detail::spmv(select_backend(queue), queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + spmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +static inline void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { + spr_precondition(queue, upper_lower, n, alpha, x, incx, a); + detail::spr(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a); + spr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +static inline void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { + spr_precondition(queue, upper_lower, n, alpha, x, incx, a); + detail::spr(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a); + spr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +static inline void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { + spr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + detail::spr2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a); + spr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +static inline void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { + spr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + detail::spr2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a); + spr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +static inline void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + detail::swap(select_backend(queue), queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +static inline void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + detail::swap(select_backend(queue), queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +static inline void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + detail::swap(select_backend(queue), queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +static inline void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + detail::swap(select_backend(queue), queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + detail::symm(select_backend(queue), queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + detail::symm(select_backend(queue), queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + detail::symm(select_backend(queue), queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + detail::symm(select_backend(queue), queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + symv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + detail::symv(select_backend(queue), queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); + symv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + symv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + detail::symv(select_backend(queue), queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); + symv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +static inline void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { + syr_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + detail::syr(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a, lda); + syr_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +static inline void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { + syr_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + detail::syr(select_backend(queue), queue, upper_lower, n, alpha, x, incx, a, lda); + syr_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +static inline void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + syr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + detail::syr2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + syr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + syr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + detail::syr2(select_backend(queue), queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + syr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::syr2k(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::syr2k(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::syr2k(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + detail::syr2k(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, float beta, cl::sycl::buffer &c, + std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + detail::syrk(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, double beta, cl::sycl::buffer &c, + std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + detail::syrk(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + detail::syrk(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + detail::syrk(select_backend(queue), queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + detail::tbsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + detail::tpsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trmm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trmm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trmm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trmm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trmv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trsm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trsm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trsm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + detail::trsm(select_backend(queue), queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + detail::trsv(select_backend(queue), queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +} //namespace blas +} //namespace onemkl + +#endif //_ONEMKL_BLAS_LOADER_HPP_ diff --git a/include/onemkl/blas/detail/blas_loader.hpp b/include/onemkl/blas/detail/blas_loader.hpp new file mode 100644 index 000000000..137c1fcf2 --- /dev/null +++ b/include/onemkl/blas/detail/blas_loader.hpp @@ -0,0 +1,756 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BLAS_LOADER_HPP_ +#define _ONEMKL_BLAS_LOADER_HPP_ + +#include +#include +#include + +#include + +namespace onemkl { +namespace blas { +namespace detail { + +void herk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void herk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx); +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx); +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void spr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); +void spr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); + +void gemm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); +void gemm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); +void gemm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); +void gemm_batch( + char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, std::int64_t stride_c, + std::int64_t batch_size); +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + double beta, cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + double beta, cl::sycl::buffer &c, std::int64_t ldc); +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void her2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void her2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void hbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void hbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, float c, float s); +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, double c, double s); +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, float c, float s); +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, double c, double s); + +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy); +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy); +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void gerc(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void gerc(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void her(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void her(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void hpr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +void hpr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + offset offsetc, std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, int8_t ao, + cl::sycl::buffer &b, std::int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, std::int64_t ldc, cl::sycl::buffer &co); +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, half beta, cl::sycl::buffer &c, std::int64_t ldc); + +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + +void hpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void hpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void spmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy); +void spmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy); + +void rotmg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m); +void rotmg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m); + +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void geru(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void geru(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); +void syr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void ger(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); +void ger(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void dotu(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +void dotu(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void hemm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); +void hemm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void hpr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); +void hpr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); + +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); + +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void dotc(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +void dotc(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void syr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); +void syr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); + +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void symv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); +void symv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); + +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void hemv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void hemv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + +void sbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +void sbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); + +void spr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); +void spr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); + +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); + +void rotm(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); +void rotm(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); + +void dot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +void dot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +void dot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + +void sdsdot(char *libname, cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &result); + +void her2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + float beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); +void her2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +} //namespace detail +} //namespace blas +} //namespace onemkl + +#endif //_ONEMKL_BLAS_LOADER_HPP_ diff --git a/include/onemkl/blas/detail/mklcpu/blas_ct.hpp b/include/onemkl/blas/detail/mklcpu/blas_ct.hpp new file mode 100644 index 000000000..3e9987614 --- /dev/null +++ b/include/onemkl/blas/detail/mklcpu/blas_ct.hpp @@ -0,0 +1,2405 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +// +// Generated based on onemkl/blas/blas.hpp +// + +#ifndef _DETAIL_MKLCPU_BLAS_HPP__ +#define _DETAIL_MKLCPU_BLAS_HPP__ + +#include +#include + +#include "onemkl/detail/backends.hpp" +#include "onemkl/detail/libraries.hpp" +#include "onemkl/types.hpp" + +#include "onemkl_blas_mklcpu.hpp" + +namespace onemkl { +namespace blas { + +template +static inline void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + syr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::syr2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + syr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + syr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::syr2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + syr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklcpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + double alpha, cl::sycl::buffer &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklcpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklcpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklcpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklcpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + double alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklcpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a); +template <> +void spr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { + spr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklcpu::spr(queue, upper_lower, n, alpha, x, incx, a); + spr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a); +template <> +void spr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { + spr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklcpu::spr(queue, upper_lower, n, alpha, x, incx, a); + spr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void hpmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + hpmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklcpu::hpmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + hpmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void hpmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + hpmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklcpu::hpmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + hpmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, float beta, cl::sycl::buffer &c, + std::int64_t ldc); +template <> +void syrk(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, + std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklcpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, double beta, cl::sycl::buffer &c, + std::int64_t ldc); +template <> +void syrk(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, + std::int64_t lda, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklcpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syrk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklcpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syrk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklcpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void her2( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + her2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::her2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + her2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void her2( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + her2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::her2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + her2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hbmv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::hbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + hbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hbmv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::hbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + hbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, float c, float s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklcpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, double c, + double s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, double c, double s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklcpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, float c, + float s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + float c, float s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklcpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + double c, double s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + double c, double s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklcpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklcpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, + double alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklcpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklcpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklcpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void gerc( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + gerc_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::gerc(queue, m, n, alpha, x, incx, y, incy, a, lda); + gerc_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void gerc( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + gerc_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::gerc(queue, m, n, alpha, x, incx, y, incy, a, lda); + gerc_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void syr2k(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void syr2k(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syr2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syr2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gemv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gemv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void gemv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void gemv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +template <> +void her(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a, + std::int64_t lda) { + her_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklcpu::her(queue, upper_lower, n, alpha, x, incx, a, lda); + her_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +template <> +void her(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a, + std::int64_t lda) { + her_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklcpu::her(queue, upper_lower, n, alpha, x, incx, a, lda); + her_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +template <> +void hpr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a) { + hpr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklcpu::hpr(queue, upper_lower, n, alpha, x, incx, a); + hpr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +template <> +void hpr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a) { + hpr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklcpu::hpr(queue, upper_lower, n, alpha, x, incx, a); + hpr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, + std::int64_t incy); +template <> +void spmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + spmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklcpu::spmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + spmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, + std::int64_t incy); +template <> +void spmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + spmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklcpu::spmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + spmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void geru( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + geru_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::geru(queue, m, n, alpha, x, incx, y, incy, a, lda); + geru_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void geru( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + geru_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::geru(queue, m, n, alpha, x, incx, y, incy, a, lda); + geru_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklcpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklcpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklcpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklcpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void gemm(cl::sycl::queue &queue, transpose transa, + transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +template <> +void gemm(cl::sycl::queue &queue, transpose transa, + transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void gemm( + cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void gemm( + cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); +template <> +void gemm(cl::sycl::queue &queue, transpose transa, + transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, + half beta, cl::sycl::buffer &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void herk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + herk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklcpu::herk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + herk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void herk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + herk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklcpu::herk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + herk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void ger(cl::sycl::queue &queue, std::int64_t m, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + ger_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::ger(queue, m, n, alpha, x, incx, y, incy, a, lda); + ger_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void ger(cl::sycl::queue &queue, std::int64_t m, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + ger_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklcpu::ger(queue, m, n, alpha, x, incx, y, incy, a, lda); + ger_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trsm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trsm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trsm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trsm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotu_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dotu(queue, n, x, incx, y, incy, result); + dotu_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotu_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dotu(queue, n, x, incx, y, incy, result); + dotu_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void hemm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + hemm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::hemm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + hemm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void hemm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + hemm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::hemm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + hemm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a); +template <> +void hpr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &a) { + hpr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklcpu::hpr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + hpr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a); +template <> +void hpr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &a) { + hpr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklcpu::hpr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + hpr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gbmv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gbmv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void gbmv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void gbmv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void tbmv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tbmv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void symm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void symm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void symm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void symm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotc_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dotc(queue, n, x, incx, y, incy, result); + dotc_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotc_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dotc(queue, n, x, incx, y, incy, result); + dotc_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { + syr_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklcpu::syr(queue, upper_lower, n, alpha, x, incx, a, lda); + syr_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { + syr_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklcpu::syr(queue, upper_lower, n, alpha, x, incx, a, lda); + syr_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trmm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trmm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trmm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trmm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklcpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m); +template <> +void rotmg(cl::sycl::queue &queue, + cl::sycl::buffer &d1, + cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m) { + rotmg_precondition(queue, d1, d2, x1, y1, param); + onemkl::mklcpu::rotmg(queue, d1, d2, x1, y1, param); + rotmg_postcondition(queue, d1, d2, x1, y1, param); +} + +template +static inline void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + double y1, cl::sycl::buffer ¶m); +template <> +void rotmg(cl::sycl::queue &queue, + cl::sycl::buffer &d1, + cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m) { + rotmg_precondition(queue, d1, d2, x1, y1, param); + onemkl::mklcpu::rotmg(queue, d1, d2, x1, y1, param); + rotmg_postcondition(queue, d1, d2, x1, y1, param); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklcpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklcpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklcpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hemv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hemv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::hemv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + hemv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hemv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hemv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::hemv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + hemv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void sbmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + sbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::sbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + sbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void sbmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + sbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::sbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + sbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklcpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklcpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklcpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklcpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void tbsv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tbsv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklcpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a); +template <> +void spr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { + spr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklcpu::spr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + spr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a); +template <> +void spr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { + spr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklcpu::spr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + spr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklcpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); +template <> +void rotm(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + rotm_precondition(queue, n, x, incx, y, incy, param); + onemkl::mklcpu::rotm(queue, n, x, incx, y, incy, param); + rotm_postcondition(queue, n, x, incx, y, incy, param); +} + +template +static inline void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); +template <> +void rotm(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + rotm_precondition(queue, n, x, incx, y, incy, param); + onemkl::mklcpu::rotm(queue, n, x, incx, y, incy, param); + rotm_postcondition(queue, n, x, incx, y, incy, param); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer &a, + cl::sycl::buffer &b, + cl::sycl::buffer &c, + cl::sycl::buffer &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklcpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer &a, + cl::sycl::buffer &b, + cl::sycl::buffer &c, + cl::sycl::buffer &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklcpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklcpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklcpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + sdsdot_precondition(queue, n, sb, x, incx, y, incy, result); + onemkl::mklcpu::sdsdot(queue, n, sb, x, incx, y, incy, result); + sdsdot_postcondition(queue, n, sb, x, incx, y, incy, result); +} + +template +static inline void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +template <> +void her2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + her2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::her2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + her2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void her2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + her2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklcpu::her2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + her2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void dot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dot(queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void dot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dot(queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void dot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklcpu::dot(queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void symv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + symv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::symv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + symv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void symv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + symv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklcpu::symv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + symv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +} //namespace blas +} //namespace onemkl + +#endif //_DETAIL_MKLCPU_BLAS_HPP_ diff --git a/include/onemkl/blas/detail/mklcpu/onemkl_blas_mklcpu.hpp b/include/onemkl/blas/detail/mklcpu/onemkl_blas_mklcpu.hpp new file mode 100644 index 000000000..0383955ae --- /dev/null +++ b/include/onemkl/blas/detail/mklcpu/onemkl_blas_mklcpu.hpp @@ -0,0 +1,854 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BLAS_MKLCPU_HPP_ +#define _ONEMKL_BLAS_MKLCPU_HPP_ + +#include + +#include +#include + +#include "onemkl/types.hpp" + +namespace onemkl { + +using onemkl::transpose; +using onemkl::uplo; +using onemkl::side; +using onemkl::diag; +using onemkl::offset; + +namespace mklcpu { + +// Level 1 + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + +void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); + +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + double c, double s); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, float c, float s); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, double c, double s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m); + +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m); + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m); + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, cl::sycl::buffer ¶m); + +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +// Level 2 + +void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); + +void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); + +void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + +void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + +void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); + +void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); + +void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); + +void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); + +void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy); + +void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy); + +void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); + +void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); + +void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); + +void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); + +void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); + +void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); + +void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); + +void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); + +void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx); + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx); + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx); + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx); + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +// Level 3 + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + float beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +// Batch API + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void gemm_batch( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, float beta, cl::sycl::buffer &c, + std::int64_t ldc, std::int64_t stride_c, std::int64_t batch_size); + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, double beta, + cl::sycl::buffer &c, std::int64_t ldc, std::int64_t stride_c, + std::int64_t batch_size); + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::int64_t batch_size); + +// BLAS-like extensions + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, offset offsetc, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, int8_t ao, + cl::sycl::buffer &b, std::int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, std::int64_t ldc, cl::sycl::buffer &co); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); + +} //namespace mklcpu +} //namespace onemkl + +#endif //_ONEMKL_BLAS_MKLCPU_HPP_ diff --git a/include/onemkl/blas/detail/mklgpu/blas_ct.hpp b/include/onemkl/blas/detail/mklgpu/blas_ct.hpp new file mode 100644 index 000000000..0e16b3b1b --- /dev/null +++ b/include/onemkl/blas/detail/mklgpu/blas_ct.hpp @@ -0,0 +1,2405 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +// +// Generated based on onemkl/blas/blas.hpp +// + +#ifndef _DETAIL_MKLGPU_BLAS_HPP_ +#define _DETAIL_MKLGPU_BLAS_HPP_ + +#include +#include + +#include "onemkl/detail/backends.hpp" +#include "onemkl/detail/libraries.hpp" +#include "onemkl/types.hpp" + +#include "onemkl_blas_mklgpu.hpp" + +namespace onemkl { +namespace blas { + +template +static inline void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + syr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::syr2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + syr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void syr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + syr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::syr2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + syr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklgpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + double alpha, cl::sycl::buffer &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklgpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklgpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklgpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklgpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void scal(cl::sycl::queue &queue, std::int64_t n, + double alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + scal_precondition(queue, n, alpha, x, incx); + onemkl::mklgpu::scal(queue, n, alpha, x, incx); + scal_postcondition(queue, n, alpha, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trmv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpmv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpmv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a); +template <> +void spr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { + spr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklgpu::spr(queue, upper_lower, n, alpha, x, incx, a); + spr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void spr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a); +template <> +void spr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { + spr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklgpu::spr(queue, upper_lower, n, alpha, x, incx, a); + spr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void hpmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + hpmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklgpu::hpmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + hpmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void hpmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void hpmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + hpmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklgpu::hpmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + hpmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, float beta, cl::sycl::buffer &c, + std::int64_t ldc); +template <> +void syrk(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, + std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklgpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, double beta, cl::sycl::buffer &c, + std::int64_t ldc); +template <> +void syrk(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, + std::int64_t lda, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklgpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syrk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklgpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syrk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syrk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklgpu::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + syrk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void her2( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + her2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::her2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + her2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void her2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void her2( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + her2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::her2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); + her2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hbmv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::hbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + hbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void hbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hbmv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::hbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + hbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, float c, float s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklgpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, double c, + double s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, double c, double s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklgpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, float c, + float s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + float c, float s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklgpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + double c, double s); +template <> +void rot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + double c, double s) { + rot_precondition(queue, n, x, incx, y, incy, c, s); + onemkl::mklgpu::rot(queue, n, x, incx, y, incy, c, s); + rot_postcondition(queue, n, x, incx, y, incy, c, s); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklgpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, + double alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklgpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklgpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void axpy(cl::sycl::queue &queue, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + axpy_precondition(queue, n, alpha, x, incx, y, incy); + onemkl::mklgpu::axpy(queue, n, alpha, x, incx, y, incy); + axpy_postcondition(queue, n, alpha, x, incx, y, incy); +} + +template +static inline void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void gerc( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + gerc_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::gerc(queue, m, n, alpha, x, incx, y, incy, a, lda); + gerc_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void gerc( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + gerc_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::gerc(queue, m, n, alpha, x, incx, y, incy, a, lda); + gerc_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void syr2k(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void syr2k(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syr2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void syr2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + syr2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + syr2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gemv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gemv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void gemv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gemv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void gemv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + gemv_precondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); + gemv_postcondition(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +template <> +void her(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a, + std::int64_t lda) { + her_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklgpu::her(queue, upper_lower, n, alpha, x, incx, a, lda); + her_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void her(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +template <> +void her(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a, + std::int64_t lda) { + her_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklgpu::her(queue, upper_lower, n, alpha, x, incx, a, lda); + her_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +template <> +void hpr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a) { + hpr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklgpu::hpr(queue, upper_lower, n, alpha, x, incx, a); + hpr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void hpr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +template <> +void hpr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &a) { + hpr_precondition(queue, upper_lower, n, alpha, x, incx, a); + onemkl::mklgpu::hpr(queue, upper_lower, n, alpha, x, incx, a); + hpr_postcondition(queue, upper_lower, n, alpha, x, incx, a); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamin(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamin_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamin(queue, n, x, incx, result); + iamin_postcondition(queue, n, x, incx, result); +} + +template +static inline void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, + std::int64_t incy); +template <> +void spmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + spmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklgpu::spmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + spmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void spmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, + std::int64_t incy); +template <> +void spmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + spmv_precondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + onemkl::mklgpu::spmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); + spmv_postcondition(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void swap(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + swap_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::swap(queue, n, x, incx, y, incy); + swap_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void geru( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + geru_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::geru(queue, m, n, alpha, x, incx, y, incy, a, lda); + geru_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); +template <> +void geru( + cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + geru_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::geru(queue, m, n, alpha, x, incx, y, incy, a, lda); + geru_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklgpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklgpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklgpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void nrm2(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + nrm2_precondition(queue, n, x, incx, result); + onemkl::mklgpu::nrm2(queue, n, x, incx, result); + nrm2_postcondition(queue, n, x, incx, result); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void gemm(cl::sycl::queue &queue, transpose transa, + transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +template <> +void gemm(cl::sycl::queue &queue, transpose transa, + transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void gemm( + cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void gemm( + cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); +template <> +void gemm(cl::sycl::queue &queue, transpose transa, + transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, + half beta, cl::sycl::buffer &c, + std::int64_t ldc) { + gemm_precondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + gemm_postcondition(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void herk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + herk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklgpu::herk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + herk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void herk( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + herk_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + onemkl::mklgpu::herk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); + herk_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +template +static inline void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void ger(cl::sycl::queue &queue, std::int64_t m, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + ger_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::ger(queue, m, n, alpha, x, incx, y, incy, a, lda); + ger_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void ger(cl::sycl::queue &queue, std::int64_t m, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { + ger_precondition(queue, m, n, alpha, x, incx, y, incy, a, lda); + onemkl::mklgpu::ger(queue, m, n, alpha, x, incx, y, incy, a, lda); + ger_postcondition(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trsm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trsm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trsm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trsm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trsm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trsm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotu_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dotu(queue, n, x, incx, y, incy, result); + dotu_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotu(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotu_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dotu(queue, n, x, incx, y, incy, result); + dotu_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void hemm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + hemm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::hemm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + hemm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void hemm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + hemm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::hemm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + hemm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a); +template <> +void hpr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &a) { + hpr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklgpu::hpr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + hpr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void hpr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a); +template <> +void hpr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &a) { + hpr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklgpu::hpr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + hpr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gbmv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void gbmv(cl::sycl::queue &queue, transpose trans, + std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void gbmv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void gbmv(cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +template <> +void gbmv( + cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, std::int64_t kl, + std::int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, std::int64_t incy) { + gbmv_precondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); + gbmv_postcondition(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbmv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void tbmv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tbmv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbmv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbmv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void symm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc); +template <> +void symm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void symm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void symm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + symm_precondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); + symm_postcondition(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotc_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dotc(queue, n, x, incx, y, incy, result); + dotc_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +template <> +void dotc(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy, + cl::sycl::buffer, 1> &result) { + dotc_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dotc(queue, n, x, incx, y, incy, result); + dotc_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { + syr_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklgpu::syr(queue, upper_lower, n, alpha, x, incx, a, lda); + syr_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void syr(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda); +template <> +void syr(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { + syr_precondition(queue, upper_lower, n, alpha, x, incx, a, lda); + onemkl::mklgpu::syr(queue, upper_lower, n, alpha, x, incx, a, lda); + syr_postcondition(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trmm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +template <> +void trmm(cl::sycl::queue &queue, side left_right, + uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trmm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +template <> +void trmm( + cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + trmm_precondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + onemkl::mklgpu::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); + trmm_postcondition(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, lda, b, + ldb); +} + +template +static inline void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m); +template <> +void rotmg(cl::sycl::queue &queue, + cl::sycl::buffer &d1, + cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m) { + rotmg_precondition(queue, d1, d2, x1, y1, param); + onemkl::mklgpu::rotmg(queue, d1, d2, x1, y1, param); + rotmg_postcondition(queue, d1, d2, x1, y1, param); +} + +template +static inline void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + double y1, cl::sycl::buffer ¶m); +template <> +void rotmg(cl::sycl::queue &queue, + cl::sycl::buffer &d1, + cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m) { + rotmg_precondition(queue, d1, d2, x1, y1, param); + onemkl::mklgpu::rotmg(queue, d1, d2, x1, y1, param); + rotmg_postcondition(queue, d1, d2, x1, y1, param); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tpsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + tpsv_precondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); + onemkl::mklgpu::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); + tpsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void trsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + std::int64_t lda, + cl::sycl::buffer, 1> &x, + std::int64_t incx) { + trsv_precondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + onemkl::mklgpu::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); + trsv_postcondition(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void copy(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer, 1> &y, + std::int64_t incy) { + copy_precondition(queue, n, x, incx, y, incy); + onemkl::mklgpu::copy(queue, n, x, incx, y, incy); + copy_postcondition(queue, n, x, incx, y, incy); +} + +template +static inline void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hemv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hemv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::hemv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + hemv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void hemv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +template <> +void hemv( + cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + hemv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::hemv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + hemv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void sbmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + sbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::sbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + sbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void sbmv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void sbmv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + sbmv_precondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::sbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); + sbmv_postcondition(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklgpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklgpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklgpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void asum(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + asum_precondition(queue, n, x, incx, result); + onemkl::mklgpu::asum(queue, n, x, incx, result); + asum_postcondition(queue, n, x, incx, result); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); +template <> +void tbsv(cl::sycl::queue &queue, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx); +template <> +void tbsv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +template <> +void tbsv( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, std::int64_t n, + std::int64_t k, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + tbsv_precondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + onemkl::mklgpu::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); + tbsv_postcondition(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +template +static inline void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a); +template <> +void spr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { + spr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklgpu::spr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + spr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void spr2(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a); +template <> +void spr2(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { + spr2_precondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); + onemkl::mklgpu::spr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); + spr2_postcondition(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); +template <> +void iamax(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, + std::int64_t incx, + cl::sycl::buffer &result) { + iamax_precondition(queue, n, x, incx, result); + onemkl::mklgpu::iamax(queue, n, x, incx, result); + iamax_postcondition(queue, n, x, incx, result); +} + +template +static inline void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); +template <> +void rotm(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + rotm_precondition(queue, n, x, incx, y, incy, param); + onemkl::mklgpu::rotm(queue, n, x, incx, y, incy, param); + rotm_postcondition(queue, n, x, incx, y, incy, param); +} + +template +static inline void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); +template <> +void rotm(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + rotm_precondition(queue, n, x, incx, y, incy, param); + onemkl::mklgpu::rotm(queue, n, x, incx, y, incy, param); + rotm_postcondition(queue, n, x, incx, y, incy, param); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer &a, + cl::sycl::buffer &b, + cl::sycl::buffer &c, + cl::sycl::buffer &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklgpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer &a, + cl::sycl::buffer &b, + cl::sycl::buffer &c, + cl::sycl::buffer &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklgpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklgpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +template <> +void rotg(cl::sycl::queue &queue, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + rotg_precondition(queue, a, b, c, s); + onemkl::mklgpu::rotg(queue, a, b, c, s); + rotg_postcondition(queue, a, b, c, s); +} + +template +static inline void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + sdsdot_precondition(queue, n, sb, x, incx, y, incy, result); + onemkl::mklgpu::sdsdot(queue, n, sb, x, incx, y, incy, result); + sdsdot_postcondition(queue, n, sb, x, incx, y, incy, result); +} + +template +static inline void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +template <> +void her2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + her2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::her2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + her2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +template <> +void her2k( + cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + her2k_precondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + onemkl::mklgpu::her2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); + her2k_postcondition(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +template +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void dot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dot(queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void dot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dot(queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +template <> +void dot(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + dot_precondition(queue, n, x, incx, y, incy, result); + onemkl::mklgpu::dot(queue, n, x, incx, y, incy, result); + dot_postcondition(queue, n, x, incx, y, incy, result); +} + +template +static inline void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void symv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, + std::int64_t incy) { + symv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::symv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + symv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +template +static inline void symv(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +template <> +void symv(cl::sycl::queue &queue, uplo upper_lower, + std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, + std::int64_t incy) { + symv_precondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + onemkl::mklgpu::symv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); + symv_postcondition(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +} //namespace blas +} //namespace onemkl + +#endif //_DETAIL_MKLGPU_BLAS_HPP_ diff --git a/include/onemkl/blas/detail/mklgpu/onemkl_blas_mklgpu.hpp b/include/onemkl/blas/detail/mklgpu/onemkl_blas_mklgpu.hpp new file mode 100644 index 000000000..f42be1db2 --- /dev/null +++ b/include/onemkl/blas/detail/mklgpu/onemkl_blas_mklgpu.hpp @@ -0,0 +1,874 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BLAS_MKLGPU_HPP_ +#define _ONEMKL_BLAS_MKLGPU_HPP_ + +#include +#include +#include + +#include "onemkl/types.hpp" + +namespace onemkl { +namespace mklgpu { + +// Level 3 + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + double beta, cl::sycl::buffer &c, std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb); + +// Level 2 + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void hbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void hbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void hemv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void hemv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void her(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void her(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void her2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void her2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); + +void hpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void hpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + +void hpr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + +void hpr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + +void hpr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); + +void hpr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); + +void sbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + +void sbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + +void symv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); + +void symv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); + +void syr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); + +void syr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); + +void syr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void syr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); + +void spmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy); + +void spmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy); + +void spr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); + +void spr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); + +void spr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); + +void spr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); + +// Level 1 + +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + +void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); + +void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + double c, double s); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, float c, float s); + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, double c, double s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m); + +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m); + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m); + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, cl::sycl::buffer ¶m); + +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); + +// Batch API + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &k, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + cl::sycl::buffer &beta, cl::sycl::buffer &c, + cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size); + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &k, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + cl::sycl::buffer &beta, cl::sycl::buffer &c, + cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size); + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void gemm_batch( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, std::int64_t stride_c, + std::int64_t batch_size); + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + double beta, cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); + +// BLAS like + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + onemkl::offset offsetc, std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, int8_t ao, + cl::sycl::buffer &b, std::int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, std::int64_t ldc, cl::sycl::buffer &co); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, half beta, cl::sycl::buffer &c, std::int64_t ldc); + +} //namespace mklgpu +} //namespace onemkl + +#endif //_ONEMKL_BLAS_MKLGPU_HPP_ diff --git a/include/onemkl/blas/predicates.hpp b/include/onemkl/blas/predicates.hpp new file mode 100644 index 000000000..d485bd1c6 --- /dev/null +++ b/include/onemkl/blas/predicates.hpp @@ -0,0 +1,3412 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BLAS_PREDICATES_HPP_ +#define _ONEMKL_BLAS_PREDICATES_HPP_ + +#include +#include +#include + +#include "onemkl/detail/exceptions.hpp" +#include "onemkl/types.hpp" + +namespace onemkl { +namespace blas { + +inline void herk_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void herk_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void herk_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void herk_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void scal_precondition(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void scal_postcondition(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void scal_precondition(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void scal_postcondition(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void scal_precondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void scal_postcondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void scal_precondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void scal_postcondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void scal_precondition(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void scal_postcondition(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void scal_precondition(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void scal_postcondition(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void spr_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void spr_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void spr_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void spr_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, double beta, + cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, double beta, + cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_batch_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syrk_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syrk_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syrk_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syrk_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syrk_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syrk_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syrk_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syrk_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void her2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void her2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void her2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void her2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rot_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + float c, float s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rot_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + float c, float s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rot_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + double c, double s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rot_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + double c, double s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rot_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + float c, float s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rot_postcondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + float c, float s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rot_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + double c, double s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rot_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, double c, + double s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void axpy_precondition(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void axpy_postcondition(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void axpy_precondition(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void axpy_postcondition(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void axpy_precondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void axpy_postcondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void axpy_precondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void axpy_postcondition(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gerc_precondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gerc_postcondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gerc_precondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gerc_postcondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr2k_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr2k_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr2k_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr2k_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr2k_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr2k_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr2k_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr2k_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void her_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void her_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void her_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void her_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &a, + std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hpr_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hpr_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hpr_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hpr_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + offset offsetc, std::int64_t m, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + int8_t ao, cl::sycl::buffer &b, std::int64_t ldb, + uint8_t bo, float beta, cl::sycl::buffer &c, + std::int64_t ldc, cl::sycl::buffer &co) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + offset offsetc, std::int64_t m, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + int8_t ao, cl::sycl::buffer &b, std::int64_t ldb, + uint8_t bo, float beta, cl::sycl::buffer &c, + std::int64_t ldc, cl::sycl::buffer &co) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_ext_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamin_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamin_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamin_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamin_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamin_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamin_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamin_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamin_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hpmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hpmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hpmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hpmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void spmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, + std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void spmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void spmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void spmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotmg_precondition(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + float y1, cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotmg_postcondition(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + float y1, cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotmg_precondition(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + double y1, cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotmg_postcondition(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, + double y1, cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void swap_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void swap_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void swap_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void swap_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void swap_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void swap_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void swap_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void swap_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void geru_precondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void geru_postcondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void geru_precondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void geru_postcondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void nrm2_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void nrm2_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void nrm2_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void nrm2_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void nrm2_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void nrm2_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void nrm2_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void nrm2_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemmt_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemmt_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemmt_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemmt_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemmt_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemmt_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemmt_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemmt_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gemm_precondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gemm_postcondition(cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void ger_precondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void ger_postcondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void ger_precondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void ger_postcondition(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dotu_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dotu_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dotu_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dotu_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hemm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hemm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hemm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hemm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hpr2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hpr2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hpr2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hpr2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gbmv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gbmv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gbmv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gbmv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gbmv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gbmv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void gbmv_precondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void gbmv_postcondition(cl::sycl::queue &queue, transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void symm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void symm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void symm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void symm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void symm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void symm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void symm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void symm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dotc_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dotc_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dotc_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dotc_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void syr_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void syr_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trmm_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trmm_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void symv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void symv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void symv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void symv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tpsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tpsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void copy_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void copy_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void copy_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void copy_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void copy_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void copy_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void copy_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void copy_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hemv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hemv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void hemv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void hemv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamax_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamax_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamax_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamax_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamax_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamax_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void iamax_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void iamax_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void sbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void sbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void sbmv_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void sbmv_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void asum_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void asum_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void asum_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void asum_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void asum_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void asum_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void asum_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void asum_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void tbsv_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void tbsv_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void spr2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void spr2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void spr2_precondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void spr2_postcondition(cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_precondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void trsm_batch_postcondition(cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotm_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotm_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotm_precondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotm_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dot_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dot_postcondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dot_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dot_postcondition(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void dot_precondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void dot_postcondition(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void sdsdot_precondition(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void sdsdot_postcondition(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void her2k_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void her2k_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void her2k_precondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void her2k_postcondition(cl::sycl::queue &queue, uplo upper_lower, transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotg_precondition(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotg_postcondition(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotg_precondition(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotg_postcondition(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotg_precondition(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotg_postcondition(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +inline void rotg_precondition(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add prechecks to queue here for input args. */ +#endif +} + +inline void rotg_postcondition(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { +#ifndef ONEMKL_DISABLE_PREDICATES + /* add postchecks to queue here for input args. */ +#endif +} + +} //namespace blas +} //namespace onemkl + +#endif //_ONEMKL_BLAS_PREDICATES_HPP_ diff --git a/include/onemkl/blas/tools/blas_func_gen.py b/include/onemkl/blas/tools/blas_func_gen.py new file mode 100755 index 000000000..b7754774c --- /dev/null +++ b/include/onemkl/blas/tools/blas_func_gen.py @@ -0,0 +1,104 @@ +#!/usr/bin/env python +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +from sys import argv, exit, stdin +from pprint import pprint +from collections import defaultdict +import re + +from func_parser import create_db + +def usage(err = None): + if err: + print 'error: %s' % err + print '''\ +usage: {script} + +usage example: +{script} blas.hpp new_blas.hpp intelmkl intelcpu mkl_sycl_blas.hpp onemkl::intelmkl::blas +'''.format(script = argv[0]) + +if len(argv) <= 1: + usage() + exit(0) + +if re.search(r'[-]*\b[h]([e][l][p])?\b' ,argv[1]): + usage() + exit(0) + +in_filename = argv[1] +out_filename = argv[2] +lib = argv[3] +backend = argv[4] +include = argv[5] +namespace = argv[6] + +header_db = create_db(in_filename) + +def print_funcs(func_list): + code="" + for data in func_list: + code +=""" +template static inline {ret_type} {name}{par_str}; +template<> +{ret_type} {name}{par_str} {{ + {name}_precondition{call_str}; + {namespace}::{name}{call_str}; + {name}_postcondition{call_str}; +}} +""".format(lib=lib, namespace=namespace, backend=backend, **data) + return code + +print "Generate " + out_filename + "..." +out_file = open(out_filename, "w+") +out_file.write("""// +// Generated based on onemkl/blas/blas.hpp +// + +#ifndef _{gard}_HPP_ +#define _{gard}_HPP_ + +#include +#include + +#include "onemkl/types.hpp" +#include "onemkl/detail/backends.hpp" +#include "onemkl/detail/libraries.hpp" + +#include "{include}" + +""".format(gard=(out_filename.split('.', 1)[0].upper()).replace('/','_'), include=include)) + + +out_file.write(""" + +namespace onemkl { +namespace blas {""") + +for func_name, func_list in header_db.iteritems(): + out_file.write(""" +{funcs}""".format(funcs=print_funcs(func_list))) + +out_file.write(""" +}} //namespace blas +}} //namespace onemkl + +#endif //_{gard}_HPP_""".format(gard=(out_filename.split('.', 1)[0].upper()).replace('/','_'))) +out_file.close() diff --git a/include/onemkl/blas/tools/func_parser.py b/include/onemkl/blas/tools/func_parser.py new file mode 100755 index 000000000..e51e91302 --- /dev/null +++ b/include/onemkl/blas/tools/func_parser.py @@ -0,0 +1,174 @@ +#!/usr/bin/env python +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +from sys import argv, exit, stdin +from collections import defaultdict +import re + +def parse_item(item): + func_type_and_name, par_str = item.split('(', 1) + ret_type, func_name = func_type_and_name.strip().rsplit(' ', 1) + """remove macros callback""" + ret_type = re.sub('^[A-Z_]+ ','',ret_type) + """remove macros calling convention""" + ret_type = re.sub(' [A-Z_]+$','',ret_type.strip()) + if func_name[0] == '*': + ret_type += ' *' + func_name = func_name[1:] + + par_str = re.sub('^\s*(void)\s*$', '', par_str.strip(');')) + + """Extract callback calls from parameter list and replace them by temp 'cllbck' param""" + clbck_type = None + clbck_param = None + if par_str.find('(') != -1: + par_str = re.sub('[)]\s*[(]', ')(', par_str) + clbck_re = re.compile(r'\w*\s*[(]\s*[A-Z_]*\s*[*]\s*\w*\s*[)]\s*[(][\w*,\[\]\s]*[)]') + clbck_list = clbck_re.findall(par_str) + clbck_type = [ re.sub('[*]\s*\w*', '*', x.split(')(')[0]) for x in clbck_list] + clbck_param = [ x.split(')(')[1] for x in clbck_list] + + par_str = re.sub('[)]\s*[(][\w*,\[\]\s]*[)]', '', par_str) + par_str = re.sub('\w*\s*[(]\s*[A-Z_]*\s*[\w]*\s*[*]\s*', 'cllbck ', par_str) + par_str_ = re.sub('[,]+(?![^<]+>)', '@', par_str) + par_list = [x.strip() for x in par_str_.split('@') \ + if len(x.strip()) > 0 ] + """Split list of parameters to types and names""" + if len(par_list) > 0: + """Add parameter names (param1, param2, etc) if the declaration includes only types""" + if re.search('(,|^)+\s*(const)*\s*[\w:]+\s*[*]*\s*(,|$)+', re.sub('<[\s\w\d,]*>', '', par_str)) is not None: + par_list = [(x + ' param' + str(idx)).replace(" * ", " *" \ + ).replace("[] param" + str(idx), "param" + str(idx) + "[]") \ + for idx, x in enumerate(par_list)] + + """Extract names to call_list""" + call_list = [x.rsplit(' ', 1)[1].strip(' *').strip('\[\]').strip('&') \ + for x in par_list] + + """Extract types to sig_list""" + par_list_wo_st_arrays = [(x.rsplit(' ', 1)[0] + \ + (lambda x: '* ' if x.find('[]') != -1 else ' ')(x.rsplit(' ', 1)[1]) + \ + (x.rsplit(' ', 1)[1]).strip('\[\]')) for x in par_list] + sig_list = [(x.rsplit(' ', 1)[0] + \ + (x.rsplit(' ', 1)[1].startswith('*') \ + and (' ' + x.rsplit(' ', 1)[1].count('*') * '*') or '')) \ + for x in par_list_wo_st_arrays] + else: + call_list = list() + sig_list = list() + par_str = '(' + ', '.join(par_list) + ')' + call_str = '(' + ', '.join(call_list) + ')' + sig_str = '(' + ', '.join(sig_list) + ')' + + """Put real callback call types back to the param_list and sig_str """ + if clbck_param is not None: + for idx, x in enumerate(clbck_param): + par_str = re.sub(r'(cllbck\s*\w*)[,]', r'\1(' + x + ',', par_str, idx) + sig_str = re.sub(r'(cllbck\s*\w*)[,]', r'\1(' + x + ',', sig_str, idx) + + if clbck_type is not None: + for idx, x in enumerate(clbck_type): + par_str = re.sub(r'cllbck(\s*\w*)', x + r'\1)', par_str, idx) + sig_str = re.sub(r'cllbck(\s*\w*)', x + r'\1)', sig_str, idx) + return func_name, ret_type, func_name, par_str, call_str, sig_str, call_list, sig_list + + +def to_dict(func_data): + """ convert (ret_type, 'name', par_str, call_str, sig_str, call_list, sig_list) tuple to + dict with corresponding keys """ + return dict(zip(('ret_type', 'name', 'par_str', 'call_str', 'sig_str', 'call_list', 'sig_list'), func_data)) + +is_comment = 0 +is_wrapperbody = 0 +def strip_line(l): + """ remove global variables""" + if re.search('^\s*\w+\s*\w+[;]', l) is not None: + l = '' + """ remove namespaces""" + if re.search('^\s*namespace\s*\w+\s*[{]', l) is not None: + l = '' + """ remove declaration keywords """ + l = re.sub("^extern ", "", l) + l = re.sub("^static ", "", l) + l = re.sub("^inline ", "", l) + """ remove extra whitespace and comments from input line """ + l = re.sub("[)][A-Za-z0-9\s_]*[;]", ");", l) + + """ remove simple wrapper function body""" + global is_wrapperbody + if is_wrapperbody == 1: + if re.search('^\s*}', l) is not None: + l = l.split('}', 1)[1].strip() + is_wrapperbody = 0 + else: + return "" + + if re.search(r'[)]\s*\n*\s*[{]', l): + l = l.split('{', 1)[0].strip() + ";" + is_wrapperbody = 1 + + global is_comment + if is_comment == 1: + if re.search('\*/', l) is not None: + l = l.split('*/', 1)[1].strip() + is_comment = 0 + else: + return "" + """ Delete comments """ + l1 = l.split('#', 1)[0].strip() + l2 = l1.split('//', 1)[0].strip() + l3 = l2.split('/*', 1)[0].strip() + if re.search('/\*', l2) is not None: + is_comment = 1 + if re.search('\*/', l2) is not None: + is_comment = 0 + l4 = l2.split('*/', 1)[1].strip() + l3 += l4 + """ Delete comments if there are several of them in one line """ + l3 = re.sub("[/][*][\w\s]*[*][/]", "", l3); + """Delete all tabs""" + return re.sub(' +',' ', l3) + +def create_db(filename): + with open(filename, 'r') as f: + data = f.readlines() + funcs_db = defaultdict(list) + whole_line = "" + idx = 0 + for l in data: + stripped = strip_line(l) + if not stripped: + continue + """ Check if function contains 1 line """ + whole_line += stripped + ' ' + """ Check if there is function """ + if re.search('[(][\w\s\*/\&,_\[\]():<>]*[)]\s*[;]', whole_line) is None: + """ Check if there is some other staff before the function """ + if re.search('[;{}]\s*$', whole_line) is not None: + whole_line = "" + continue + else: + stripped = whole_line.strip() + whole_line = "" + parsed = parse_item(stripped) + func_name, func_data = parsed[0], parsed[1:] + funcs_db[func_name].append(to_dict(func_data)) + idx = idx + 1 + return funcs_db diff --git a/include/onemkl/blas/tools/func_parser.pyc b/include/onemkl/blas/tools/func_parser.pyc new file mode 100644 index 000000000..64059815c Binary files /dev/null and b/include/onemkl/blas/tools/func_parser.pyc differ diff --git a/include/onemkl/detail/backends.hpp b/include/onemkl/detail/backends.hpp new file mode 100644 index 000000000..19cd80470 --- /dev/null +++ b/include/onemkl/detail/backends.hpp @@ -0,0 +1,38 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BACKENDS_HPP_ +#define _ONEMKL_BACKENDS_HPP_ + +#include +#include + +namespace onemkl { + +enum class backend { intelcpu, intelgpu, unsupported }; + +typedef std::map backendmap; + +static backendmap backend_map = { { backend::intelcpu, "intelcpu" }, + { backend::intelgpu, "intelgpu" }, + { backend::unsupported, "unsupported" } }; + +} //namespace onemkl + +#endif //_ONEMKL_BACKENDS_HPP_ diff --git a/include/onemkl/detail/backends_selector.hpp b/include/onemkl/detail/backends_selector.hpp new file mode 100644 index 000000000..3cde4f886 --- /dev/null +++ b/include/onemkl/detail/backends_selector.hpp @@ -0,0 +1,57 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_BACKENDS_SELECTOR_HPP_ +#define _ONEMKL_BACKENDS_SELECTOR_HPP_ + +#include +#include +#include +#include "onemkl/detail/backends.hpp" + +#ifdef __linux__ + #define LIB_NAME(a) "lib" a ".so" +#endif + +#define INTEL_ID 32902 + +namespace onemkl { +inline char *select_backend(cl::sycl::queue &queue) { + if (queue.is_host()) { + return (char *)LIB_NAME("onemkl_blas_mklcpu"); + } + else if (queue.get_device().is_cpu()) { + return (char *)LIB_NAME("onemkl_blas_mklcpu"); + } + else if (queue.get_device().is_gpu()) { + unsigned int vendor_id = static_cast( + queue.get_device().get_info()); + + if (vendor_id == INTEL_ID) + return (char *)LIB_NAME("onemkl_blas_mklgpu"); + return (char *)"unsupported"; + } + else { + return (char *)"unsupported"; + } +} + +} //namespace onemkl + +#endif //_ONEMKL_BACKENDS_SELECTOR_HPP_ diff --git a/include/onemkl/detail/exceptions.hpp b/include/onemkl/detail/exceptions.hpp new file mode 100644 index 000000000..6b4623ce0 --- /dev/null +++ b/include/onemkl/detail/exceptions.hpp @@ -0,0 +1,89 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_EXCEPTIONS_HPP_ +#define _ONEMKL_EXCEPTIONS_HPP_ + +#include "onemkl/detail/backends.hpp" +#include "onemkl/types.hpp" + +namespace onemkl { + +class InvalidArgumentsException : virtual public std::exception { +private: + std::string error_message = ""; + +public: + InvalidArgumentsException(const std::string& msg) + : error_message(std::string("oneMKL InvalidArgumentsException: \n") + msg + "\n") {} + + const char* what() const noexcept override { + return error_message.c_str(); + }; +}; + +class MemoryAllocationException : virtual public std::exception { +private: + std::string error_message = ""; + +public: + MemoryAllocationException(const std::string& msg) + : error_message(std::string("oneMKL MemoryAllocationException: \n") + msg + "\n") {} + + const char* what() const noexcept override { + return error_message.c_str(); + }; +}; + +class UnsupportedBackendException : virtual public std::exception { +private: + std::string error_message = ""; + +public: + UnsupportedBackendException(cl::sycl::queue& queue, const std::string& msg) + : error_message( + std::string("oneMKL UnsupportedBackendException: \n") + + std::string(" There is currently no onemkl::backend available for the \n") + + std::string(" provided queue, device and sycl::backend. \n")) {} + + const char* what() const noexcept override { + return error_message.c_str(); + }; +}; + +class BackendNotAvailableForApiException : virtual public std::exception { +private: + std::string error_message = ""; + +public: + BackendNotAvailableForApiException(cl::sycl::queue& queue, onemkl::backend& backend, + const std::string& api_description) + : error_message(std::string("oneMKL BackendNotAvailableForApiException: \n") + + std::string(" The onemkl::backend = ") + onemkl::backend_map[backend] + + std::string("\n") + std::string(" is not available for ") + + api_description + std::string("\n")) {} + + const char* what() const noexcept override { + return error_message.c_str(); + }; +}; + +} //namespace onemkl + +#endif //_ONEMKL_EXCEPTIONS_HPP_ diff --git a/include/onemkl/detail/libraries.hpp b/include/onemkl/detail/libraries.hpp new file mode 100644 index 000000000..cbb55782f --- /dev/null +++ b/include/onemkl/detail/libraries.hpp @@ -0,0 +1,37 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_LIBRARIES_HPP_ +#define _ONEMKL_LIBRARIES_HPP_ + +#include +#include +#include + +namespace onemkl { + +enum class library { intelmkl, cublas }; + +typedef std::map librarymap; + +static librarymap library_map = { { library::intelmkl, "intelmkl" } }; + +} //namespace onemkl + +#endif //_ONEMKL_LIBRARIES_HPP_ diff --git a/include/onemkl/onemkl.hpp b/include/onemkl/onemkl.hpp new file mode 100644 index 000000000..e47b1d2f1 --- /dev/null +++ b/include/onemkl/onemkl.hpp @@ -0,0 +1,27 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_HPP_ +#define _ONEMKL_HPP_ + +#include + +#include + +#endif //_ONEMKL_HPP_ diff --git a/include/onemkl/types.hpp b/include/onemkl/types.hpp new file mode 100644 index 000000000..8ada1ce62 --- /dev/null +++ b/include/onemkl/types.hpp @@ -0,0 +1,61 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _ONEMKL_TYPES_HPP_ +#define _ONEMKL_TYPES_HPP_ + +namespace onemkl { + +// BLAS flag types. +enum class transpose : char { nontrans = 0, trans = 1, conjtrans = 3, N = 0, T = 1, C = 3 }; + +enum class uplo : char { upper = 0, lower = 1, U = 0, L = 1 }; + +enum class diag : char { nonunit = 0, unit = 1, N = 0, U = 1 }; + +enum class side : char { left = 0, right = 1, L = 0, R = 1 }; + +enum class offset : char { row = 0, column = 1, fix = 2, R = 0, C = 1, F = 2 }; + +// LAPACK flag types. +enum class job : char { + novec = 0, + vec = 1, + updatevec = 2, + allvec = 3, + somevec = 4, + overwritevec = 5, + N = 0, + V = 1, + U = 2, + A = 3, + S = 4, + O = 5 +}; + +enum class generate : char { q = 0, p = 1, none = 2, both = 3, Q = 0, P = 1, N = 2, V = 3 }; + +enum class index_base : char { + zero = 0, + one = 1, +}; + +} //namespace onemkl + +#endif //_ONEMKL_TYPES_HPP_ diff --git a/legal_information.md b/legal_information.md new file mode 100644 index 000000000..120c41043 --- /dev/null +++ b/legal_information.md @@ -0,0 +1,36 @@ +# Trademark Information + +3D XPoint, Altera, APEX, AnyWAN, Arria, Avalon, Axxia, BunnyPeople, Celeron, +Cilk, CONVERGATE, Cyclone, Docea, eASIC, easicopy, Enpirion, Hyperflex, Intel, +the Intel logo, Intel Adaptix, Intel Agilex, Intel Atom, Intel CoFluent, Intel +Core, Intel Inside, the Intel Inside logo, Intel Nervana, Intel Optane, Intel +RealSense, Intel Shooting Star, Intel Sirius, Intel SpeedStep, Intel Unite, +Intel vPro, Intel Xeon Phi, Iris, Itanium, MAX, Movidius, Myriad, neon, Nios, +OpenVINO, the OpenVINO logo, Pentium, Puma, Quark, Quartus, SICOFI, Simics, +SoftSilicon, Sound Mark, StarPro, Stratix, the Stratix logo, Stay With It, the +Engineering Stay With It logo, StreamSight, Tarari, The Journey Inside, +Thunderbolt, the Thunderbolt logo, Transcede, Ultrabook, VTune, and Xeon are +trademarks of Intel Corporation or its subsidiaries. + +*Other names and brands may be claimed as the property of others. + +Microsoft, Windows, and the Windows logo are trademarks, or registered +trademarks of Microsoft Corporation in the United States and/or other countries. + +Java is a registered trademark of Oracle and/or its affiliates. + +The Bluetooth(R) word mark and logos are registered trademarks owned by the +Bluetooth SIG, Inc. and any use of such marks by Intel Corporation is under +license. + +Intel Corporation uses the Palm OS* Ready mark under license from Palm, Inc. + +OpenCL and the OpenCL logo are trademarks of Apple Inc. used by permission by +Khronos. + +Radeon and the Radeon RX Vega logo are trademarks of Advanced Micro Devices, +Inc. + +SPEC(R), SPECrate(R) and SPEC CPU(R) are registered trademarks of the Standard +Performance Evaluation Corporation. See http://www.spec.org/spec/trademarks.html +for more information. \ No newline at end of file diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt new file mode 100644 index 000000000..2571f6339 --- /dev/null +++ b/src/CMakeLists.txt @@ -0,0 +1,59 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# build blas_loader and backends +add_subdirectory(blas) + +# generate header with enabled backends for testing +configure_file(config.hpp.in "${CMAKE_CURRENT_BINARY_DIR}/onemkl/config.hpp.configured") +file(GENERATE + OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/onemkl/config.hpp" + INPUT "${CMAKE_CURRENT_BINARY_DIR}/onemkl/config.hpp.configured" +) + +# Add recipe for onemkl loader library +if(BUILD_SHARED_LIBS) +add_library(onemkl SHARED) + +target_include_directories(onemkl + PUBLIC $ + $ + $ +) +set_target_properties(onemkl PROPERTIES + SOVERSION ${PROJECT_VERSION_MAJOR} +) + +# generate header file for exporting symbols +include(GenerateExportHeader) +generate_export_header( + onemkl + EXPORT_FILE_NAME "onemkl/export.hpp" +) +# Build dispatcher library +target_link_libraries(onemkl PUBLIC onemkl_blas) + +# Add the library to install package +install(TARGETS onemkl_blas EXPORT oneMKLTargets) +install(TARGETS onemkl EXPORT oneMKLTargets + RUNTIME DESTINATION bin + ARCHIVE DESTINATION lib + LIBRARY DESTINATION lib +) +endif() diff --git a/src/blas/CMakeLists.txt b/src/blas/CMakeLists.txt new file mode 100644 index 000000000..46bd63657 --- /dev/null +++ b/src/blas/CMakeLists.txt @@ -0,0 +1,38 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# Build backends +add_subdirectory(backends) + +# Recipe for BLAS loader object +if(BUILD_SHARED_LIBS) +add_library(onemkl_blas OBJECT) +target_sources(onemkl_blas PRIVATE blas_loader.cpp) +target_include_directories(onemkl_blas + PRIVATE ${PROJECT_SOURCE_DIR}/include + ${PROJECT_SOURCE_DIR}/src + $ +) + +set_target_properties(onemkl_blas PROPERTIES + POSITION_INDEPENDENT_CODE ON +) +target_link_libraries(onemkl_blas PUBLIC ONEMKL::SYCL::SYCL) +endif() + diff --git a/src/blas/backends/CMakeLists.txt b/src/blas/backends/CMakeLists.txt new file mode 100644 index 000000000..043153e79 --- /dev/null +++ b/src/blas/backends/CMakeLists.txt @@ -0,0 +1,26 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +if(ENABLE_MKLCPU_BACKEND) + add_subdirectory(mklcpu) +endif() + +if(ENABLE_MKLGPU_BACKEND) + add_subdirectory(mklgpu) +endif() diff --git a/src/blas/backends/mklcpu/CMakeLists.txt b/src/blas/backends/mklcpu/CMakeLists.txt new file mode 100644 index 000000000..cd6ab2b3b --- /dev/null +++ b/src/blas/backends/mklcpu/CMakeLists.txt @@ -0,0 +1,69 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +set(LIB_NAME onemkl_blas_mklcpu) +set(LIB_OBJ ${LIB_NAME}_obj) + +find_package(MKL REQUIRED) + +add_library(${LIB_NAME}) +add_library(${LIB_OBJ} OBJECT + fp16.hpp cpu_common.hpp + cpu_level1.cpp cpu_level2.cpp cpu_level3.cpp cpu_batch.cpp cpu_extensions.cpp + $<$: mkl_blas_cpu_wrappers.cpp> +) + +target_include_directories(${LIB_OBJ} + PRIVATE ${PROJECT_SOURCE_DIR}/include + ${PROJECT_SOURCE_DIR}/src + ${MKL_INCLUDE} +) + +target_compile_options(${LIB_OBJ} PRIVATE ${MKL_COPT}) + +target_link_libraries(${LIB_OBJ} PUBLIC ONEMKL::SYCL::SYCL ${MKL_LINK_C}) + +target_compile_features(${LIB_OBJ} PUBLIC cxx_std_14) +set_target_properties(${LIB_OBJ} PROPERTIES + POSITION_INDEPENDENT_CODE ON +) +target_link_libraries(${LIB_NAME} PUBLIC ${LIB_OBJ}) + +#Set MKL libraries as not transitive for dynamic +if(BUILD_SHARED_LIBS) + set_target_properties(${LIB_NAME} PROPERTIES + INTERFACE_LINK_LIBRARIES ONEMKL::SYCL::SYCL + ) +endif() + +# Add major version to the library +set_target_properties(${LIB_NAME} PROPERTIES + SOVERSION ${PROJECT_VERSION_MAJOR} +) + +# Add dependencies rpath to the library +list(APPEND CMAKE_BUILD_RPATH $) + +# Add the library to install package +install(TARGETS ${LIB_OBJ} EXPORT oneMKLTargets) +install(TARGETS ${LIB_NAME} EXPORT oneMKLTargets + RUNTIME DESTINATION bin + ARCHIVE DESTINATION lib + LIBRARY DESTINATION lib +) diff --git a/src/blas/backends/mklcpu/cpu_batch.cpp b/src/blas/backends/mklcpu/cpu_batch.cpp new file mode 100644 index 000000000..e2ebddeb5 --- /dev/null +++ b/src/blas/backends/mklcpu/cpu_batch.cpp @@ -0,0 +1,1107 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include + +#include "cpu_common.hpp" + +namespace onemkl { +namespace mklcpu { + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto transa_acc = transa.get_access(cgh); + auto transb_acc = transb.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto k_acc = k.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto beta_acc = beta.get_access(cgh); + auto c_acc = c.get_access(cgh); + auto ldc_acc = ldc.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + float **a_array = (float **)::malloc(sizeof(float *) * total_size); + float **b_array = (float **)::malloc(sizeof(float *) * total_size); + float **c_array = (float **)::malloc(sizeof(float *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *k_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldc_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *transa_ = (char *)::malloc(sizeof(char) * group_count); + char *transb_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + k_[i] = k_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + ldc_[i] = ldc_acc[i]; + group_size_[i] = group_size_acc[i]; + transa_[i] = *fortran_char(transa_acc[i]); + transb_[i] = *fortran_char(transb_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + c_array[idx] = c_array[idx - 1] + offset_c; + } + idx++; + offset_a = (transa_acc[i] == transpose::nontrans) ? lda_acc[i] * k_acc[i] + : lda_acc[i] * m_acc[i]; + offset_b = (transb_acc[i] == transpose::nontrans) ? ldb_acc[i] * n_acc[i] + : ldb_acc[i] * k_acc[i]; + offset_c = ldc_acc[i] * n_acc[i]; + } + } + + ::sgemm_batch(transa_, transb_, m_, n_, k_, alpha_acc.get_pointer(), + (const float **)a_array, lda_, (const float **)b_array, ldb_, + beta_acc.get_pointer(), c_array, ldc_, (MKL_INT *)&group_count, + group_size_); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + ::free(m_); + ::free(n_); + ::free(k_); + ::free(lda_); + ::free(ldb_); + ::free(ldc_); + ::free(group_size_); + ::free(transa_); + ::free(transb_); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto transa_acc = transa.get_access(cgh); + auto transb_acc = transb.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto k_acc = k.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto beta_acc = beta.get_access(cgh); + auto c_acc = c.get_access(cgh); + auto ldc_acc = ldc.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + double **a_array = (double **)::malloc(sizeof(double *) * total_size); + double **b_array = (double **)::malloc(sizeof(double *) * total_size); + double **c_array = (double **)::malloc(sizeof(double *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *k_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldc_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *transa_ = (char *)::malloc(sizeof(char) * group_count); + char *transb_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + k_[i] = k_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + ldc_[i] = ldc_acc[i]; + group_size_[i] = group_size_acc[i]; + transa_[i] = *fortran_char(transa_acc[i]); + transb_[i] = *fortran_char(transb_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + c_array[idx] = c_array[idx - 1] + offset_c; + } + idx++; + offset_a = (transa_acc[i] == transpose::nontrans) ? lda_acc[i] * k_acc[i] + : lda_acc[i] * m_acc[i]; + offset_b = (transb_acc[i] == transpose::nontrans) ? ldb_acc[i] * n_acc[i] + : ldb_acc[i] * k_acc[i]; + offset_c = ldc_acc[i] * n_acc[i]; + } + } + + ::dgemm_batch(transa_, transb_, m_, n_, k_, alpha_acc.get_pointer(), + (const double **)a_array, lda_, (const double **)b_array, ldb_, + beta_acc.get_pointer(), c_array, ldc_, (MKL_INT *)&group_count, + group_size_); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + ::free(m_); + ::free(n_); + ::free(k_); + ::free(lda_); + ::free(ldb_); + ::free(ldc_); + ::free(group_size_); + ::free(transa_); + ::free(transb_); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto transa_acc = transa.get_access(cgh); + auto transb_acc = transb.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto k_acc = k.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto beta_acc = beta.get_access(cgh); + auto c_acc = c.get_access(cgh); + auto ldc_acc = ldc.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + MKL_Complex8 **a_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * total_size); + MKL_Complex8 **b_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * total_size); + MKL_Complex8 **c_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *k_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldc_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *transa_ = (char *)::malloc(sizeof(char) * group_count); + char *transb_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + k_[i] = k_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + ldc_[i] = ldc_acc[i]; + group_size_[i] = group_size_acc[i]; + transa_[i] = *fortran_char(transa_acc[i]); + transb_[i] = *fortran_char(transb_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + c_array[idx] = c_array[idx - 1] + offset_c; + } + idx++; + offset_a = (transa_acc[i] == transpose::nontrans) ? lda_acc[i] * k_acc[i] + : lda_acc[i] * m_acc[i]; + offset_b = (transb_acc[i] == transpose::nontrans) ? ldb_acc[i] * n_acc[i] + : ldb_acc[i] * k_acc[i]; + offset_c = ldc_acc[i] * n_acc[i]; + } + } + + ::cgemm_batch(transa_, transb_, m_, n_, k_, alpha_acc.get_pointer(), + (const MKL_Complex8 **)a_array, lda_, (const MKL_Complex8 **)b_array, + ldb_, beta_acc.get_pointer(), c_array, ldc_, (MKL_INT *)&group_count, + group_size_); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + ::free(m_); + ::free(n_); + ::free(k_); + ::free(lda_); + ::free(ldb_); + ::free(ldc_); + ::free(group_size_); + ::free(transa_); + ::free(transb_); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto transa_acc = transa.get_access(cgh); + auto transb_acc = transb.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto k_acc = k.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto beta_acc = beta.get_access(cgh); + auto c_acc = c.get_access(cgh); + auto ldc_acc = ldc.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + MKL_Complex16 **a_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * total_size); + MKL_Complex16 **b_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * total_size); + MKL_Complex16 **c_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *k_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldc_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *transa_ = (char *)::malloc(sizeof(char) * group_count); + char *transb_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + k_[i] = k_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + ldc_[i] = ldc_acc[i]; + group_size_[i] = group_size_acc[i]; + transa_[i] = *fortran_char(transa_acc[i]); + transb_[i] = *fortran_char(transb_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + c_array[idx] = c_array[idx - 1] + offset_c; + } + idx++; + offset_a = (transa_acc[i] == transpose::nontrans) ? lda_acc[i] * k_acc[i] + : lda_acc[i] * m_acc[i]; + offset_b = (transb_acc[i] == transpose::nontrans) ? ldb_acc[i] * n_acc[i] + : ldb_acc[i] * k_acc[i]; + offset_c = ldc_acc[i] * n_acc[i]; + } + } + + ::zgemm_batch(transa_, transb_, m_, n_, k_, alpha_acc.get_pointer(), + (const MKL_Complex16 **)a_array, lda_, (const MKL_Complex16 **)b_array, + ldb_, beta_acc.get_pointer(), c_array, ldc_, (MKL_INT *)&group_count, + group_size_); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + ::free(m_); + ::free(n_); + ::free(k_); + ::free(lda_); + ::free(ldb_); + ::free(ldc_); + ::free(group_size_); + ::free(transa_); + ::free(transb_); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + int64_t stride_a, cl::sycl::buffer &b, int64_t ldb, int64_t stride_b, + float beta, cl::sycl::buffer &c, int64_t ldc, int64_t stride_c, + int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto c_acc = c.get_access(cgh); + char transa_ = *fortran_char(transa); + char transb_ = *fortran_char(transb); + MKL_INT one = 1; + + host_task(cgh, [=]() { + float **a_array = (float **)::malloc(sizeof(float *) * batch_size); + float **b_array = (float **)::malloc(sizeof(float *) * batch_size); + float **c_array = (float **)::malloc(sizeof(float *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + c_array[i] = c_array[i - 1] + stride_c; + } + } + + ::sgemm_batch(&transa_, &transb_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&k, &alpha, (const float **)a_array, + (const MKL_INT *)&lda, (const float **)b_array, (const MKL_INT *)&ldb, + &beta, c_array, (const MKL_INT *)&ldc, (const MKL_INT *)&one, + (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + int64_t stride_a, cl::sycl::buffer &b, int64_t ldb, int64_t stride_b, + double beta, cl::sycl::buffer &c, int64_t ldc, int64_t stride_c, + int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto c_acc = c.get_access(cgh); + char transa_ = *fortran_char(transa); + char transb_ = *fortran_char(transb); + MKL_INT one = 1; + + host_task(cgh, [=]() { + double **a_array = (double **)::malloc(sizeof(double *) * batch_size); + double **b_array = (double **)::malloc(sizeof(double *) * batch_size); + double **c_array = (double **)::malloc(sizeof(double *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + c_array[i] = c_array[i - 1] + stride_c; + } + } + + ::dgemm_batch(&transa_, &transb_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&k, &alpha, (const double **)a_array, + (const MKL_INT *)&lda, (const double **)b_array, (const MKL_INT *)&ldb, + &beta, c_array, (const MKL_INT *)&ldc, (const MKL_INT *)&one, + (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, int64_t stride_a, cl::sycl::buffer, 1> &b, + int64_t ldb, int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc, int64_t stride_c, + int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto c_acc = c.get_access(cgh); + char transa_ = *fortran_char(transa); + char transb_ = *fortran_char(transb); + MKL_INT one = 1; + + host_task(cgh, [=]() { + MKL_Complex8 **a_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * batch_size); + MKL_Complex8 **b_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * batch_size); + MKL_Complex8 **c_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + c_array[i] = c_array[i - 1] + stride_c; + } + } + + ::cgemm_batch(&transa_, &transb_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&k, &alpha, (const MKL_Complex8 **)a_array, + (const MKL_INT *)&lda, (const MKL_Complex8 **)b_array, + (const MKL_INT *)&ldb, &beta, c_array, (const MKL_INT *)&ldc, + (const MKL_INT *)&one, (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + }); + }); +} + +void gemm_batch(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, int64_t stride_a, cl::sycl::buffer, 1> &b, + int64_t ldb, int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc, int64_t stride_c, + int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto c_acc = c.get_access(cgh); + char transa_ = *fortran_char(transa); + char transb_ = *fortran_char(transb); + MKL_INT one = 1; + + host_task(cgh, [=]() { + MKL_Complex16 **a_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * batch_size); + MKL_Complex16 **b_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * batch_size); + MKL_Complex16 **c_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + c_array[0] = c_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + c_array[i] = c_array[i - 1] + stride_c; + } + } + + ::zgemm_batch(&transa_, &transb_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&k, &alpha, (const MKL_Complex16 **)a_array, + (const MKL_INT *)&lda, (const MKL_Complex16 **)b_array, + (const MKL_INT *)&ldb, &beta, c_array, (const MKL_INT *)&ldc, + (const MKL_INT *)&one, (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + ::free(c_array); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto side_acc = left_right.get_access(cgh); + auto uplo_acc = upper_lower.get_access(cgh); + auto trans_acc = trans.get_access(cgh); + auto diag_acc = unit_diag.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + float **a_array = (float **)::malloc(sizeof(float *) * total_size); + float **b_array = (float **)::malloc(sizeof(float *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *side_ = (char *)::malloc(sizeof(char) * group_count); + char *uplo_ = (char *)::malloc(sizeof(char) * group_count); + char *trans_ = (char *)::malloc(sizeof(char) * group_count); + char *diag_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + group_size_[i] = group_size_acc[i]; + trans_[i] = *fortran_char(trans_acc[i]); + side_[i] = *fortran_char(side_acc[i]); + uplo_[i] = *fortran_char(uplo_acc[i]); + diag_[i] = *fortran_char(diag_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + } + idx++; + offset_a = + (side_acc[i] == side::left) ? lda_acc[i] * m_acc[i] : lda_acc[i] * n_acc[i]; + offset_b = ldb_acc[i] * n_acc[i]; + } + } + + ::strsm_batch(side_, uplo_, trans_, diag_, m_, n_, alpha_acc.get_pointer(), + (const float **)a_array, lda_, (float **)b_array, ldb_, + (MKL_INT *)&group_count, group_size_); + + ::free(a_array); + ::free(b_array); + ::free(m_); + ::free(n_); + ::free(lda_); + ::free(ldb_); + ::free(group_size_); + ::free(side_); + ::free(uplo_); + ::free(trans_); + ::free(diag_); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, int64_t m, int64_t n, float alpha, cl::sycl::buffer &a, + int64_t lda, int64_t stride_a, cl::sycl::buffer &b, int64_t ldb, + int64_t stride_b, int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + char trans_ = *fortran_char(trans); + char side_ = *fortran_char(left_right); + char uplo_ = *fortran_char(upper_lower); + char diag_ = *fortran_char(unit_diag); + MKL_INT one = 1; + + host_task(cgh, [=]() { + float **a_array = (float **)::malloc(sizeof(float *) * batch_size); + float **b_array = (float **)::malloc(sizeof(float *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + } + } + + ::strsm_batch(&side_, &uplo_, &trans_, &diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + &alpha, (const float **)a_array, (const MKL_INT *)&lda, (float **)b_array, + (const MKL_INT *)&ldb, (const MKL_INT *)&one, + (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto side_acc = left_right.get_access(cgh); + auto uplo_acc = upper_lower.get_access(cgh); + auto trans_acc = trans.get_access(cgh); + auto diag_acc = unit_diag.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + double **a_array = (double **)::malloc(sizeof(double *) * total_size); + double **b_array = (double **)::malloc(sizeof(double *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *side_ = (char *)::malloc(sizeof(char) * group_count); + char *uplo_ = (char *)::malloc(sizeof(char) * group_count); + char *trans_ = (char *)::malloc(sizeof(char) * group_count); + char *diag_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + group_size_[i] = group_size_acc[i]; + trans_[i] = *fortran_char(trans_acc[i]); + side_[i] = *fortran_char(side_acc[i]); + uplo_[i] = *fortran_char(uplo_acc[i]); + diag_[i] = *fortran_char(diag_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + } + idx++; + offset_a = + (side_acc[i] == side::left) ? lda_acc[i] * m_acc[i] : lda_acc[i] * n_acc[i]; + offset_b = ldb_acc[i] * n_acc[i]; + } + } + + ::dtrsm_batch(side_, uplo_, trans_, diag_, m_, n_, alpha_acc.get_pointer(), + (const double **)a_array, lda_, (double **)b_array, ldb_, + (MKL_INT *)&group_count, group_size_); + + ::free(a_array); + ::free(b_array); + ::free(m_); + ::free(n_); + ::free(lda_); + ::free(ldb_); + ::free(group_size_); + ::free(side_); + ::free(uplo_); + ::free(trans_); + ::free(diag_); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, int64_t m, int64_t n, double alpha, cl::sycl::buffer &a, + int64_t lda, int64_t stride_a, cl::sycl::buffer &b, int64_t ldb, + int64_t stride_b, int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + char trans_ = *fortran_char(trans); + char side_ = *fortran_char(left_right); + char uplo_ = *fortran_char(upper_lower); + char diag_ = *fortran_char(unit_diag); + MKL_INT one = 1; + + host_task(cgh, [=]() { + double **a_array = (double **)::malloc(sizeof(double *) * batch_size); + double **b_array = (double **)::malloc(sizeof(double *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + } + } + + ::dtrsm_batch(&side_, &uplo_, &trans_, &diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + &alpha, (const double **)a_array, (const MKL_INT *)&lda, + (double **)b_array, (const MKL_INT *)&ldb, (const MKL_INT *)&one, + (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto side_acc = left_right.get_access(cgh); + auto uplo_acc = upper_lower.get_access(cgh); + auto trans_acc = trans.get_access(cgh); + auto diag_acc = unit_diag.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + MKL_Complex8 **a_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * total_size); + MKL_Complex8 **b_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *side_ = (char *)::malloc(sizeof(char) * group_count); + char *uplo_ = (char *)::malloc(sizeof(char) * group_count); + char *trans_ = (char *)::malloc(sizeof(char) * group_count); + char *diag_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + group_size_[i] = group_size_acc[i]; + trans_[i] = *fortran_char(trans_acc[i]); + side_[i] = *fortran_char(side_acc[i]); + uplo_[i] = *fortran_char(uplo_acc[i]); + diag_[i] = *fortran_char(diag_acc[i]); + + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + } + idx++; + offset_a = + (side_acc[i] == side::left) ? lda_acc[i] * m_acc[i] : lda_acc[i] * n_acc[i]; + offset_b = ldb_acc[i] * n_acc[i]; + } + } + + ::ctrsm_batch(side_, uplo_, trans_, diag_, m_, n_, alpha_acc.get_pointer(), + (const MKL_Complex8 **)a_array, lda_, (MKL_Complex8 **)b_array, ldb_, + (MKL_INT *)&group_count, group_size_); + + ::free(a_array); + ::free(b_array); + ::free(m_); + ::free(n_); + ::free(lda_); + ::free(ldb_); + ::free(group_size_); + ::free(side_); + ::free(uplo_); + ::free(trans_); + ::free(diag_); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, int64_t stride_a, + cl::sycl::buffer, 1> &b, int64_t ldb, int64_t stride_b, + int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + char trans_ = *fortran_char(trans); + char side_ = *fortran_char(left_right); + char uplo_ = *fortran_char(upper_lower); + char diag_ = *fortran_char(unit_diag); + MKL_INT one = 1; + + host_task(cgh, [=]() { + MKL_Complex8 **a_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * batch_size); + MKL_Complex8 **b_array = (MKL_Complex8 **)::malloc(sizeof(MKL_Complex8 *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + } + } + + ::ctrsm_batch(&side_, &uplo_, &trans_, &diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + &alpha, (const MKL_Complex8 **)a_array, (const MKL_INT *)&lda, + (MKL_Complex8 **)b_array, (const MKL_INT *)&ldb, (const MKL_INT *)&one, + (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + int64_t group_count, cl::sycl::buffer &group_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto side_acc = left_right.get_access(cgh); + auto uplo_acc = upper_lower.get_access(cgh); + auto trans_acc = trans.get_access(cgh); + auto diag_acc = unit_diag.get_access(cgh); + auto m_acc = m.get_access(cgh); + auto n_acc = n.get_access(cgh); + auto alpha_acc = alpha.get_access(cgh); + auto a_acc = a.get_access(cgh); + auto lda_acc = lda.get_access(cgh); + auto b_acc = b.get_access(cgh); + auto ldb_acc = ldb.get_access(cgh); + auto group_size_acc = group_size.get_access(cgh); + + host_task(cgh, [=]() { + int64_t total_size = 0; + + for (int64_t i = 0; i < group_count; i++) { + total_size += group_size_acc[i]; + } + + MKL_Complex16 **a_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * total_size); + MKL_Complex16 **b_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * total_size); + MKL_INT *m_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *n_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *lda_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *ldb_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + MKL_INT *group_size_ = (MKL_INT *)::malloc(sizeof(MKL_INT) * group_count); + int64_t offset_a = 0, offset_b = 0, offset_c = 0, idx = 0; + char *side_ = (char *)::malloc(sizeof(char) * group_count); + char *uplo_ = (char *)::malloc(sizeof(char) * group_count); + char *trans_ = (char *)::malloc(sizeof(char) * group_count); + char *diag_ = (char *)::malloc(sizeof(char) * group_count); + + for (int64_t i = 0; i < group_count; i++) { + m_[i] = m_acc[i]; + n_[i] = n_acc[i]; + lda_[i] = lda_acc[i]; + ldb_[i] = ldb_acc[i]; + group_size_[i] = group_size_acc[i]; + trans_[i] = *fortran_char(trans_acc[i]); + side_[i] = *fortran_char(side_acc[i]); + uplo_[i] = *fortran_char(uplo_acc[i]); + diag_[i] = *fortran_char(diag_acc[i]); + for (int64_t j = 0; j < group_size_acc[i]; j++) { + if (idx == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[idx] = a_array[idx - 1] + offset_a; + b_array[idx] = b_array[idx - 1] + offset_b; + } + idx++; + offset_a = + (side_acc[i] == side::left) ? lda_acc[i] * m_acc[i] : lda_acc[i] * n_acc[i]; + offset_b = ldb_acc[i] * n_acc[i]; + } + } + + ::ztrsm_batch(side_, uplo_, trans_, diag_, m_, n_, alpha_acc.get_pointer(), + (const MKL_Complex16 **)a_array, lda_, (MKL_Complex16 **)b_array, ldb_, + (MKL_INT *)&group_count, group_size_); + + ::free(a_array); + ::free(b_array); + ::free(m_); + ::free(n_); + ::free(lda_); + ::free(ldb_); + ::free(group_size_); + ::free(side_); + ::free(uplo_); + ::free(trans_); + ::free(diag_); + }); + }); +} + +void trsm_batch(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, int64_t stride_a, + cl::sycl::buffer, 1> &b, int64_t ldb, int64_t stride_b, + int64_t batch_size) { + queue.submit([&](cl::sycl::handler &cgh) { + auto a_acc = a.get_access(cgh); + auto b_acc = b.get_access(cgh); + char trans_ = *fortran_char(trans); + char side_ = *fortran_char(left_right); + char uplo_ = *fortran_char(upper_lower); + char diag_ = *fortran_char(unit_diag); + MKL_INT one = 1; + host_task(cgh, [=]() { + MKL_Complex16 **a_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * batch_size); + MKL_Complex16 **b_array = + (MKL_Complex16 **)::malloc(sizeof(MKL_Complex16 *) * batch_size); + + for (int64_t i = 0; i < batch_size; i++) { + if (i == 0) { + a_array[0] = a_acc.get_pointer(); + b_array[0] = b_acc.get_pointer(); + } + else { + a_array[i] = a_array[i - 1] + stride_a; + b_array[i] = b_array[i - 1] + stride_b; + } + } + + ::ztrsm_batch(&side_, &uplo_, &trans_, &diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + &alpha, (const MKL_Complex16 **)a_array, (const MKL_INT *)&lda, + (MKL_Complex16 **)b_array, (const MKL_INT *)&ldb, (const MKL_INT *)&one, + (const MKL_INT *)&batch_size); + + ::free(a_array); + ::free(b_array); + }); + }); +} + +} // namespace mklcpu +} // namespace onemkl diff --git a/src/blas/backends/mklcpu/cpu_common.hpp b/src/blas/backends/mklcpu/cpu_common.hpp new file mode 100644 index 000000000..43aeab312 --- /dev/null +++ b/src/blas/backends/mklcpu/cpu_common.hpp @@ -0,0 +1,103 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _MKL_CPU_COMMON_HPP_ +#define _MKL_CPU_COMMON_HPP_ + +#define MKL_Complex8 std::complex +#define MKL_Complex16 std::complex + +#include +#include + +#include "mkl_blas.h" +#include "mkl_cblas.h" + +#include "onemkl/blas/detail/mklcpu/onemkl_blas_mklcpu.hpp" +#include "onemkl/types.hpp" + +namespace onemkl { +namespace mklcpu { + +// host_task automatically uses run_on_host_intel if it is supported by the +// compiler. Otherwise, it falls back to single_task. +template +static inline auto host_task_internal(H &cgh, F f, int) -> decltype(cgh.run_on_host_intel(f)) { + return cgh.run_on_host_intel(f); +} + +template +static inline void host_task_internal(H &cgh, F f, long) { + cgh.template single_task(f); +} + +template +static inline void host_task(H &cgh, F f) { + (void)host_task_internal(cgh, f, 0); +} + +// Conversion functions to traditional Fortran characters. +inline const char *fortran_char(transpose t) { + if (t == transpose::nontrans) + return "N"; + if (t == transpose::trans) + return "T"; + if (t == transpose::conjtrans) + return "C"; + return "N"; +} + +inline const char *fortran_char(offset t) { + if (t == offset::fix) + return "F"; + if (t == offset::row) + return "R"; + if (t == offset::column) + return "C"; + return "N"; +} + +inline const char *fortran_char(uplo u) { + if (u == uplo::upper) + return "U"; + if (u == uplo::lower) + return "L"; + return "U"; +} + +inline const char *fortran_char(diag d) { + if (d == diag::nonunit) + return "N"; + if (d == diag::unit) + return "U"; + return "N"; +} + +inline const char *fortran_char(side s) { + if (s == side::left) + return "L"; + if (s == side::right) + return "R"; + return "L"; +} + +} // namespace mklcpu +} // namespace onemkl + +#endif //_MKL_CPU_COMMON_HPP_ diff --git a/src/blas/backends/mklcpu/cpu_extensions.cpp b/src/blas/backends/mklcpu/cpu_extensions.cpp new file mode 100644 index 000000000..d053a0d4b --- /dev/null +++ b/src/blas/backends/mklcpu/cpu_extensions.cpp @@ -0,0 +1,312 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include + +#include "cpu_common.hpp" +#include "fp16.hpp" + +namespace onemkl { +namespace mklcpu { + +template +static inline void copy_mat(T_src &src, transpose trans, int64_t row, int64_t col, int64_t ld, + T_dest off, T_dest *&dest) { + int64_t i, j; + if (trans == transpose::N) { + for (j = 0; j < col; j++) { + for (i = 0; i < row; i++) { + dest[i + ld * j] = (T_dest)src[i + ld * j] - off; + } + } + } + else { + for (i = 0; i < row; i++) { + for (j = 0; j < col; j++) { + dest[i * ld + j] = (T_dest)src[i * ld + j] - off; + } + } + } +} + +template +static inline void copy_mat(T_src &src, int64_t row, int64_t col, int64_t ld, offset off_kind, + T_off off, T_dest &dest) { + using T_data = typename std::remove_reference::type; + int64_t i, j; + T_data tmp; + + if (off_kind == offset::F) { + tmp = off[0]; + for (j = 0; j < col; j++) { + for (i = 0; i < row; i++) { + dest[i + ld * j] = tmp + (T_data)src[i + ld * j]; + } + } + } + else if (off_kind == offset::C) { + for (j = 0; j < col; j++) { + for (i = 0; i < row; i++) { + tmp = off[i]; + dest[i + ld * j] = tmp + (T_data)src[i + ld * j]; + } + } + } + else { + for (j = 0; j < col; j++) { + tmp = off[j]; + for (i = 0; i < row; i++) { + dest[i + ld * j] = tmp + (T_data)src[i + ld * j]; + } + } + } +} + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, half alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, half beta, cl::sycl::buffer &c, + int64_t ldc) { + auto a_fp16 = a.reinterpret(a.get_range()); + auto b_fp16 = b.reinterpret(b.get_range()); + auto c_fp16 = c.reinterpret(c.get_range()); + + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + float f32_alpha = (float)alpha; + float f32_beta = (float)beta; + auto accessor_a = a_fp16.get_access(cgh); + auto accessor_b = b_fp16.get_access(cgh); + auto accessor_c = c_fp16.get_access(cgh); + host_task(cgh, [=]() { + int64_t sizea, sizeb, sizec; + sizea = (transa == transpose::N) ? lda * k : lda * m; + sizeb = (transb == transpose::N) ? ldb * n : ldb * k; + sizec = ldc * n; + // copy A, B and C to float + float *f32_a = (float *)::malloc(sizeof(float) * sizea); + float *f32_b = (float *)::malloc(sizeof(float) * sizeb); + float *f32_c = (float *)::malloc(sizeof(float) * sizec); + copy_mat(accessor_a, transa, m, k, lda, 0.0f, f32_a); + copy_mat(accessor_b, transb, k, n, ldb, 0.0f, f32_b); + copy_mat(accessor_c, transpose::N, m, n, ldc, 0.0f, f32_c); + ::sgemm((const char *)&transa_, (const char *)&transb_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const float *)&f32_alpha, f32_a, + (const MKL_INT *)&lda, f32_b, (const MKL_INT *)&ldb, (const float *)&f32_beta, + f32_c, (const MKL_INT *)&ldc); + // copy C back to half + fp16 co = 0.0f; + copy_mat(f32_c, m, n, ldc, offset::F, &co, accessor_c); + ::free(f32_a); + ::free(f32_b); + ::free(f32_c); + }); + }); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, + cl::sycl::buffer &c, int64_t ldc) { + gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + int64_t ldc) { + gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, half alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, half beta, cl::sycl::buffer &c, + int64_t ldc) { + gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + auto a_fp16 = a.reinterpret(a.get_range()); + auto b_fp16 = b.reinterpret(b.get_range()); + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + auto accessor_a = a_fp16.get_access(cgh); + auto accessor_b = b_fp16.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + int64_t sizea, sizeb; + sizea = (transa == transpose::N) ? lda * k : lda * m; + sizeb = (transb == transpose::N) ? ldb * n : ldb * k; + // copy A and B to float + float *f32_a = (float *)::malloc(sizeof(float) * sizea); + float *f32_b = (float *)::malloc(sizeof(float) * sizeb); + copy_mat(accessor_a, transa, m, k, lda, 0.0f, f32_a); + copy_mat(accessor_b, transb, k, n, ldb, 0.0f, f32_b); + ::sgemm((const char *)&transa_, (const char *)&transb_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const float *)&alpha, f32_a, + (const MKL_INT *)&lda, f32_b, (const MKL_INT *)&ldb, (const float *)&beta, + accessor_c.get_pointer(), (const MKL_INT *)&ldc); + ::free(f32_a); + ::free(f32_b); + }); + }); +} + +void gemm_ext(cl::sycl::queue &queue, transpose transa, transpose transb, offset offsetc, int64_t m, + int64_t n, int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + int8_t ao, cl::sycl::buffer &b, int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, int64_t ldc, cl::sycl::buffer &co) { + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + const char offsetc_ = *fortran_char(offsetc); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + auto accessor_co = co.get_access(cgh); + host_task(cgh, [=]() { + MKL_INT8 *a_mat = + static_cast(static_cast(accessor_a.get_pointer())); + MKL_UINT8 *b_mat = + static_cast(static_cast(accessor_b.get_pointer())); + MKL_INT8 bo_internal = -bo; + MKL_INT8 ao_internal = -ao; + ::gemm_s8u8s32((const char *)&transa_, (const char *)&transb_, (const char *)&offsetc_, + (const MKL_INT *)&m, (const MKL_INT *)&n, (const MKL_INT *)&k, + (const float *)&alpha, a_mat, (const MKL_INT *)&lda, &ao_internal, b_mat, + (const MKL_INT *)&ldb, &bo_internal, (const float *)&beta, + (MKL_INT32 *)accessor_c.get_pointer(), (const MKL_INT *)&ldc, + (const MKL_INT32 *)accessor_co.get_pointer()); + }); + }); +} + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::sgemmt((const char *)&upper_lower_, (const char *)&transa_, (const char *)&transb_, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const float *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const float *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::dgemmt((const char *)&upper_lower_, (const char *)&transa_, (const char *)&transb_, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const double *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const double *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::cgemmt((const char *)&upper_lower_, (const char *)&transa_, (const char *)&transb_, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const MKL_Complex8 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const MKL_Complex8 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void gemmt(cl::sycl::queue &queue, uplo upper_lower, transpose transa, transpose transb, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zgemmt((const char *)&upper_lower_, (const char *)&transa_, (const char *)&transb_, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const MKL_Complex16 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const MKL_Complex16 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +} // namespace mklcpu +} // namespace onemkl diff --git a/src/blas/backends/mklcpu/cpu_level1.cpp b/src/blas/backends/mklcpu/cpu_level1.cpp new file mode 100644 index 000000000..74e9d4d61 --- /dev/null +++ b/src/blas/backends/mklcpu/cpu_level1.cpp @@ -0,0 +1,715 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include + +#include "cpu_common.hpp" + +namespace onemkl { +namespace mklcpu { + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::sasum((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::dasum((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::scasum((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::dzasum((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void axpy(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::saxpy((const MKL_INT *)&n, (const float *)&alpha, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void axpy(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::daxpy((const MKL_INT *)&n, (const double *)&alpha, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void axpy(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::caxpy((const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void axpy(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zaxpy((const MKL_INT *)&n, (const MKL_Complex16 *)&alpha_, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::scopy((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dcopy((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::ccopy((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::zcopy((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::sdot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::ddot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::dsdot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dotc(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + ::cdotc(accessor_result.get_pointer(), (const MKL_INT *)&n, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dotc(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + ::zdotc(accessor_result.get_pointer(), (const MKL_INT *)&n, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dotu(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + ::cdotu(accessor_result.get_pointer(), (const MKL_INT *)&n, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void dotu(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + ::zdotu(accessor_result.get_pointer(), (const MKL_INT *)&n, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_isamin((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.template get_access(cgh); + auto accessor_result = result.template get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_idamin((const MKL_INT)n, accessor_x.get_pointer(), (const MKL_INT)incx); + }); + }); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_icamin((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_izamin((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_isamax((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_idamax((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_icamax((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::cblas_izamax((MKL_INT)n, accessor_x.get_pointer(), (MKL_INT)incx); + }); + }); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.template get_access(cgh); + auto accessor_result = result.template get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::snrm2((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::dnrm2((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::scnrm2((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::dznrm2((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, float c, float s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::srot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy, &c, &s); + }); + }); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, double c, double s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::drot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy, &c, &s); + }); + }); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, float c, + float s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::csrot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy, &c, &s); + }); + }); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, double c, + double s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::zdrot((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy, &c, &s); + }); + }); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + auto accessor_s = s.get_access(cgh); + host_task(cgh, [=]() { + ::srotg(accessor_a.get_pointer(), accessor_b.get_pointer(), accessor_c.get_pointer(), + accessor_s.get_pointer()); + }); + }); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + auto accessor_s = s.get_access(cgh); + host_task(cgh, [=]() { + ::drotg(accessor_a.get_pointer(), accessor_b.get_pointer(), accessor_c.get_pointer(), + accessor_s.get_pointer()); + }); + }); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + auto accessor_s = s.get_access(cgh); + host_task(cgh, [=]() { + ::crotg(accessor_a.get_pointer(), accessor_b.get_pointer(), accessor_c.get_pointer(), + accessor_s.get_pointer()); + }); + }); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + auto accessor_s = s.get_access(cgh); + host_task(cgh, [=]() { + ::zrotg(accessor_a.get_pointer(), accessor_b.get_pointer(), accessor_c.get_pointer(), + accessor_s.get_pointer()); + }); + }); +} + +void rotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_param = param.get_access(cgh); + host_task(cgh, [=]() { + ::srotm((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy, accessor_param.get_pointer()); + }); + }); +} + +void rotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_param = param.get_access(cgh); + host_task(cgh, [=]() { + ::drotm((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy, accessor_param.get_pointer()); + }); + }); +} + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_d1 = d1.get_access(cgh); + auto accessor_d2 = d2.get_access(cgh); + auto accessor_x1 = x1.get_access(cgh); + auto accessor_param = param.get_access(cgh); + host_task(cgh, [=]() { + ::srotmg(accessor_d1.get_pointer(), accessor_d2.get_pointer(), + accessor_x1.get_pointer(), (float *)&y1, accessor_param.get_pointer()); + }); + }); +} + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, cl::sycl::buffer ¶m) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_d1 = d1.get_access(cgh); + auto accessor_d2 = d2.get_access(cgh); + auto accessor_x1 = x1.get_access(cgh); + auto accessor_param = param.get_access(cgh); + host_task(cgh, [=]() { + ::drotmg(accessor_d1.get_pointer(), accessor_d2.get_pointer(), + accessor_x1.get_pointer(), (double *)&y1, accessor_param.get_pointer()); + }); + }); +} + +void scal(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::sscal((const MKL_INT *)&n, (const float *)&alpha, accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void scal(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::dscal((const MKL_INT *)&n, (const double *)&alpha, accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void scal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::cscal((const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void scal(cl::sycl::queue &queue, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::csscal((const MKL_INT *)&n, (const float *)&alpha, accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void scal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zscal((const MKL_INT *)&n, (const MKL_Complex16 *)&alpha_, accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void scal(cl::sycl::queue &queue, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::zdscal((const MKL_INT *)&n, (const double *)&alpha, accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void sdsdot(cl::sycl::queue &queue, int64_t n, float sb, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &result) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_result = result.get_access(cgh); + host_task(cgh, [=]() { + accessor_result[0] = + ::sdsdot((const MKL_INT *)&n, (const float *)&sb, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::sswap((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dswap((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::cswap((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::zswap((const MKL_INT *)&n, accessor_x.get_pointer(), (const MKL_INT *)&incx, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +} // namespace mklcpu +} // namespace onemkl diff --git a/src/blas/backends/mklcpu/cpu_level2.cpp b/src/blas/backends/mklcpu/cpu_level2.cpp new file mode 100644 index 000000000..4dc5c4cd7 --- /dev/null +++ b/src/blas/backends/mklcpu/cpu_level2.cpp @@ -0,0 +1,1178 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include + +#include "cpu_common.hpp" + +namespace onemkl { +namespace mklcpu { + +void gbmv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, int64_t kl, int64_t ku, + float alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx, float beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::sgbmv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&kl, (const MKL_INT *)&ku, (const float *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const float *)&beta, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void gbmv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, int64_t kl, int64_t ku, + double alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx, double beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dgbmv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&kl, (const MKL_INT *)&ku, (const double *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const double *)&beta, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void gbmv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, int64_t kl, int64_t ku, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::cgbmv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&kl, (const MKL_INT *)&ku, (const MKL_Complex8 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const MKL_Complex8 *)&beta_, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void gbmv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, int64_t kl, int64_t ku, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zgbmv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_INT *)&kl, (const MKL_INT *)&ku, (const MKL_Complex16 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const MKL_Complex16 *)&beta_, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void gemv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::sgemv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const float *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const float *)&beta, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void gemv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dgemv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const double *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const double *)&beta, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void gemv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::cgemv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const MKL_Complex8 *)&beta_, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void gemv(cl::sycl::queue &queue, transpose trans, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char trans_ = *fortran_char(trans); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zgemv((const char *)&trans_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const MKL_Complex16 *)&beta_, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void ger(cl::sycl::queue &queue, int64_t m, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &a, + int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::sger((const MKL_INT *)&m, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void ger(cl::sycl::queue &queue, int64_t m, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &a, + int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::dger((const MKL_INT *)&m, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void gerc(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::cgerc((const MKL_INT *)&m, (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void gerc(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zgerc((const MKL_INT *)&m, (const MKL_INT *)&n, (const MKL_Complex16 *)&alpha_, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void geru(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::cgeru((const MKL_INT *)&m, (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void geru(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zgeru((const MKL_INT *)&m, (const MKL_INT *)&n, (const MKL_Complex16 *)&alpha_, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void hbmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::chbmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_INT *)&k, + (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const MKL_Complex8 *)&beta_, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void hbmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zhbmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_INT *)&k, + (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const MKL_Complex16 *)&beta_, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void hemv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::chemv((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const MKL_Complex8 *)&beta_, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void hemv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zhemv((const char *)&upper_lower_, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const MKL_Complex16 *)&beta_, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void her(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::cher((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_a.get_pointer(), + (const MKL_INT *)&lda); + }); + }); +} + +void her(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::zher((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_a.get_pointer(), + (const MKL_INT *)&lda); + }); + }); +} + +void her2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::cher2((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void her2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zher2((const char *)&upper_lower_, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy, + accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void hpmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &ap, cl::sycl::buffer, 1> &x, + int64_t incx, std::complex beta, cl::sycl::buffer, 1> &y, + int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::chpmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, + accessor_ap.get_pointer(), accessor_x.get_pointer(), (const MKL_INT *)&incx, + (const MKL_Complex8 *)&beta_, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void hpmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &ap, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zhpmv((const char *)&upper_lower_, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_ap.get_pointer(), + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const MKL_Complex16 *)&beta_, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void hpr(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + ::chpr((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_ap.get_pointer()); + }); + }); +} + +void hpr(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + ::zhpr((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_ap.get_pointer()); + }); + }); +} + +void hpr2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::chpr2((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_ap.get_pointer()); + }); + }); +} + +void hpr2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zhpr2((const char *)&upper_lower_, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_x.get_pointer(), + (const MKL_INT *)&incx, accessor_y.get_pointer(), (const MKL_INT *)&incy, + accessor_ap.get_pointer()); + }); + }); +} + +void sbmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, int64_t k, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::ssbmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_INT *)&k, + (const float *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const float *)&beta, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void sbmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, int64_t k, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dsbmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const MKL_INT *)&k, + (const double *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx, (const double *)&beta, + accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void spmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer &ap, cl::sycl::buffer &x, int64_t incx, float beta, + cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::sspmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_ap.get_pointer(), accessor_x.get_pointer(), (const MKL_INT *)&incx, + (const float *)&beta, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void spmv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer &ap, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dspmv((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_ap.get_pointer(), accessor_x.get_pointer(), (const MKL_INT *)&incx, + (const double *)&beta, accessor_y.get_pointer(), (const MKL_INT *)&incy); + }); + }); +} + +void spr(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + ::sspr((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_ap.get_pointer()); + }); + }); +} + +void spr(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + ::dspr((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_ap.get_pointer()); + }); + }); +} + +void spr2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + ::sspr2((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_ap.get_pointer()); + }); + }); +} + +void spr2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &ap) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_ap = ap.get_access(cgh); + host_task(cgh, [=]() { + ::dspr2((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_ap.get_pointer()); + }); + }); +} + +void symv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::ssymv((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const float *)&beta, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void symv(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + host_task(cgh, [=]() { + ::dsymv((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_x.get_pointer(), + (const MKL_INT *)&incx, (const double *)&beta, accessor_y.get_pointer(), + (const MKL_INT *)&incy); + }); + }); +} + +void syr(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::ssyr((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_a.get_pointer(), + (const MKL_INT *)&lda); + }); + }); +} + +void syr(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a, + int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::dsyr((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_a.get_pointer(), + (const MKL_INT *)&lda); + }); + }); +} + +void syr2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::ssyr2((const char *)&upper_lower_, (const MKL_INT *)&n, (const float *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void syr2(cl::sycl::queue &queue, uplo upper_lower, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &a, int64_t lda) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_x = x.get_access(cgh); + auto accessor_y = y.get_access(cgh); + auto accessor_a = a.get_access(cgh); + host_task(cgh, [=]() { + ::dsyr2((const char *)&upper_lower_, (const MKL_INT *)&n, (const double *)&alpha, + accessor_x.get_pointer(), (const MKL_INT *)&incx, accessor_y.get_pointer(), + (const MKL_INT *)&incy, accessor_a.get_pointer(), (const MKL_INT *)&lda); + }); + }); +} + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::stbmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::dtbmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ctbmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ztbmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::stbsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::dtbsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ctbsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tbsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ztbsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, (const MKL_INT *)&k, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer &ap, cl::sycl::buffer &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::stpmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer &ap, cl::sycl::buffer &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::dtpmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &ap, cl::sycl::buffer, 1> &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ctpmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpmv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &ap, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ztpmv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer &ap, cl::sycl::buffer &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::stpsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer &ap, cl::sycl::buffer &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::dtpsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &ap, cl::sycl::buffer, 1> &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ctpsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void tpsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &ap, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_ap = ap.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ztpsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_ap.get_pointer(), accessor_x.get_pointer(), + (const MKL_INT *)&incx); + }); + }); +} + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose transa, diag unit_diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::strmv((const char *)&upper_lower_, (const char *)&transa_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose transa, diag unit_diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::dtrmv((const char *)&upper_lower_, (const char *)&transa_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose transa, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::ctrmv((const char *)&upper_lower_, (const char *)&transa_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trmv(cl::sycl::queue &queue, uplo upper_lower, transpose transa, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::ztrmv((const char *)&upper_lower_, (const char *)&transa_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::strsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::dtrsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ctrsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +void trsv(cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_x = x.get_access(cgh); + host_task(cgh, [=]() { + ::ztrsv((const char *)&upper_lower_, (const char *)&trans_, (const char *)&unit_diag_, + (const MKL_INT *)&n, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_x.get_pointer(), (const MKL_INT *)&incx); + }); + }); +} + +} // namespace mklcpu +} // namespace onemkl diff --git a/src/blas/backends/mklcpu/cpu_level3.cpp b/src/blas/backends/mklcpu/cpu_level3.cpp new file mode 100644 index 000000000..493db2fe8 --- /dev/null +++ b/src/blas/backends/mklcpu/cpu_level3.cpp @@ -0,0 +1,644 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include + +#include "cpu_common.hpp" + +namespace onemkl { +namespace mklcpu { + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::sgemm((const char *)&transa_, (const char *)&transb_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const float *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const float *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::dgemm((const char *)&transa_, (const char *)&transb_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const double *)&alpha, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const double *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::cgemm((const char *)&transa_, (const char *)&transb_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const MKL_Complex8 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const MKL_Complex8 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void gemm(cl::sycl::queue &queue, transpose transa, transpose transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char transa_ = *fortran_char(transa); + const char transb_ = *fortran_char(transb); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zgemm((const char *)&transa_, (const char *)&transb_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_INT *)&k, (const MKL_Complex16 *)&alpha_, + accessor_a.get_pointer(), (const MKL_INT *)&lda, accessor_b.get_pointer(), + (const MKL_INT *)&ldb, (const MKL_Complex16 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::chemm((const char *)&left_right_, (const char *)&upper_lower_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const MKL_Complex8 *)&beta_, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void hemm(cl::sycl::queue &queue, side left_right, uplo upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zhemm((const char *)&left_right_, (const char *)&upper_lower_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const MKL_Complex16 *)&beta_, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + float alpha, cl::sycl::buffer, 1> &a, int64_t lda, float beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::cherk((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const float *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, (const float *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void herk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + double alpha, cl::sycl::buffer, 1> &a, int64_t lda, double beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::zherk((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const double *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, (const double *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::cher2k((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const float *)&beta, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void her2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::zher2k((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const double *)&beta, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, int64_t m, int64_t n, + float alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb, float beta, cl::sycl::buffer &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::ssymm((const char *)&left_right_, (const char *)&upper_lower_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const float *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const float *)&beta, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, int64_t m, int64_t n, + double alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb, double beta, cl::sycl::buffer &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::dsymm((const char *)&left_right_, (const char *)&upper_lower_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const double *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const double *)&beta, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::csymm((const char *)&left_right_, (const char *)&upper_lower_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const MKL_Complex8 *)&beta_, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void symm(cl::sycl::queue &queue, side left_right, uplo upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zsymm((const char *)&left_right_, (const char *)&upper_lower_, (const MKL_INT *)&m, + (const MKL_INT *)&n, (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const MKL_Complex16 *)&beta_, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + float alpha, cl::sycl::buffer &a, int64_t lda, float beta, + cl::sycl::buffer &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::ssyrk((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const float *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, (const float *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + double alpha, cl::sycl::buffer &a, int64_t lda, double beta, + cl::sycl::buffer &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::dsyrk((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const double *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, (const double *)&beta, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::csyrk((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, (const MKL_Complex8 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void syrk(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zsyrk((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, (const MKL_Complex16 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + float alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb, float beta, cl::sycl::buffer &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::ssyr2k((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const float *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const float *)&beta, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + ::dsyr2k((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const double *)&alpha, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const double *)&beta, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + float beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex8 beta_ = { beta_real, beta_imag }; + ::csyr2k((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const MKL_Complex8 *)&beta_, accessor_c.get_pointer(), (const MKL_INT *)&ldc); + }); + }); +} + +void syr2k(cl::sycl::queue &queue, uplo upper_lower, transpose trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + queue.submit([&](cl::sycl::handler &cgh) { + const char upper_lower_ = *fortran_char(upper_lower); + const char trans_ = *fortran_char(trans); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + double beta_real = beta.real(), beta_imag = beta.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + auto accessor_c = c.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + MKL_Complex16 beta_ = { beta_real, beta_imag }; + ::zsyr2k((const char *)&upper_lower_, (const char *)&trans_, (const MKL_INT *)&n, + (const MKL_INT *)&k, (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), + (const MKL_INT *)&lda, accessor_b.get_pointer(), (const MKL_INT *)&ldb, + (const MKL_Complex16 *)&beta_, accessor_c.get_pointer(), + (const MKL_INT *)&ldc); + }); + }); +} + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, float alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::strmm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const float *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, double alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::dtrmm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const double *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::ctrmm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trmm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::ztrmm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, float alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::strsm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const float *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, double alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + ::dtrsm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const double *)&alpha, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + float alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex8 alpha_ = { alpha_real, alpha_imag }; + ::ctrsm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_Complex8 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +void trsm(cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose transa, + diag unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + queue.submit([&](cl::sycl::handler &cgh) { + const char left_right_ = *fortran_char(left_right); + const char upper_lower_ = *fortran_char(upper_lower); + const char transa_ = *fortran_char(transa); + const char unit_diag_ = *fortran_char(unit_diag); + double alpha_real = alpha.real(), alpha_imag = alpha.imag(); + auto accessor_a = a.get_access(cgh); + auto accessor_b = b.get_access(cgh); + host_task(cgh, [=]() { + MKL_Complex16 alpha_ = { alpha_real, alpha_imag }; + ::ztrsm((const char *)&left_right_, (const char *)&upper_lower_, (const char *)&transa_, + (const char *)&unit_diag_, (const MKL_INT *)&m, (const MKL_INT *)&n, + (const MKL_Complex16 *)&alpha_, accessor_a.get_pointer(), (const MKL_INT *)&lda, + accessor_b.get_pointer(), (const MKL_INT *)&ldb); + }); + }); +} + +} // namespace mklcpu +} // namespace onemkl diff --git a/src/blas/backends/mklcpu/fp16.hpp b/src/blas/backends/mklcpu/fp16.hpp new file mode 100644 index 000000000..c4e29de36 --- /dev/null +++ b/src/blas/backends/mklcpu/fp16.hpp @@ -0,0 +1,344 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _FP16_HPP_ +#define _FP16_HPP_ + +#include +#include +#include +#include + +namespace onemkl { +namespace mklcpu { + +union float_raw { + float f; + uint32_t i; +}; + +static inline uint32_t float_to_raw(float f) { + float_raw r; + r.f = f; + return r.i; +} + +static inline float raw_to_float(uint32_t i) { + float_raw r; + r.i = i; + return r.f; +} + +namespace fp16_impl { + +template +struct is_float_double { + static constexpr bool value = false; +}; +template <> +struct is_float_double { + static constexpr bool value = true; +}; +template <> +struct is_float_double { + static constexpr bool value = true; +}; + +} // namespace fp16_impl + +// fp16: seeeeemm'mmmmmmmm + +struct fp16 { + uint16_t raw; + + fp16(int raw_, bool) : raw(raw_) {} + + fp16() {} + fp16(float f); + fp16(double d) : fp16(float(d)) {} + template + fp16(T i, typename std::enable_if::value>::type *_ = nullptr) + : fp16(float(i)) {} + + inline operator float() const; + + fp16 operator+() const { + return *this; + } + fp16 operator-() const { + fp16 h = *this; + h.raw ^= 0x8000; + return h; + } + + fp16 operator++() { + return (*this = *this + 1); + } + fp16 operator++(int) { + fp16 h = *this; + ++*this; + return h; + } + fp16 operator--() { + return (*this = *this - 1); + } + fp16 operator--(int) { + fp16 h = *this; + --*this; + return h; + } + + friend float operator+(const fp16 &h1, const fp16 &h2) { + return float(h1) + float(h2); + } + friend float operator-(const fp16 &h1, const fp16 &h2) { + return float(h1) - float(h2); + } + friend float operator*(const fp16 &h1, const fp16 &h2) { + return float(h1) * float(h2); + } + friend float operator/(const fp16 &h1, const fp16 &h2) { + return float(h1) / float(h2); + } + + template + friend typename std::enable_if::value, float>::type operator+(const fp16 &h, + const T &o) { + return float(h) + float(o); + } + template + friend typename std::enable_if::value, float>::type operator-(const fp16 &h, + const T &o) { + return float(h) - float(o); + } + template + friend typename std::enable_if::value, float>::type operator*(const fp16 &h, + const T &o) { + return float(h) * float(o); + } + template + friend typename std::enable_if::value, float>::type operator/(const fp16 &h, + const T &o) { + return float(h) / float(o); + } + template + friend typename std::enable_if::value, float>::type operator+( + const T &o, const fp16 &h) { + return float(o) + float(h); + } + template + friend typename std::enable_if::value, float>::type operator-( + const T &o, const fp16 &h) { + return float(o) - float(h); + } + template + friend typename std::enable_if::value, float>::type operator*( + const T &o, const fp16 &h) { + return float(o) * float(h); + } + template + friend typename std::enable_if::value, float>::type operator/( + const T &o, const fp16 &h) { + return float(o) / float(h); + } + + template + friend typename std::enable_if::value, T>::type operator+( + const fp16 &h, const T &o) { + return float(h) + o; + } + template + friend typename std::enable_if::value, T>::type operator-( + const fp16 &h, const T &o) { + return float(h) - o; + } + template + friend typename std::enable_if::value, T>::type operator*( + const fp16 &h, const T &o) { + return float(h) * o; + } + template + friend typename std::enable_if::value, T>::type operator/( + const fp16 &h, const T &o) { + return float(h) / o; + } + template + friend typename std::enable_if::value, T>::type operator+( + const T &o, const fp16 &h) { + return o + float(h); + } + template + friend typename std::enable_if::value, T>::type operator-( + const T &o, const fp16 &h) { + return o - float(h); + } + template + friend typename std::enable_if::value, T>::type operator*( + const T &o, const fp16 &h) { + return o * float(h); + } + template + friend typename std::enable_if::value, T>::type operator/( + const T &o, const fp16 &h) { + return o / float(h); + } +}; + +fp16::fp16(float f) { + uint32_t i = float_to_raw(f); + uint32_t s = i >> 31; + uint32_t e = (i >> 23) & 0xFF; + uint32_t m = i & 0x7FFFFF; + + uint32_t ss = s; + uint32_t mm = m >> 13; + uint32_t r = m & 0x1FFF; + uint32_t ee = 0; + int32_t eee = (e - 127) + 15; + + if (e == 0) { + // Denormal/zero floats all become zero. + ee = 0; + mm = 0; + } + else if (e == 0xFF) { + // Preserve inf/nan. + ee = 0x1F; + if (m != 0 && mm == 0) + mm = 1; + } + else if (eee > 0 && eee < 0x1F) { + // Normal range. Perform round to even on mantissa. + ee = eee; + if (r > (0x1000 - (mm & 1))) { + // Round up. + mm++; + if (mm == 0x400) { + // Rounds up to next dyad (or inf). + mm = 0; + ee++; + } + } + } + else if (eee >= 0x1F) { + // Overflow. + ee = 0x1F; + mm = 0; + } + else { + // Underflow. Scale the input float, converting it + // into an equivalent denormal. + float ff = f * raw_to_float(0x01000000); + uint32_t ii = float_to_raw(ff); + ; + ee = 0; + mm = ii; + } + + raw = (ss << 15) | (ee << 10) | mm; +} + +inline fp16::operator float() const { + uint32_t ss = raw >> 15; + uint32_t ee = (raw >> 10) & 0x1F; + uint32_t mm = raw & 0x3FF; + + uint32_t s = ss; + uint32_t eee = ee - 15 + 127; + uint32_t m = mm << 13; + uint32_t e; + + if (ee == 0) { + if (mm == 0) + e = 0; + else { + // Half denormal -> float normal + return (ss ? -1 : 1) * std::scalbn((float)mm, -24); + } + } + else if (ee == 0x1F) { + // inf/nan + e = 0xFF; + } + else + e = eee; + + uint32_t f = (s << 31) | (e << 23) | m; + + return raw_to_float(f); +} + +} // namespace mklcpu +} // namespace onemkl + +namespace std { + +bool isfinite(onemkl::mklcpu::fp16 h) { + return (~h.raw & 0x7C00); +} + +onemkl::mklcpu::fp16 abs(onemkl::mklcpu::fp16 h) { + onemkl::mklcpu::fp16 a = h; + a.raw &= ~0x8000; + return a; +} + +onemkl::mklcpu::fp16 real(onemkl::mklcpu::fp16 h) { + return h; +} + +float imag(onemkl::mklcpu::fp16 h) { + return 0.0f; +} + +template <> +class numeric_limits { +public: + static onemkl::mklcpu::fp16 min() { + return onemkl::mklcpu::fp16(0x0100, false); + } + static onemkl::mklcpu::fp16 lowest() { + return onemkl::mklcpu::fp16(0xFBFF, false); + } + static onemkl::mklcpu::fp16 max() { + return onemkl::mklcpu::fp16(0x7BFF, false); + } + static onemkl::mklcpu::fp16 epsilon() { + return onemkl::mklcpu::fp16(0x1400, false); + } + static onemkl::mklcpu::fp16 round_error() { + return onemkl::mklcpu::fp16(0x3800, false); + } // 0.5ulp + static onemkl::mklcpu::fp16 infinity() { + return onemkl::mklcpu::fp16(0x7C00, false); + } + static onemkl::mklcpu::fp16 quiet_NaN() { + return onemkl::mklcpu::fp16(0x7D00, false); + } + static onemkl::mklcpu::fp16 signaling_NaN() { + return onemkl::mklcpu::fp16(0x7E00, false); + } + static onemkl::mklcpu::fp16 denorm_min() { + return onemkl::mklcpu::fp16(0x0001, false); + } +}; + +} // namespace std + +#endif //_FP16_HPP_ diff --git a/src/blas/backends/mklcpu/mkl_blas_cpu_wrappers.cpp b/src/blas/backends/mklcpu/mkl_blas_cpu_wrappers.cpp new file mode 100644 index 000000000..7d9379224 --- /dev/null +++ b/src/blas/backends/mklcpu/mkl_blas_cpu_wrappers.cpp @@ -0,0 +1,205 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include "blas/function_table.hpp" +#include "onemkl/blas/detail/mklcpu/onemkl_blas_mklcpu.hpp" + +#define WRAPPER_VERSION 1 + +extern "C" function_table_t mkl_blas_table = { + WRAPPER_VERSION, + onemkl::mklcpu::asum, + onemkl::mklcpu::asum, + onemkl::mklcpu::asum, + onemkl::mklcpu::asum, + onemkl::mklcpu::axpy, + onemkl::mklcpu::axpy, + onemkl::mklcpu::axpy, + onemkl::mklcpu::axpy, + onemkl::mklcpu::copy, + onemkl::mklcpu::copy, + onemkl::mklcpu::copy, + onemkl::mklcpu::copy, + onemkl::mklcpu::dot, + onemkl::mklcpu::dot, + onemkl::mklcpu::dot, + onemkl::mklcpu::dotc, + onemkl::mklcpu::dotc, + onemkl::mklcpu::dotu, + onemkl::mklcpu::dotu, + onemkl::mklcpu::iamin, + onemkl::mklcpu::iamin, + onemkl::mklcpu::iamin, + onemkl::mklcpu::iamin, + onemkl::mklcpu::iamax, + onemkl::mklcpu::iamax, + onemkl::mklcpu::iamax, + onemkl::mklcpu::iamax, + onemkl::mklcpu::nrm2, + onemkl::mklcpu::nrm2, + onemkl::mklcpu::nrm2, + onemkl::mklcpu::nrm2, + onemkl::mklcpu::rot, + onemkl::mklcpu::rot, + onemkl::mklcpu::rot, + onemkl::mklcpu::rot, + onemkl::mklcpu::rotg, + onemkl::mklcpu::rotg, + onemkl::mklcpu::rotg, + onemkl::mklcpu::rotg, + onemkl::mklcpu::rotm, + onemkl::mklcpu::rotm, + onemkl::mklcpu::rotmg, + onemkl::mklcpu::rotmg, + onemkl::mklcpu::scal, + onemkl::mklcpu::scal, + onemkl::mklcpu::scal, + onemkl::mklcpu::scal, + onemkl::mklcpu::scal, + onemkl::mklcpu::scal, + onemkl::mklcpu::sdsdot, + onemkl::mklcpu::swap, + onemkl::mklcpu::swap, + onemkl::mklcpu::swap, + onemkl::mklcpu::swap, + onemkl::mklcpu::gbmv, + onemkl::mklcpu::gbmv, + onemkl::mklcpu::gbmv, + onemkl::mklcpu::gbmv, + onemkl::mklcpu::gemv, + onemkl::mklcpu::gemv, + onemkl::mklcpu::gemv, + onemkl::mklcpu::gemv, + onemkl::mklcpu::ger, + onemkl::mklcpu::ger, + onemkl::mklcpu::gerc, + onemkl::mklcpu::gerc, + onemkl::mklcpu::geru, + onemkl::mklcpu::geru, + onemkl::mklcpu::hbmv, + onemkl::mklcpu::hbmv, + onemkl::mklcpu::hemv, + onemkl::mklcpu::hemv, + onemkl::mklcpu::her, + onemkl::mklcpu::her, + onemkl::mklcpu::her2, + onemkl::mklcpu::her2, + onemkl::mklcpu::hpmv, + onemkl::mklcpu::hpmv, + onemkl::mklcpu::hpr, + onemkl::mklcpu::hpr, + onemkl::mklcpu::hpr2, + onemkl::mklcpu::hpr2, + onemkl::mklcpu::sbmv, + onemkl::mklcpu::sbmv, + onemkl::mklcpu::spmv, + onemkl::mklcpu::spmv, + onemkl::mklcpu::spr, + onemkl::mklcpu::spr, + onemkl::mklcpu::spr2, + onemkl::mklcpu::spr2, + onemkl::mklcpu::symv, + onemkl::mklcpu::symv, + onemkl::mklcpu::syr, + onemkl::mklcpu::syr, + onemkl::mklcpu::syr2, + onemkl::mklcpu::syr2, + onemkl::mklcpu::tbmv, + onemkl::mklcpu::tbmv, + onemkl::mklcpu::tbmv, + onemkl::mklcpu::tbmv, + onemkl::mklcpu::tbsv, + onemkl::mklcpu::tbsv, + onemkl::mklcpu::tbsv, + onemkl::mklcpu::tbsv, + onemkl::mklcpu::tpmv, + onemkl::mklcpu::tpmv, + onemkl::mklcpu::tpmv, + onemkl::mklcpu::tpmv, + onemkl::mklcpu::tpsv, + onemkl::mklcpu::tpsv, + onemkl::mklcpu::tpsv, + onemkl::mklcpu::tpsv, + onemkl::mklcpu::trmv, + onemkl::mklcpu::trmv, + onemkl::mklcpu::trmv, + onemkl::mklcpu::trmv, + onemkl::mklcpu::trsv, + onemkl::mklcpu::trsv, + onemkl::mklcpu::trsv, + onemkl::mklcpu::trsv, + onemkl::mklcpu::gemm, + onemkl::mklcpu::gemm, + onemkl::mklcpu::gemm, + onemkl::mklcpu::gemm, + onemkl::mklcpu::gemm, + onemkl::mklcpu::hemm, + onemkl::mklcpu::hemm, + onemkl::mklcpu::herk, + onemkl::mklcpu::herk, + onemkl::mklcpu::her2k, + onemkl::mklcpu::her2k, + onemkl::mklcpu::symm, + onemkl::mklcpu::symm, + onemkl::mklcpu::symm, + onemkl::mklcpu::symm, + onemkl::mklcpu::syrk, + onemkl::mklcpu::syrk, + onemkl::mklcpu::syrk, + onemkl::mklcpu::syrk, + onemkl::mklcpu::syr2k, + onemkl::mklcpu::syr2k, + onemkl::mklcpu::syr2k, + onemkl::mklcpu::syr2k, + onemkl::mklcpu::trmm, + onemkl::mklcpu::trmm, + onemkl::mklcpu::trmm, + onemkl::mklcpu::trmm, + onemkl::mklcpu::trsm, + onemkl::mklcpu::trsm, + onemkl::mklcpu::trsm, + onemkl::mklcpu::trsm, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::gemm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::trsm_batch, + onemkl::mklcpu::gemmt, + onemkl::mklcpu::gemmt, + onemkl::mklcpu::gemmt, + onemkl::mklcpu::gemmt, + onemkl::mklcpu::gemm_ext, + onemkl::mklcpu::gemm_ext, + onemkl::mklcpu::gemm_ext, + onemkl::mklcpu::gemm_ext, + onemkl::mklcpu::gemm_ext, + onemkl::mklcpu::gemm_ext, + onemkl::mklcpu::gemm_ext, +}; diff --git a/src/blas/backends/mklgpu/CMakeLists.txt b/src/blas/backends/mklgpu/CMakeLists.txt new file mode 100644 index 000000000..731a42658 --- /dev/null +++ b/src/blas/backends/mklgpu/CMakeLists.txt @@ -0,0 +1,68 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +set(LIB_NAME onemkl_blas_mklgpu) +set(LIB_OBJ ${LIB_NAME}_obj) +find_package(MKL REQUIRED) + +add_library(${LIB_NAME}) +add_library(${LIB_OBJ} OBJECT + mkl_internal_blas_gpu_wrappers.cpp + mkl_blas_sycl_buffer.cpp + $<$: mkl_blas_gpu_wrappers.cpp> +) + +target_include_directories(${LIB_OBJ} + PRIVATE ${PROJECT_SOURCE_DIR}/include + ${PROJECT_SOURCE_DIR}/src + ${MKL_INCLUDE} +) + +target_compile_options(${LIB_OBJ} PRIVATE ${MKL_COPT}) + +target_link_libraries(${LIB_OBJ} PUBLIC ONEMKL::SYCL::SYCL ${MKL_LINK_SYCL}) + +target_compile_features(${LIB_OBJ} PUBLIC cxx_std_14) +set_target_properties(${LIB_OBJ} PROPERTIES + POSITION_INDEPENDENT_CODE ON +) +target_link_libraries(${LIB_NAME} PUBLIC ${LIB_OBJ}) + +#Set MKL libraries as not transitive for dynamic +if(BUILD_SHARED_LIBS) + set_target_properties(${LIB_NAME} PROPERTIES + INTERFACE_LINK_LIBRARIES ONEMKL::SYCL::SYCL + ) +endif() + +# Add major version to the library +set_target_properties(${LIB_NAME} PROPERTIES + SOVERSION ${PROJECT_VERSION_MAJOR} +) + +# Add dependencies rpath to the library +list(APPEND CMAKE_BUILD_RPATH $) + +# Add the library to install package +install(TARGETS ${LIB_OBJ} EXPORT oneMKLTargets) +install(TARGETS ${LIB_NAME} EXPORT oneMKLTargets + RUNTIME DESTINATION bin + ARCHIVE DESTINATION lib + LIBRARY DESTINATION lib +) diff --git a/src/blas/backends/mklgpu/mkl_blas_gpu_wrappers.cpp b/src/blas/backends/mklgpu/mkl_blas_gpu_wrappers.cpp new file mode 100644 index 000000000..7d9a7aeb5 --- /dev/null +++ b/src/blas/backends/mklgpu/mkl_blas_gpu_wrappers.cpp @@ -0,0 +1,205 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include "blas/function_table.hpp" +#include "onemkl/blas/detail/mklgpu/onemkl_blas_mklgpu.hpp" + +#define WRAPPER_VERSION 1 + +extern "C" function_table_t mkl_blas_table = { + WRAPPER_VERSION, + onemkl::mklgpu::asum, + onemkl::mklgpu::asum, + onemkl::mklgpu::asum, + onemkl::mklgpu::asum, + onemkl::mklgpu::axpy, + onemkl::mklgpu::axpy, + onemkl::mklgpu::axpy, + onemkl::mklgpu::axpy, + onemkl::mklgpu::copy, + onemkl::mklgpu::copy, + onemkl::mklgpu::copy, + onemkl::mklgpu::copy, + onemkl::mklgpu::dot, + onemkl::mklgpu::dot, + onemkl::mklgpu::dot, + onemkl::mklgpu::dotc, + onemkl::mklgpu::dotc, + onemkl::mklgpu::dotu, + onemkl::mklgpu::dotu, + onemkl::mklgpu::iamin, + onemkl::mklgpu::iamin, + onemkl::mklgpu::iamin, + onemkl::mklgpu::iamin, + onemkl::mklgpu::iamax, + onemkl::mklgpu::iamax, + onemkl::mklgpu::iamax, + onemkl::mklgpu::iamax, + onemkl::mklgpu::nrm2, + onemkl::mklgpu::nrm2, + onemkl::mklgpu::nrm2, + onemkl::mklgpu::nrm2, + onemkl::mklgpu::rot, + onemkl::mklgpu::rot, + onemkl::mklgpu::rot, + onemkl::mklgpu::rot, + onemkl::mklgpu::rotg, + onemkl::mklgpu::rotg, + onemkl::mklgpu::rotg, + onemkl::mklgpu::rotg, + onemkl::mklgpu::rotm, + onemkl::mklgpu::rotm, + onemkl::mklgpu::rotmg, + onemkl::mklgpu::rotmg, + onemkl::mklgpu::scal, + onemkl::mklgpu::scal, + onemkl::mklgpu::scal, + onemkl::mklgpu::scal, + onemkl::mklgpu::scal, + onemkl::mklgpu::scal, + onemkl::mklgpu::sdsdot, + onemkl::mklgpu::swap, + onemkl::mklgpu::swap, + onemkl::mklgpu::swap, + onemkl::mklgpu::swap, + onemkl::mklgpu::gbmv, + onemkl::mklgpu::gbmv, + onemkl::mklgpu::gbmv, + onemkl::mklgpu::gbmv, + onemkl::mklgpu::gemv, + onemkl::mklgpu::gemv, + onemkl::mklgpu::gemv, + onemkl::mklgpu::gemv, + onemkl::mklgpu::ger, + onemkl::mklgpu::ger, + onemkl::mklgpu::gerc, + onemkl::mklgpu::gerc, + onemkl::mklgpu::geru, + onemkl::mklgpu::geru, + onemkl::mklgpu::hbmv, + onemkl::mklgpu::hbmv, + onemkl::mklgpu::hemv, + onemkl::mklgpu::hemv, + onemkl::mklgpu::her, + onemkl::mklgpu::her, + onemkl::mklgpu::her2, + onemkl::mklgpu::her2, + onemkl::mklgpu::hpmv, + onemkl::mklgpu::hpmv, + onemkl::mklgpu::hpr, + onemkl::mklgpu::hpr, + onemkl::mklgpu::hpr2, + onemkl::mklgpu::hpr2, + onemkl::mklgpu::sbmv, + onemkl::mklgpu::sbmv, + onemkl::mklgpu::spmv, + onemkl::mklgpu::spmv, + onemkl::mklgpu::spr, + onemkl::mklgpu::spr, + onemkl::mklgpu::spr2, + onemkl::mklgpu::spr2, + onemkl::mklgpu::symv, + onemkl::mklgpu::symv, + onemkl::mklgpu::syr, + onemkl::mklgpu::syr, + onemkl::mklgpu::syr2, + onemkl::mklgpu::syr2, + onemkl::mklgpu::tbmv, + onemkl::mklgpu::tbmv, + onemkl::mklgpu::tbmv, + onemkl::mklgpu::tbmv, + onemkl::mklgpu::tbsv, + onemkl::mklgpu::tbsv, + onemkl::mklgpu::tbsv, + onemkl::mklgpu::tbsv, + onemkl::mklgpu::tpmv, + onemkl::mklgpu::tpmv, + onemkl::mklgpu::tpmv, + onemkl::mklgpu::tpmv, + onemkl::mklgpu::tpsv, + onemkl::mklgpu::tpsv, + onemkl::mklgpu::tpsv, + onemkl::mklgpu::tpsv, + onemkl::mklgpu::trmv, + onemkl::mklgpu::trmv, + onemkl::mklgpu::trmv, + onemkl::mklgpu::trmv, + onemkl::mklgpu::trsv, + onemkl::mklgpu::trsv, + onemkl::mklgpu::trsv, + onemkl::mklgpu::trsv, + onemkl::mklgpu::gemm, + onemkl::mklgpu::gemm, + onemkl::mklgpu::gemm, + onemkl::mklgpu::gemm, + onemkl::mklgpu::gemm, + onemkl::mklgpu::hemm, + onemkl::mklgpu::hemm, + onemkl::mklgpu::herk, + onemkl::mklgpu::herk, + onemkl::mklgpu::her2k, + onemkl::mklgpu::her2k, + onemkl::mklgpu::symm, + onemkl::mklgpu::symm, + onemkl::mklgpu::symm, + onemkl::mklgpu::symm, + onemkl::mklgpu::syrk, + onemkl::mklgpu::syrk, + onemkl::mklgpu::syrk, + onemkl::mklgpu::syrk, + onemkl::mklgpu::syr2k, + onemkl::mklgpu::syr2k, + onemkl::mklgpu::syr2k, + onemkl::mklgpu::syr2k, + onemkl::mklgpu::trmm, + onemkl::mklgpu::trmm, + onemkl::mklgpu::trmm, + onemkl::mklgpu::trmm, + onemkl::mklgpu::trsm, + onemkl::mklgpu::trsm, + onemkl::mklgpu::trsm, + onemkl::mklgpu::trsm, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::gemm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::trsm_batch, + onemkl::mklgpu::gemmt, + onemkl::mklgpu::gemmt, + onemkl::mklgpu::gemmt, + onemkl::mklgpu::gemmt, + onemkl::mklgpu::gemm_ext, + onemkl::mklgpu::gemm_ext, + onemkl::mklgpu::gemm_ext, + onemkl::mklgpu::gemm_ext, + onemkl::mklgpu::gemm_ext, + onemkl::mklgpu::gemm_ext, + onemkl::mklgpu::gemm_ext, +}; diff --git a/src/blas/backends/mklgpu/mkl_blas_sycl_buffer.cpp b/src/blas/backends/mklgpu/mkl_blas_sycl_buffer.cpp new file mode 100644 index 000000000..a87a4e641 --- /dev/null +++ b/src/blas/backends/mklgpu/mkl_blas_sycl_buffer.cpp @@ -0,0 +1,1252 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +/********************************************************* + * AUTOGENERATED FILE; DO NOT EDIT + ********************************************************/ + +#include + +#include "mkl_internal_blas_gpu_wrappers.hpp" +#include "onemkl/types.hpp" + +namespace onemkl { +namespace mklgpu { + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemm(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::symm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::hemm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::hemm(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + double beta, cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::syrk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::herk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::herk(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::syr2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::her2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::her2k(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, beta, c, + ldc); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trmm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb) { + onemkl::mklgpu::internal::trsm(queue, left_right, upper_lower, trans, unit_diag, m, n, alpha, a, + lda, b, ldb); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + onemkl::mklgpu::internal::gemv(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + onemkl::mklgpu::internal::gbmv(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + onemkl::mklgpu::internal::ger(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + onemkl::mklgpu::internal::ger(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::gerc(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::gerc(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::geru(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::geru(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void hbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::hbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void hbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + onemkl::mklgpu::internal::hbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void hemv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::hemv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void hemv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + onemkl::mklgpu::internal::hemv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void her(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::her(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void her(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::her(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void her2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::her2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void her2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + onemkl::mklgpu::internal::her2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void hpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::hpmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void hpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + onemkl::mklgpu::internal::hpmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void hpr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { + onemkl::mklgpu::internal::hpr(queue, upper_lower, n, alpha, x, incx, a); +} + +void hpr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { + onemkl::mklgpu::internal::hpr(queue, upper_lower, n, alpha, x, incx, a); +} + +void hpr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { + onemkl::mklgpu::internal::hpr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void hpr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { + onemkl::mklgpu::internal::hpr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void sbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::sbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void sbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::sbmv(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void symv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::symv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void symv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::symv(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void syr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda) { + onemkl::mklgpu::internal::syr(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void syr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda) { + onemkl::mklgpu::internal::syr(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void syr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + onemkl::mklgpu::internal::syr2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void syr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + onemkl::mklgpu::internal::syr2(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void spmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::spmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void spmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::spmv(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void spr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a) { + onemkl::mklgpu::internal::spr(queue, upper_lower, n, alpha, x, incx, a); +} + +void spr(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a) { + onemkl::mklgpu::internal::spr(queue, upper_lower, n, alpha, x, incx, a); +} + +void spr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a) { + onemkl::mklgpu::internal::spr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void spr2(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a) { + onemkl::mklgpu::internal::spr2(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbmv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tbsv(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpmv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::tpsv(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::trmv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + onemkl::mklgpu::internal::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::trsv(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + onemkl::mklgpu::internal::dotc(queue, n, x, incx, y, incy, result); +} + +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + onemkl::mklgpu::internal::dotc(queue, n, x, incx, y, incy, result); +} + +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + onemkl::mklgpu::internal::dotu(queue, n, x, incx, y, incy, result); +} + +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + onemkl::mklgpu::internal::dotu(queue, n, x, incx, y, incy, result); +} + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamax(queue, n, x, incx, result); +} + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamax(queue, n, x, incx, result); +} + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamax(queue, n, x, incx, result); +} + +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamax(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamin(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamin(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamin(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::iamin(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::asum(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::asum(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::asum(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::asum(queue, n, x, incx, result); +} + +void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::axpy(queue, n, alpha, x, incx, y, incy); +} + +void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::axpy(queue, n, alpha, x, incx, y, incy); +} + +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::axpy(queue, n, alpha, x, incx, y, incy); +} + +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::axpy(queue, n, alpha, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::copy(queue, n, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::copy(queue, n, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::copy(queue, n, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::copy(queue, n, x, incx, y, incy); +} + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::dot(queue, n, x, incx, y, incy, result); +} + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::dot(queue, n, x, incx, y, incy, result); +} + +void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::sdsdot(queue, n, sb, x, incx, y, incy, result); +} + +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::dot(queue, n, x, incx, y, incy, result); +} + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::nrm2(queue, n, x, incx, result); +} + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result) { + onemkl::mklgpu::internal::nrm2(queue, n, x, incx, result); +} + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::nrm2(queue, n, x, incx, result); +} + +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result) { + onemkl::mklgpu::internal::nrm2(queue, n, x, incx, result); +} + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s) { + onemkl::mklgpu::internal::rot(queue, n, x, incx, y, incy, c, s); +} + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + double c, double s) { + onemkl::mklgpu::internal::rot(queue, n, x, incx, y, incy, c, s); +} + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, float c, float s) { + onemkl::mklgpu::internal::rot(queue, n, x, incx, y, incy, c, s); +} + +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, double c, double s) { + onemkl::mklgpu::internal::rot(queue, n, x, incx, y, incy, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s) { + onemkl::mklgpu::internal::rotg(queue, a, b, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s) { + onemkl::mklgpu::internal::rotg(queue, a, b, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + onemkl::mklgpu::internal::rotg(queue, a, b, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + onemkl::mklgpu::internal::rotg(queue, a, b, c, s); +} + +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m) { + onemkl::mklgpu::internal::rotm(queue, n, x, incx, y, incy, param); +} + +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m) { + onemkl::mklgpu::internal::rotm(queue, n, x, incx, y, incy, param); +} + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m) { + onemkl::mklgpu::internal::rotmg(queue, d1, d2, x1, y1, param); +} + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, cl::sycl::buffer ¶m) { + onemkl::mklgpu::internal::rotmg(queue, d1, d2, x1, y1, param); +} + +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx) { + onemkl::mklgpu::internal::scal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx) { + onemkl::mklgpu::internal::scal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::scal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::scal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::scal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + onemkl::mklgpu::internal::scal(queue, n, alpha, x, incx); +} + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::swap(queue, n, x, incx, y, incy); +} + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy) { + onemkl::mklgpu::internal::swap(queue, n, x, incx, y, incy); +} + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::swap(queue, n, x, incx, y, incy); +} + +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy) { + onemkl::mklgpu::internal::swap(queue, n, x, incx, y, incy); +} + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &k, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + cl::sycl::buffer &beta, cl::sycl::buffer &c, + cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &k, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + cl::sycl::buffer &beta, cl::sycl::buffer &c, + cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void gemm_batch(cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void gemm_batch( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, std::int64_t stride_c, + std::int64_t batch_size) { + //UNSUPPORTED +} + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + double beta, cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { + //UNSUPPORTED +} + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { + //UNSUPPORTED +} + +void gemm_batch(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void trsm_batch( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { + //UNSUPPORTED +} + +void trsm_batch(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { + //UNSUPPORTED +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemmt(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemmt(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemmt(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + onemkl::mklgpu::internal::gemmt(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + onemkl::mklgpu::internal::gemm_ext(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + onemkl::offset offsetc, std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, int8_t ao, + cl::sycl::buffer &b, std::int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, std::int64_t ldc, cl::sycl::buffer &co) { + //UNSUPPORTED +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + //UNSUPPORTED +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc) { + //UNSUPPORTED +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + //UNSUPPORTED +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + //UNSUPPORTED +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, half beta, cl::sycl::buffer &c, std::int64_t ldc) { + //UNSUPPORTED +} + +} // namespace mklgpu +} // namespace onemkl diff --git a/src/blas/backends/mklgpu/mkl_internal_blas_gpu_wrappers.cpp b/src/blas/backends/mklgpu/mkl_internal_blas_gpu_wrappers.cpp new file mode 100644 index 000000000..c8ea56c71 --- /dev/null +++ b/src/blas/backends/mklgpu/mkl_internal_blas_gpu_wrappers.cpp @@ -0,0 +1,1047 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +/********************************************************* + * AUTOGENERATED FILE; DO NOT EDIT + ********************************************************/ + +#include + +#include "mkl_internal_blas_gpu_wrappers.hpp" +#include "mkl_internal_blas_sycl_gpu.hpp" + +namespace onemkl { +namespace mklgpu { +namespace internal { + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, int64_t m, + int64_t n, int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::sgemm(queue, mkl::cblas_convert(transa), mkl::cblas_convert(transb), m, n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, int64_t m, + int64_t n, int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::dgemm(queue, mkl::cblas_convert(transa), mkl::cblas_convert(transb), m, n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, int64_t m, + int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::cgemm(queue, mkl::cblas_convert(transa), mkl::cblas_convert(transb), m, n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, int64_t m, + int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zgemm(queue, mkl::cblas_convert(transa), mkl::cblas_convert(transb), m, n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, int64_t m, + int64_t n, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::ssymm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), m, n, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, int64_t m, + int64_t n, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::dsymm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), m, n, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, int64_t m, + int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::csymm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), m, n, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, int64_t m, + int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zsymm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), m, n, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, int64_t m, + int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::chemm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), m, n, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, int64_t m, + int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zhemm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), m, n, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, float beta, + cl::sycl::buffer &c, int64_t ldc) { + mkl::gpu::ssyrk(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, double beta, + cl::sycl::buffer &c, int64_t ldc) { + mkl::gpu::dsyrk(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + int64_t ldc) { + mkl::gpu::csyrk(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, beta, c, ldc); +} + +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + int64_t ldc) { + mkl::gpu::zsyrk(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, beta, c, ldc); +} + +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, float alpha, cl::sycl::buffer, 1> &a, int64_t lda, + float beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::cherk(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, beta, c, ldc); +} + +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, double alpha, cl::sycl::buffer, 1> &a, int64_t lda, + double beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zherk(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, beta, c, ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::ssyr2k(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::dsyr2k(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::csyr2k(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zsyr2k(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::cher2k(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zher2k(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(trans), n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, int64_t ldb) { + mkl::gpu::strmm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb) { + mkl::gpu::dtrmm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + mkl::gpu::ctrmm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + mkl::gpu::ztrmm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, int64_t ldb) { + mkl::gpu::strsm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb) { + mkl::gpu::dtrsm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + mkl::gpu::ctrsm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb) { + mkl::gpu::ztrsm(queue, mkl::cblas_convert(left_right), mkl::cblas_convert(upper_lower), + mkl::cblas_convert(transa), mkl::cblas_convert(unit_diag), m, n, alpha, a, lda, + b, ldb); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::sgemv(queue, mkl::cblas_convert(trans), m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::dgemv(queue, mkl::cblas_convert(trans), m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::cgemv(queue, mkl::cblas_convert(trans), m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zgemv(queue, mkl::cblas_convert(trans), m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx, float beta, cl::sycl::buffer &y, + int64_t incy) { + mkl::gpu::sgbmv(queue, mkl::cblas_convert(trans), m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx, double beta, cl::sycl::buffer &y, + int64_t incy) { + mkl::gpu::dgbmv(queue, mkl::cblas_convert(trans), m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &x, int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::cgbmv(queue, mkl::cblas_convert(trans), m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &x, int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zgbmv(queue, mkl::cblas_convert(trans), m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void ger(cl::sycl::queue &queue, int64_t m, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &a, + int64_t lda) { + mkl::gpu::sger(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void ger(cl::sycl::queue &queue, int64_t m, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &a, + int64_t lda) { + mkl::gpu::dger(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void gerc(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::cgerc(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void gerc(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::zgerc(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void geru(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::cgeru(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void geru(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::zgeru(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void hbmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::chbmv(queue, mkl::cblas_convert(uplo), n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void hbmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zhbmv(queue, mkl::cblas_convert(uplo), n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void hemv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::chemv(queue, mkl::cblas_convert(uplo), n, alpha, a, lda, x, incx, beta, y, incy); +} + +void hemv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zhemv(queue, mkl::cblas_convert(uplo), n, alpha, a, lda, x, incx, beta, y, incy); +} + +void her(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::cher(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a, lda); +} + +void her(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::zher(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a, lda); +} + +void her2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::cher2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a, lda); +} + +void her2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda) { + mkl::gpu::zher2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a, lda); +} + +void hpmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + int64_t incx, std::complex beta, cl::sycl::buffer, 1> &y, + int64_t incy) { + mkl::gpu::chpmv(queue, mkl::cblas_convert(uplo), n, alpha, a, x, incx, beta, y, incy); +} + +void hpmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zhpmv(queue, mkl::cblas_convert(uplo), n, alpha, a, x, incx, beta, y, incy); +} + +void hpr(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a) { + mkl::gpu::chpr(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a); +} + +void hpr(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a) { + mkl::gpu::zhpr(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a); +} + +void hpr2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a) { + mkl::gpu::chpr2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a); +} + +void hpr2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a) { + mkl::gpu::zhpr2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a); +} + +void sbmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, int64_t k, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::ssbmv(queue, mkl::cblas_convert(uplo), n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void sbmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, int64_t k, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::dsbmv(queue, mkl::cblas_convert(uplo), n, k, alpha, a, lda, x, incx, beta, y, incy); +} + +void spmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx, float beta, + cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::sspmv(queue, mkl::cblas_convert(uplo), n, alpha, a, x, incx, beta, y, incy); +} + +void spmv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx, double beta, + cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::dspmv(queue, mkl::cblas_convert(uplo), n, alpha, a, x, incx, beta, y, incy); +} + +void spr(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a) { + mkl::gpu::sspr(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a); +} + +void spr(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a) { + mkl::gpu::dspr(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a); +} + +void spr2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &a) { + mkl::gpu::sspr2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a); +} + +void spr2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &a) { + mkl::gpu::dspr2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a); +} + +void symv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::ssymv(queue, mkl::cblas_convert(uplo), n, alpha, a, lda, x, incx, beta, y, incy); +} + +void symv(cl::sycl::queue &queue, onemkl::uplo uplo, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::dsymv(queue, mkl::cblas_convert(uplo), n, alpha, a, lda, x, incx, beta, y, incy); +} + +void syr(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a, int64_t lda) { + mkl::gpu::ssyr(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a, lda); +} + +void syr(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a, + int64_t lda) { + mkl::gpu::dsyr(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, a, lda); +} + +void syr2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &a, int64_t lda) { + mkl::gpu::ssyr2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a, lda); +} + +void syr2(cl::sycl::queue &queue, onemkl::uplo upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &a, int64_t lda) { + mkl::gpu::dsyr2(queue, mkl::cblas_convert(upplo), n, alpha, x, incx, y, incy, a, lda); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::stbmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::dtbmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ctbmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ztbmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::stbsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::dtbsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ctbsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ztbsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, k, a, lda, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::stpmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::dtpmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ctpmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ztpmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::stpsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx) { + mkl::gpu::dtpsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ctpsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ztpsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + mkl::gpu::strmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + mkl::gpu::dtrmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ctrmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ztrmv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + mkl::gpu::strsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx) { + mkl::gpu::dtrsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ctrsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + int64_t n, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::ztrsv(queue, mkl::cblas_convert(upplo), mkl::cblas_convert(trans), + mkl::cblas_convert(diag), n, a, lda, x, incx); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::scasum(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::dzasum(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::sasum(queue, n, x, incx, result); +} + +void asum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::dasum(queue, n, x, incx, result); +} + +void axpy(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::saxpy(queue, n, alpha, x, incx, y, incy); +} + +void axpy(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::daxpy(queue, n, alpha, x, incx, y, incy); +} + +void axpy(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::caxpy(queue, n, alpha, x, incx, y, incy); +} + +void axpy(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zaxpy(queue, n, alpha, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::scopy(queue, n, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::dcopy(queue, n, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::ccopy(queue, n, x, incx, y, incy); +} + +void copy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zcopy(queue, n, x, incx, y, incy); +} + +void dot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result) { + mkl::gpu::sdot(queue, n, x, incx, y, incy, result); +} + +void dot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result) { + mkl::gpu::ddot(queue, n, x, incx, y, incy, result); +} + +void sdsdot(cl::sycl::queue &queue, int64_t n, float sb, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &result) { + mkl::gpu::sdsdot(queue, n, sb, x, incx, y, incy, result); +} + +void dot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result) { + mkl::gpu::dsdot(queue, n, x, incx, y, incy, result); +} + +void dotc(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + mkl::gpu::cdotc(queue, n, x, incx, y, incy, result); +} + +void dotc(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + mkl::gpu::zdotc(queue, n, x, incx, y, incy, result); +} + +void dotu(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + mkl::gpu::cdotu(queue, n, x, incx, y, incy, result); +} + +void dotu(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result) { + mkl::gpu::zdotu(queue, n, x, incx, y, incy, result); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::scnrm2(queue, n, x, incx, result); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::dznrm2(queue, n, x, incx, result); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::snrm2(queue, n, x, incx, result); +} + +void nrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::dnrm2(queue, n, x, incx, result); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, float c, + float s) { + mkl::gpu::csrot(queue, n, x, incx, y, incy, c, s); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, double c, + double s) { + mkl::gpu::zdrot(queue, n, x, incx, y, incy, c, s); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, float c, float s) { + mkl::gpu::srot(queue, n, x, incx, y, incy, c, s); +} + +void rot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, double c, double s) { + mkl::gpu::drot(queue, n, x, incx, y, incy, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s) { + mkl::gpu::srotg(queue, a, b, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s) { + mkl::gpu::drotg(queue, a, b, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + mkl::gpu::crotg(queue, a, b, c, s); +} + +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + mkl::gpu::zrotg(queue, a, b, c, s); +} + +void rotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m) { + mkl::gpu::srotm(queue, n, x, incx, y, incy, param); +} + +void rotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m) { + mkl::gpu::drotm(queue, n, x, incx, y, incy, param); +} + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m) { + mkl::gpu::srotmg(queue, d1, d2, x1, y1, param); +} + +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, cl::sycl::buffer ¶m) { + mkl::gpu::drotmg(queue, d1, d2, x1, y1, param); +} + +void scal(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx) { + mkl::gpu::sscal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx) { + mkl::gpu::dscal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::cscal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::zscal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::csscal(queue, n, alpha, x, incx); +} + +void scal(cl::sycl::queue &queue, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx) { + mkl::gpu::zdscal(queue, n, alpha, x, incx); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::sswap(queue, n, x, incx, y, incy); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy) { + mkl::gpu::dswap(queue, n, x, incx, y, incy); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::cswap(queue, n, x, incx, y, incy); +} + +void swap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy) { + mkl::gpu::zswap(queue, n, x, incx, y, incy); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::isamax(queue, n, x, incx, result); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::idamax(queue, n, x, incx, result); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::icamax(queue, n, x, incx, result); +} + +void iamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::izamax(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::isamin(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result) { + mkl::gpu::idamin(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::icamin(queue, n, x, incx, result); +} + +void iamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result) { + mkl::gpu::izamin(queue, n, x, incx, result); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, int64_t n, int64_t k, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, int64_t ldb, + float beta, cl::sycl::buffer &c, int64_t ldc) { + mkl::gpu::sgemmt(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(transa), + mkl::cblas_convert(transb), n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, int64_t n, int64_t k, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, int64_t ldb, + double beta, cl::sycl::buffer &c, int64_t ldc) { + mkl::gpu::dgemmt(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(transa), + mkl::cblas_convert(transb), n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::zgemmt(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(transa), + mkl::cblas_convert(transb), n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc) { + mkl::gpu::cgemmt(queue, mkl::cblas_convert(upper_lower), mkl::cblas_convert(transa), + mkl::cblas_convert(transb), n, k, alpha, a, lda, b, ldb, beta, c, ldc); +} + +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, int64_t m, + int64_t n, int64_t k, half alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, half beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::hgemm(queue, mkl::cblas_convert(transa), mkl::cblas_convert(transb), m, n, k, alpha, + a, lda, b, ldb, beta, c, ldc); +} + +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, int64_t m, + int64_t n, int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc) { + mkl::gpu::gemm_f16f16f32(queue, mkl::cblas_convert(transa), mkl::cblas_convert(transb), m, n, k, + alpha, a, lda, b, ldb, beta, c, ldc); +} + +} //namespace internal +} //namespace mklgpu +} //namespace onemkl diff --git a/src/blas/backends/mklgpu/mkl_internal_blas_gpu_wrappers.hpp b/src/blas/backends/mklgpu/mkl_internal_blas_gpu_wrappers.hpp new file mode 100644 index 000000000..33854f577 --- /dev/null +++ b/src/blas/backends/mklgpu/mkl_internal_blas_gpu_wrappers.hpp @@ -0,0 +1,533 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ +#ifndef _MKL_INTERNAL_BLAS_GPU_WRAPPERS_HPP_ +#define _MKL_INTERNAL_BLAS_GPU_WRAPPERS_HPP_ + +#include +#include +#include + +#include "onemkl/types.hpp" + +namespace onemkl { +namespace mklgpu { +namespace internal { +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); +void symm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc); +void hemm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + double beta, cl::sycl::buffer &c, std::int64_t ldc); +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void syrk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void herk(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void syr2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void her2k(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +void trmm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb); +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); +void trsm(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb); +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void gemv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void gbmv(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); +void ger(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void gerc(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void geru(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void hbmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void hbmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +void hemv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void hemv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +void her(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void her(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void her2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void her2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda); +void hpmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +void hpmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); +void hpr(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +void hpr(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); +void hpr2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); +void hpr2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a); +void sbmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); +void sbmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); +void spmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy); +void spmv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy); +void spr(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); +void spr(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a); +void spr2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); +void spr2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a); +void symv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy); +void symv(cl::sycl::queue &queue, onemkl::uplo uplo, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy); +void syr(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); +void syr(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda); +void syr2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); +void syr2(cl::sycl::queue &queue, onemkl::uplo upplo, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda); +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); +void tbmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); +void tbsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx); +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void tpmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx); +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void tpsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void trmv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void trsv(cl::sycl::queue &queue, onemkl::uplo upplo, onemkl::transpose trans, onemkl::diag diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); +void asum(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); +void axpy(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +void axpy(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void axpy(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); +void copy(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); +void sdsdot(cl::sycl::queue &queue, std::int64_t n, float sb, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); +void dot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer &result); +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +void dotc(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +void dotu(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); +void nrm2(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s); +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + double c, double s); +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, float c, float s); +void rot(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, double c, double s); +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); +void rotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +void rotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m); +void rotm(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, cl::sycl::buffer ¶m); +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m); +void rotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, double y1, cl::sycl::buffer ¶m); +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, cl::sycl::buffer &x, + std::int64_t incx); +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, cl::sycl::buffer &x, + std::int64_t incx); +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void scal(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void scal(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void scal(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); +void swap(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy); +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamax(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &result); +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void iamin(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer &result); +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc); +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); +void gemmt(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); +void gemm(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); +void gemm_ext(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc); + +} //namespace internal +} //namespace mklgpu +} //namespace onemkl +#endif //_MKL_INTERNAL_BLAS_GPU_WRAPPERS_HPP_ diff --git a/src/blas/backends/mklgpu/mkl_internal_blas_sycl_gpu.hpp b/src/blas/backends/mklgpu/mkl_internal_blas_sycl_gpu.hpp new file mode 100644 index 000000000..09c05c29e --- /dev/null +++ b/src/blas/backends/mklgpu/mkl_internal_blas_sycl_gpu.hpp @@ -0,0 +1,872 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _MKL_INTERNAL_BLAS_SYCL_GPU_HPP_ +#define _MKL_INTERNAL_BLAS_SYCL_GPU_HPP_ + +#include +#include + +typedef enum { MKL_ROW_MAJOR = 101, MKL_COL_MAJOR = 102 } MKL_LAYOUT; + +typedef enum { MKL_NOTRANS = 111, MKL_TRANS = 112, MKL_CONJTRANS = 113 } MKL_TRANSPOSE; + +typedef enum { MKL_UPPER = 121, MKL_LOWER = 122 } MKL_UPLO; + +typedef enum { MKL_NONUNIT = 131, MKL_UNIT = 132 } MKL_DIAG; + +typedef enum { MKL_LEFT = 141, MKL_RIGHT = 142 } MKL_SIDE; + +typedef enum { + MKL_COMPACT_SSE = 181, + MKL_COMPACT_AVX = 182, + MKL_COMPACT_AVX512 = 183 +} MKL_COMPACT_PACK; + +enum CBLAS_OFFSET { CblasRowOffset = 171, CblasColOffset = 172, CblasFixOffset = 173 }; +typedef enum CBLAS_OFFSET CBLAS_OFFSET; + +namespace mkl { + +inline MKL_TRANSPOSE cblas_convert(onemkl::transpose t) { + if (t == onemkl::transpose::nontrans) + return MKL_NOTRANS; + if (t == onemkl::transpose::trans) + return MKL_TRANS; + if (t == onemkl::transpose::conjtrans) + return MKL_CONJTRANS; + return MKL_NOTRANS; +} + +inline MKL_UPLO cblas_convert(onemkl::uplo u) { + if (u == onemkl::uplo::upper) + return MKL_UPPER; + if (u == onemkl::uplo::lower) + return MKL_LOWER; + return MKL_UPPER; +} + +inline MKL_DIAG cblas_convert(onemkl::diag d) { + if (d == onemkl::diag::nonunit) + return MKL_NONUNIT; + if (d == onemkl::diag::unit) + return MKL_UNIT; + return MKL_NONUNIT; +} + +inline MKL_SIDE cblas_convert(onemkl::side s) { + if (s == onemkl::side::left) + return MKL_LEFT; + if (s == onemkl::side::right) + return MKL_RIGHT; + return MKL_LEFT; +} + +namespace gpu { + +// gemm + +void sgemm(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, int64_t n, + int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, cl::sycl::buffer &c, + int64_t ldc); + +void dgemm(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, int64_t n, + int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc); + +void cgemm(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc); + +void zgemm(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, int64_t n, + int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &b, int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc); + +// symm + +void ssymm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, int64_t m, int64_t n, + float alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb, float beta, cl::sycl::buffer &c, int64_t ldc); + +void dsymm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, int64_t m, int64_t n, + double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, cl::sycl::buffer &c, + int64_t ldc); + +void csymm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void zsymm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +// hemm + +void chemm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void zhemm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +// syrk +void ssyrk(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + float alpha, cl::sycl::buffer &a, int64_t lda, float beta, + cl::sycl::buffer &c, int64_t ldc); + +void dsyrk(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + double alpha, cl::sycl::buffer &a, int64_t lda, double beta, + cl::sycl::buffer &c, int64_t ldc); + +void csyrk(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc); + +void zsyrk(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc); + +// herk + +void cherk(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + float alpha, cl::sycl::buffer, 1> &a, int64_t lda, float beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void zherk(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + double alpha, cl::sycl::buffer, 1> &a, int64_t lda, double beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +// syr2k + +void ssyr2k(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + float alpha, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb, float beta, cl::sycl::buffer &c, int64_t ldc); + +void dsyr2k(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, double beta, + cl::sycl::buffer &c, int64_t ldc); + +void csyr2k(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void zsyr2k(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +// her2k + +void cher2k(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void zher2k(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE trans, int64_t n, int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +// trmm + +void strmm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, float alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb); + +void dtrmm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, double alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb); + +void ctrmm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb); + +void ztrmm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb); + +// trsm +void strsm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, float alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb); + +void dtrsm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, double alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb); + +void ctrsm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb); + +void ztrsm(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_DIAG unit_diag, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb); + +// Level2 + +void sgemv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy); + +void dgemv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx, double beta, cl::sycl::buffer &y, int64_t incy); + +void cgemv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void zgemv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void sgbmv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx, float beta, cl::sycl::buffer &y, + int64_t incy); + +void dgbmv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, double alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &x, int64_t incx, double beta, + cl::sycl::buffer &y, int64_t incy); + +void cgbmv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &x, int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, int64_t incy); + +void zgbmv(cl::sycl::queue &queue, MKL_TRANSPOSE trans, int64_t m, int64_t n, int64_t kl, + int64_t ku, std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, cl::sycl::buffer, 1> &x, int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, int64_t incy); + +void sger(cl::sycl::queue &queue, int64_t m, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &a, + int64_t lda); + +void dger(cl::sycl::queue &queue, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &a, int64_t lda); + +void cgerc(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda); + +void zgerc(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda); + +void cgeru(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda); + +void zgeru(cl::sycl::queue &queue, int64_t m, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda); + +void chbmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void zhbmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void chemv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void zhemv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void cher(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a, int64_t lda); + +void zher(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a, int64_t lda); + +void cher2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda); + +void zher2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a, int64_t lda); + +void chpmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + int64_t incx, std::complex beta, cl::sycl::buffer, 1> &y, + int64_t incy); + +void zhpmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, int64_t incy); + +void chpr(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a); + +void zhpr(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &a); + +void chpr2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a); + +void zhpr2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &a); + +void ssbmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, int64_t k, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy); + +void dsbmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, int64_t k, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx, double beta, cl::sycl::buffer &y, int64_t incy); + +void sspmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx, float beta, + cl::sycl::buffer &y, int64_t incy); + +void dspmv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx, + double beta, cl::sycl::buffer &y, int64_t incy); + +void sspr(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a); + +void dspr(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a); + +void sspr2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &a); + +void dspr2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &a); + +void ssymv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx, + float beta, cl::sycl::buffer &y, int64_t incy); + +void dsymv(cl::sycl::queue &queue, MKL_UPLO uplo, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx, double beta, cl::sycl::buffer &y, int64_t incy); + +void ssyr(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a, int64_t lda); + +void dsyr(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &a, + int64_t lda); + +void ssyr2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, float alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &a, int64_t lda); + +void dsyr2(cl::sycl::queue &queue, MKL_UPLO upplo, int64_t n, double alpha, + cl::sycl::buffer &x, int64_t incx, cl::sycl::buffer &y, + int64_t incy, cl::sycl::buffer &a, int64_t lda); + +void stbmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx); + +void dtbmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx); + +void ctbmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void ztbmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void stbsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx); + +void dtbsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx); + +void ctbsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void ztbsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + int64_t k, cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void stpmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx); + +void dtpmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx); + +void ctpmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + int64_t incx); + +void ztpmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx); + +void stpsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx); + +void dtpsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, cl::sycl::buffer &x, int64_t incx); + +void ctpsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, cl::sycl::buffer, 1> &x, + int64_t incx); + +void ztpsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, int64_t incx); + +void strmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx); + +void dtrmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx); + +void ctrmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void ztrmv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void strsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, int64_t incx); + +void dtrsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &x, + int64_t incx); + +void ctrsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +void ztrsv(cl::sycl::queue &queue, MKL_UPLO upplo, MKL_TRANSPOSE trans, MKL_DIAG diag, int64_t n, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &x, int64_t incx); + +// Level1 + +void scasum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void dzasum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void sasum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void dasum(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void saxpy(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy); + +void daxpy(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy); + +void caxpy(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy); + +void zaxpy(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx, + cl::sycl::buffer, 1> &y, int64_t incy); + +void scopy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy); + +void dcopy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy); + +void ccopy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy); + +void zcopy(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy); + +void sdot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result); + +void ddot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result); + +void sdsdot(cl::sycl::queue &queue, int64_t n, float sb, cl::sycl::buffer &x, + int64_t incx, cl::sycl::buffer &y, int64_t incy, + cl::sycl::buffer &result); + +void dsdot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer &result); + +void cdotc(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result); + +void zdotc(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result); + +void cdotu(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result); + +void zdotu(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, + cl::sycl::buffer, 1> &result); + +void scnrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void dznrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void snrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void dnrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void csrot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, float c, + float s); + +void zdrot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, double c, + double s); + +void srot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, float c, float s); + +void drot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, double c, double s); + +void srotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void drotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void crotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void zrotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void srotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m); + +void drotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m); + +void srotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m); + +void drotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m); + +void sscal(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx); + +void dscal(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx); + +void cscal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void zscal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void csscal(cl::sycl::queue &queue, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void zdscal(cl::sycl::queue &queue, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void sswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy); + +void dswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy); + +void cswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy); + +void zswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy); + +void isamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void idamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void icamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void dnrm2(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void csrot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, float c, + float s); + +void zdrot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy, double c, + double s); + +void srot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, float c, float s); + +void drot(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, double c, double s); + +void srotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void drotg(cl::sycl::queue &queue, cl::sycl::buffer &a, cl::sycl::buffer &b, + cl::sycl::buffer &c, cl::sycl::buffer &s); + +void crotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void zrotg(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + +void srotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m); + +void drotm(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy, cl::sycl::buffer ¶m); + +void srotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, cl::sycl::buffer &d2, + cl::sycl::buffer &x1, float y1, cl::sycl::buffer ¶m); + +void drotmg(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m); + +void sscal(cl::sycl::queue &queue, int64_t n, float alpha, cl::sycl::buffer &x, + int64_t incx); + +void dscal(cl::sycl::queue &queue, int64_t n, double alpha, cl::sycl::buffer &x, + int64_t incx); + +void cscal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void zscal(cl::sycl::queue &queue, int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void csscal(cl::sycl::queue &queue, int64_t n, float alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void zdscal(cl::sycl::queue &queue, int64_t n, double alpha, + cl::sycl::buffer, 1> &x, int64_t incx); + +void sswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy); + +void dswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &y, int64_t incy); + +void cswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy); + +void zswap(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer, 1> &y, int64_t incy); + +void isamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void idamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void icamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void izamax(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void isamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void idamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer &x, int64_t incx, + cl::sycl::buffer &result); + +void icamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +void izamin(cl::sycl::queue &queue, int64_t n, cl::sycl::buffer, 1> &x, + int64_t incx, cl::sycl::buffer &result); + +// batch api + +void sgemm_batch(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, + int64_t n, int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + int64_t stride_a, cl::sycl::buffer &b, int64_t ldb, int64_t stride_b, + float beta, cl::sycl::buffer &c, int64_t ldc, int64_t stride_c, + int64_t batch_size, int64_t offset_a = 0, int64_t offset_b = 0, + int64_t offset_c = 0); + +void dgemm_batch(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, + int64_t n, int64_t k, double alpha, cl::sycl::buffer &a, int64_t lda, + int64_t stride_a, cl::sycl::buffer &b, int64_t ldb, int64_t stride_b, + double beta, cl::sycl::buffer &c, int64_t ldc, int64_t stride_c, + int64_t batch_size, int64_t offset_a = 0, int64_t offset_b = 0, + int64_t offset_c = 0); + +void cgemm_batch(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, + int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, int64_t stride_a, + cl::sycl::buffer, 1> &b, int64_t ldb, int64_t stride_b, + std::complex beta, cl::sycl::buffer, 1> &c, int64_t ldc, + int64_t stride_c, int64_t batch_size, int64_t offset_a = 0, int64_t offset_b = 0, + int64_t offset_c = 0); + +void zgemm_batch(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, + int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, int64_t stride_a, + cl::sycl::buffer, 1> &b, int64_t ldb, int64_t stride_b, + std::complex beta, cl::sycl::buffer, 1> &c, + int64_t ldc, int64_t stride_c, int64_t batch_size, int64_t offset_a = 0, + int64_t offset_b = 0, int64_t offset_c = 0); + +void strsm_batch(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, + MKL_TRANSPOSE trans, MKL_DIAG unit_diag, int64_t m, int64_t n, float alpha, + cl::sycl::buffer &a, int64_t lda, int64_t stride_a, + cl::sycl::buffer &b, int64_t ldb, int64_t stride_b, int64_t batch_size, + int64_t offset_a = 0, int64_t offset_b = 0); + +void dtrsm_batch(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, + MKL_TRANSPOSE trans, MKL_DIAG unit_diag, int64_t m, int64_t n, double alpha, + cl::sycl::buffer &a, int64_t lda, int64_t stride_a, + cl::sycl::buffer &b, int64_t ldb, int64_t stride_b, int64_t batch_size, + int64_t offset_a = 0, int64_t offset_b = 0); + +void ctrsm_batch(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, + MKL_TRANSPOSE trans, MKL_DIAG unit_diag, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, int64_t stride_a, cl::sycl::buffer, 1> &b, + int64_t ldb, int64_t stride_b, int64_t batch_size, int64_t offset_a = 0, + int64_t offset_b = 0); + +void ztrsm_batch(cl::sycl::queue &queue, MKL_SIDE left_right, MKL_UPLO upper_lower, + MKL_TRANSPOSE trans, MKL_DIAG unit_diag, int64_t m, int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + int64_t lda, int64_t stride_a, cl::sycl::buffer, 1> &b, + int64_t ldb, int64_t stride_b, int64_t batch_size, int64_t offset_a = 0, + int64_t offset_b = 0); + +// BLAS like extension + +void sgemmt(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_TRANSPOSE transb, int64_t n, int64_t k, float alpha, cl::sycl::buffer &a, + int64_t lda, cl::sycl::buffer &b, int64_t ldb, float beta, + cl::sycl::buffer &c, int64_t ldc); + +void dgemmt(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_TRANSPOSE transb, int64_t n, int64_t k, double alpha, + cl::sycl::buffer &a, int64_t lda, cl::sycl::buffer &b, + int64_t ldb, double beta, cl::sycl::buffer &c, int64_t ldc); + +void zgemmt(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_TRANSPOSE transb, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void cgemmt(cl::sycl::queue &queue, MKL_UPLO upper_lower, MKL_TRANSPOSE transa, + MKL_TRANSPOSE transb, int64_t n, int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, int64_t lda, + cl::sycl::buffer, 1> &b, int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, int64_t ldc); + +void hgemm(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, int64_t n, + int64_t k, half alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, half beta, cl::sycl::buffer &c, + int64_t ldc); + +void gemm_f16f16f32(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, int64_t m, + int64_t n, int64_t k, float alpha, cl::sycl::buffer &a, int64_t lda, + cl::sycl::buffer &b, int64_t ldb, float beta, + cl::sycl::buffer &c, int64_t ldc); + +void gemm_s8u8s32(cl::sycl::queue &queue, MKL_TRANSPOSE transa, MKL_TRANSPOSE transb, + CBLAS_OFFSET offsetc, int64_t m, int64_t n, int64_t k, float alpha, + cl::sycl::buffer &a, int64_t lda, int8_t ao, + cl::sycl::buffer &b, int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, int64_t ldc, cl::sycl::buffer &co); +} // namespace gpu +} // namespace mkl +#endif //_MKL_INTERNAL_BLAS_SYCL_GPU_HPP_ diff --git a/src/blas/blas_loader.cpp b/src/blas/blas_loader.cpp new file mode 100644 index 000000000..f98329194 --- /dev/null +++ b/src/blas/blas_loader.cpp @@ -0,0 +1,1319 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include "loader.hpp" + +namespace onemkl { +namespace blas { +namespace detail { + +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].scasum_sycl(queue, n, x, incx, result); +} + +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].dzasum_sycl(queue, n, x, incx, result); +} + +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].sasum_sycl(queue, n, x, incx, result); +} + +void asum(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].dasum_sycl(queue, n, x, incx, result); +} + +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy) { + function_tables[libname].saxpy_sycl(queue, n, alpha, x, incx, y, incy); +} + +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy) { + function_tables[libname].daxpy_sycl(queue, n, alpha, x, incx, y, incy); +} + +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].caxpy_sycl(queue, n, alpha, x, incx, y, incy); +} + +void axpy(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].zaxpy_sycl(queue, n, alpha, x, incx, y, incy); +} + +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].scopy_sycl(queue, n, x, incx, y, incy); +} + +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dcopy_sycl(queue, n, x, incx, y, incy); +} + +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].ccopy_sycl(queue, n, x, incx, y, incy); +} + +void copy(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].zcopy_sycl(queue, n, x, incx, y, incy); +} + +void dot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + function_tables[libname].sdot_sycl(queue, n, x, incx, y, incy, result); +} + +void dot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + function_tables[libname].ddot_sycl(queue, n, x, incx, y, incy, result); +} + +void dot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result) { + function_tables[libname].dsdot_sycl(queue, n, x, incx, y, incy, result); +} + +void dotc(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + function_tables[libname].cdotc_sycl(queue, n, x, incx, y, incy, result); +} + +void dotc(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + function_tables[libname].zdotc_sycl(queue, n, x, incx, y, incy, result); +} + +void dotu(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + function_tables[libname].cdotu_sycl(queue, n, x, incx, y, incy, result); +} + +void dotu(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result) { + function_tables[libname].zdotu_sycl(queue, n, x, incx, y, incy, result); +} + +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].isamin_sycl(queue, n, x, incx, result); +} + +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].idamin_sycl(queue, n, x, incx, result); +} + +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].icamin_sycl(queue, n, x, incx, result); +} + +void iamin(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].izamin_sycl(queue, n, x, incx, result); +} + +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].isamax_sycl(queue, n, x, incx, result); +} + +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].idamax_sycl(queue, n, x, incx, result); +} + +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].icamax_sycl(queue, n, x, incx, result); +} + +void iamax(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].izamax_sycl(queue, n, x, incx, result); +} + +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].snrm2_sycl(queue, n, x, incx, result); +} + +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result) { + function_tables[libname].dnrm2_sycl(queue, n, x, incx, result); +} + +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].scnrm2_sycl(queue, n, x, incx, result); +} + +void nrm2(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result) { + function_tables[libname].dznrm2_sycl(queue, n, x, incx, result); +} + +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, float c, float s) { + function_tables[libname].srot_sycl(queue, n, x, incx, y, incy, c, s); +} + +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, double c, double s) { + function_tables[libname].drot_sycl(queue, n, x, incx, y, incy, c, s); +} + +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, float c, float s) { + function_tables[libname].csrot_sycl(queue, n, x, incx, y, incy, c, s); +} + +void rot(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, double c, double s) { + function_tables[libname].zdrot_sycl(queue, n, x, incx, y, incy, c, s); +} + +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { + function_tables[libname].srotg_sycl(queue, a, b, c, s); +} + +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s) { + function_tables[libname].drotg_sycl(queue, a, b, c, s); +} + +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + function_tables[libname].crotg_sycl(queue, a, b, c, s); +} + +void rotg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s) { + function_tables[libname].zrotg_sycl(queue, a, b, c, s); +} + +void rotm(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + function_tables[libname].srotm_sycl(queue, n, x, incx, y, incy, param); +} + +void rotm(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m) { + function_tables[libname].drotm_sycl(queue, n, x, incx, y, incy, param); +} + +void rotmg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m) { + function_tables[libname].srotmg_sycl(queue, d1, d2, x1, y1, param); +} + +void rotmg(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m) { + function_tables[libname].drotmg_sycl(queue, d1, d2, x1, y1, param); +} + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].sscal_sycl(queue, n, alpha, x, incx); +} + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].dscal_sycl(queue, n, alpha, x, incx); +} + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].cscal_sycl(queue, n, alpha, x, incx); +} + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].csscal_sycl(queue, n, alpha, x, incx); +} + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].zscal_sycl(queue, n, alpha, x, incx); +} + +void scal(char *libname, cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].zdscal_sycl(queue, n, alpha, x, incx); +} + +void sdsdot(char *libname, cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &result) { + function_tables[libname].sdsdot_sycl(queue, n, sb, x, incx, y, incy, result); +} + +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].sswap_sycl(queue, n, x, incx, y, incy); +} + +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dswap_sycl(queue, n, x, incx, y, incy); +} + +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].cswap_sycl(queue, n, x, incx, y, incy); +} + +void swap(char *libname, cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].zswap_sycl(queue, n, x, incx, y, incy); +} + +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].sgbmv_sycl(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dgbmv_sycl(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].cgbmv_sycl(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gbmv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + function_tables[libname].zgbmv_sycl(queue, trans, m, n, kl, ku, alpha, a, lda, x, incx, beta, y, + incy); +} + +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].sgemv_sycl(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dgemv_sycl(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].cgemv_sycl(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void gemv(char *libname, cl::sycl::queue &queue, transpose trans, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + function_tables[libname].zgemv_sycl(queue, trans, m, n, alpha, a, lda, x, incx, beta, y, incy); +} + +void ger(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + function_tables[libname].sger_sycl(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void ger(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + function_tables[libname].dger_sycl(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void gerc(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].cgerc_sycl(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void gerc(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].zgerc_sycl(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void geru(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].cgeru_sycl(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void geru(char *libname, cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].zgeru_sycl(queue, m, n, alpha, x, incx, y, incy, a, lda); +} + +void hbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].chbmv_sycl(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); +} + +void hbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + function_tables[libname].zhbmv_sycl(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); +} + +void hemv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].chemv_sycl(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); +} + +void hemv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + function_tables[libname].zhemv_sycl(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); +} + +void her(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].cher_sycl(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void her(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].zher_sycl(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void her2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].cher2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void her2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a, std::int64_t lda) { + function_tables[libname].zher2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void hpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy) { + function_tables[libname].chpmv_sycl(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void hpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy) { + function_tables[libname].zhpmv_sycl(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void hpr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { + function_tables[libname].chpr_sycl(queue, upper_lower, n, alpha, x, incx, a); +} + +void hpr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a) { + function_tables[libname].zhpr_sycl(queue, upper_lower, n, alpha, x, incx, a); +} + +void hpr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { + function_tables[libname].chpr2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void hpr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &a) { + function_tables[libname].zhpr2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void sbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].ssbmv_sycl(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); +} + +void sbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dsbmv_sycl(queue, upper_lower, n, k, alpha, a, lda, x, incx, beta, y, + incy); +} + +void spmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + float beta, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].sspmv_sycl(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void spmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dspmv_sycl(queue, upper_lower, n, alpha, a, x, incx, beta, y, incy); +} + +void spr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a) { + function_tables[libname].sspr_sycl(queue, upper_lower, n, alpha, x, incx, a); +} + +void spr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a) { + function_tables[libname].dspr_sycl(queue, upper_lower, n, alpha, x, incx, a); +} + +void spr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a) { + function_tables[libname].sspr2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void spr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a) { + function_tables[libname].dspr2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a); +} + +void symv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].ssymv_sycl(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); +} + +void symv(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, std::int64_t incy) { + function_tables[libname].dsymv_sycl(queue, upper_lower, n, alpha, a, lda, x, incx, beta, y, + incy); +} + +void syr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda) { + function_tables[libname].ssyr_sycl(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void syr(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &a, + std::int64_t lda) { + function_tables[libname].dsyr_sycl(queue, upper_lower, n, alpha, x, incx, a, lda); +} + +void syr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + function_tables[libname].ssyr2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void syr2(char *libname, cl::sycl::queue &queue, uplo upper_lower, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, cl::sycl::buffer &y, + std::int64_t incy, cl::sycl::buffer &a, std::int64_t lda) { + function_tables[libname].dsyr2_sycl(queue, upper_lower, n, alpha, x, incx, y, incy, a, lda); +} + +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].stbmv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].dtbmv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ctbmv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ztbmv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].stbsv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].dtbsv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ctbsv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tbsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, std::int64_t k, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ztbsv_sycl(queue, upper_lower, trans, unit_diag, n, k, a, lda, x, + incx); +} + +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx) { + function_tables[libname].stpmv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx) { + function_tables[libname].dtpmv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ctpmv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ztpmv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx) { + function_tables[libname].stpsv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx) { + function_tables[libname].dtpsv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ctpsv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void tpsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ztpsv_sycl(queue, upper_lower, trans, unit_diag, n, a, x, incx); +} + +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].strmv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].dtrmv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ctrmv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trmv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ztrmv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].strsv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx) { + function_tables[libname].dtrsv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ctrsv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void trsv(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, diag unit_diag, + std::int64_t n, cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx) { + function_tables[libname].ztrsv_sycl(queue, upper_lower, trans, unit_diag, n, a, lda, x, incx); +} + +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].sgemm_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); +} + +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].dgemm_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); +} + +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].cgemm_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); +} + +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].zgemm_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); +} + +void gemm(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, std::int64_t m, + std::int64_t n, std::int64_t k, half alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].hgemm_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, beta, + c, ldc); +} + +void hemm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].chemm_sycl(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void hemm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zhemm_sycl(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void herk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, float beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].cherk_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); +} + +void herk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, double beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zherk_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); +} + +void her2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + float beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].cher2k_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void her2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + double beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].zher2k_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].ssymm_sycl(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].dsymm_sycl(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].csymm_sycl(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void symm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, std::int64_t m, + std::int64_t n, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zsymm_sycl(queue, left_right, upper_lower, m, n, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].ssyrk_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); +} + +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + double beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].dsyrk_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); +} + +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].csyrk_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); +} + +void syrk(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zsyrk_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, beta, c, + ldc); +} + +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].ssyr2k_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].dsyr2k_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].csyr2k_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void syr2k(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose trans, std::int64_t n, + std::int64_t k, std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zsyr2k_sycl(queue, upper_lower, trans, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb) { + function_tables[libname].strmm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb) { + function_tables[libname].dtrmm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + function_tables[libname].ctrmm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trmm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + function_tables[libname].ztrmm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb) { + function_tables[libname].strsm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb) { + function_tables[libname].dtrsm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + function_tables[libname].ctrsm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void trsm(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, transpose trans, + diag unit_diag, std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb) { + function_tables[libname].ztrsm_sycl(queue, left_right, upper_lower, trans, unit_diag, m, n, + alpha, a, lda, b, ldb); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].sgemm_batch_group_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + b, ldb, beta, c, ldc, group_count, group_size); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].dgemm_batch_group_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + b, ldb, beta, c, ldc, group_count, group_size); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].cgemm_batch_group_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + b, ldb, beta, c, ldc, group_count, group_size); +} + +void gemm_batch( + char *libname, cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].zgemm_batch_group_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + b, ldb, beta, c, ldc, group_count, group_size); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, std::int64_t stride_c, + std::int64_t batch_size) { + function_tables[libname].sgemm_batch_strided_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + stride_a, b, ldb, stride_b, beta, c, ldc, + stride_c, batch_size); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + double beta, cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { + function_tables[libname].dgemm_batch_strided_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + stride_a, b, ldb, stride_b, beta, c, ldc, + stride_c, batch_size); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { + function_tables[libname].cgemm_batch_strided_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + stride_a, b, ldb, stride_b, beta, c, ldc, + stride_c, batch_size); +} + +void gemm_batch(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size) { + function_tables[libname].zgemm_batch_strided_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + stride_a, b, ldb, stride_b, beta, c, ldc, + stride_c, batch_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].strsm_batch_group_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, b, ldb, + group_count, group_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &alpha, + cl::sycl::buffer &a, cl::sycl::buffer &lda, + cl::sycl::buffer &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].dtrsm_batch_group_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, b, ldb, + group_count, group_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size) { + function_tables[libname].ctrsm_batch_group_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, b, ldb, + group_count, group_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, cl::sycl::buffer &trans, + cl::sycl::buffer &unit_diag, cl::sycl::buffer &m, + cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, + cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size) { + function_tables[libname].ztrsm_batch_group_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, b, ldb, + group_count, group_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size) { + function_tables[libname].strsm_batch_strided_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, stride_a, b, + ldb, stride_b, batch_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &a, std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size) { + function_tables[libname].dtrsm_batch_strided_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, stride_a, b, + ldb, stride_b, batch_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { + function_tables[libname].ctrsm_batch_strided_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, stride_a, b, + ldb, stride_b, batch_size); +} + +void trsm_batch(char *libname, cl::sycl::queue &queue, side left_right, uplo upper_lower, + transpose trans, diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size) { + function_tables[libname].ztrsm_batch_strided_sycl(queue, left_right, upper_lower, trans, + unit_diag, m, n, alpha, a, lda, stride_a, b, + ldb, stride_b, batch_size); +} + +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].sgemmt_sycl(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, + ldb, beta, c, ldc); +} + +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].dgemmt_sycl(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, + ldb, beta, c, ldc); +} + +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc) { + function_tables[libname].cgemmt_sycl(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, + ldb, beta, c, ldc); +} + +void gemmt(char *libname, cl::sycl::queue &queue, uplo upper_lower, transpose transa, + transpose transb, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zgemmt_sycl(queue, upper_lower, transa, transb, n, k, alpha, a, lda, b, + ldb, beta, c, ldc); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].gemm_f16f16f32_ext_sycl(queue, transa, transb, m, n, k, alpha, a, lda, + b, ldb, beta, c, ldc); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + offset offsetc, std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, int8_t ao, + cl::sycl::buffer &b, std::int64_t ldb, uint8_t bo, float beta, + cl::sycl::buffer &c, std::int64_t ldc, cl::sycl::buffer &co) { + function_tables[libname].gemm_s8u8s32_ext_sycl(queue, transa, transb, offsetc, m, n, k, alpha, + a, lda, ao, b, ldb, bo, beta, c, ldc, co); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].sgemm_ext_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, double beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].dgemm_ext_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].cgemm_ext_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc) { + function_tables[libname].zgemm_ext_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +void gemm_ext(char *libname, cl::sycl::queue &queue, transpose transa, transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, half beta, cl::sycl::buffer &c, std::int64_t ldc) { + function_tables[libname].hgemm_ext_sycl(queue, transa, transb, m, n, k, alpha, a, lda, b, ldb, + beta, c, ldc); +} + +} /*namespace detail */ +} /* namespace blas */ +} /* namespace onemkl */ diff --git a/src/blas/function_table.hpp b/src/blas/function_table.hpp new file mode 100644 index 000000000..69d9128b7 --- /dev/null +++ b/src/blas/function_table.hpp @@ -0,0 +1,806 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _BLAS_FUNCTION_TABLE_HPP_ +#define _BLAS_FUNCTION_TABLE_HPP_ + +#include +#include +#include +#include "onemkl/types.hpp" + +typedef struct { + int version; + void (*scasum_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*dzasum_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*sasum_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*dasum_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*saxpy_sycl)(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + void (*daxpy_sycl)(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy); + void (*caxpy_sycl)(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*zaxpy_sycl)(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*scopy_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + void (*dcopy_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + void (*ccopy_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*zcopy_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*sdot_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + void (*ddot_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + void (*dsdot_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + void (*cdotc_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + void (*zdotc_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + void (*cdotu_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + void (*zdotu_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, + cl::sycl::buffer, 1> &result); + void (*isamin_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*idamin_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*icamin_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*izamin_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*isamax_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*idamax_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*icamax_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*izamax_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*snrm2_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*dnrm2_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer &result); + void (*scnrm2_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*dznrm2_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &result); + void (*srot_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, float c, + float s); + void (*drot_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy, double c, + double s); + void (*csrot_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, float c, + float s); + void (*zdrot_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + double c, double s); + void (*srotg_sycl)(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); + void (*drotg_sycl)(cl::sycl::queue &queue, cl::sycl::buffer &a, + cl::sycl::buffer &b, cl::sycl::buffer &c, + cl::sycl::buffer &s); + void (*crotg_sycl)(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + void (*zrotg_sycl)(cl::sycl::queue &queue, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &c, + cl::sycl::buffer, 1> &s); + void (*srotm_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); + void (*drotm_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer ¶m); + void (*srotmg_sycl)(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, float y1, + cl::sycl::buffer ¶m); + void (*drotmg_sycl)(cl::sycl::queue &queue, cl::sycl::buffer &d1, + cl::sycl::buffer &d2, cl::sycl::buffer &x1, double y1, + cl::sycl::buffer ¶m); + void (*sscal_sycl)(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx); + void (*dscal_sycl)(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx); + void (*cscal_sycl)(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*csscal_sycl)(cl::sycl::queue &queue, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*zscal_sycl)(cl::sycl::queue &queue, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*zdscal_sycl)(cl::sycl::queue &queue, std::int64_t n, double alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*sdsdot_sycl)(cl::sycl::queue &queue, std::int64_t n, float sb, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &result); + void (*sswap_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + void (*dswap_sycl)(cl::sycl::queue &queue, std::int64_t n, cl::sycl::buffer &x, + std::int64_t incx, cl::sycl::buffer &y, std::int64_t incy); + void (*cswap_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*zswap_sycl)(cl::sycl::queue &queue, std::int64_t n, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*sgbmv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + void (*dgbmv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + void (*cgbmv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*zgbmv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, std::int64_t kl, std::int64_t ku, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*sgemv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + void (*dgemv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy); + void (*cgemv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*zgemv_sycl)(cl::sycl::queue &queue, onemkl::transpose trans, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*sger_sycl)(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); + void (*dger_sycl)(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, double alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); + void (*cgerc_sycl)(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); + void (*zgerc_sycl)(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); + void (*cgeru_sycl)(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); + void (*zgeru_sycl)(cl::sycl::queue &queue, std::int64_t m, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); + void (*chbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*zhbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*chemv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*zhemv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &x, + std::int64_t incx, std::complex beta, + cl::sycl::buffer, 1> &y, std::int64_t incy); + void (*cher_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + void (*zher_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a, std::int64_t lda); + void (*cher2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); + void (*zher2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a, + std::int64_t lda); + void (*chpmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*zhpmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx, + std::complex beta, cl::sycl::buffer, 1> &y, + std::int64_t incy); + void (*chpr_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + void (*zhpr_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer, 1> &x, std::int64_t incx, + cl::sycl::buffer, 1> &a); + void (*chpr2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a); + void (*zhpr2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::complex alpha, cl::sycl::buffer, 1> &x, + std::int64_t incx, cl::sycl::buffer, 1> &y, + std::int64_t incy, cl::sycl::buffer, 1> &a); + void (*ssbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + void (*dsbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx, + double beta, cl::sycl::buffer &y, std::int64_t incy); + void (*sspmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, float beta, cl::sycl::buffer &y, + std::int64_t incy); + void (*dspmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &a, cl::sycl::buffer &x, + std::int64_t incx, double beta, cl::sycl::buffer &y, + std::int64_t incy); + void (*sspr_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a); + void (*dspr_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a); + void (*sspr2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a); + void (*dspr2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a); + void (*ssymv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, float beta, + cl::sycl::buffer &y, std::int64_t incy); + void (*dsymv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx, double beta, + cl::sycl::buffer &y, std::int64_t incy); + void (*ssyr_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, float alpha, + cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda); + void (*dsyr_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &a, std::int64_t lda); + void (*ssyr2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + float alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); + void (*dsyr2_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, std::int64_t n, + double alpha, cl::sycl::buffer &x, std::int64_t incx, + cl::sycl::buffer &y, std::int64_t incy, + cl::sycl::buffer &a, std::int64_t lda); + void (*stbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + void (*dtbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + void (*ctbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*ztbmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*stbsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + void (*dtbsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &x, std::int64_t incx); + void (*ctbsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*ztbsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, std::int64_t k, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*stpmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + void (*dtpmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + void (*ctpmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*ztpmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*stpsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + void (*dtpsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + cl::sycl::buffer &x, std::int64_t incx); + void (*ctpsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*ztpsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*strmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + void (*dtrmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + void (*ctrmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*ztrmv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*strsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + void (*dtrsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &x, std::int64_t incx); + void (*ctrsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*ztrsv_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t n, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &x, std::int64_t incx); + void (*sgemm_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*dgemm_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*cgemm_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*zgemm_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*hgemm_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, onemkl::transpose transb, + std::int64_t m, std::int64_t n, std::int64_t k, half alpha, + cl::sycl::buffer &a, std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, half beta, cl::sycl::buffer &c, std::int64_t ldc); + void (*chemm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*zhemm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*cherk_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*zherk_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*cher2k_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, float beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*zher2k_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, double beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*ssymm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + std::int64_t m, std::int64_t n, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); + void (*dsymm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + std::int64_t m, std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + double beta, cl::sycl::buffer &c, std::int64_t ldc); + void (*csymm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*zsymm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + std::int64_t m, std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*ssyrk_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, float beta, cl::sycl::buffer &c, + std::int64_t ldc); + void (*dsyrk_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, double beta, cl::sycl::buffer &c, + std::int64_t ldc); + void (*csyrk_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*zsyrk_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*ssyr2k_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb, + float beta, cl::sycl::buffer &c, std::int64_t ldc); + void (*dsyr2k_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*csyr2k_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*zsyr2k_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose trans, + std::int64_t n, std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::complex beta, cl::sycl::buffer, 1> &c, + std::int64_t ldc); + void (*strmm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); + void (*dtrmm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb); + void (*ctrmm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + void (*ztrmm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + void (*strsm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb); + void (*dtrsm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, double alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, std::int64_t ldb); + void (*ctrsm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + void (*ztrsm_sycl)(cl::sycl::queue &queue, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose trans, onemkl::diag unit_diag, std::int64_t m, + std::int64_t n, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + cl::sycl::buffer, 1> &b, std::int64_t ldb); + void (*sgemm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + void (*dgemm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, cl::sycl::buffer &beta, + cl::sycl::buffer &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + void (*cgemm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, cl::sycl::buffer, 1> &c, + cl::sycl::buffer &ldc, std::int64_t group_count, + cl::sycl::buffer &group_size); + void (*zgemm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &transa, + cl::sycl::buffer &transb, cl::sycl::buffer &m, + cl::sycl::buffer &n, cl::sycl::buffer &k, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + cl::sycl::buffer, 1> &beta, + cl::sycl::buffer, 1> &c, cl::sycl::buffer &ldc, + std::int64_t group_count, cl::sycl::buffer &group_size); + void (*sgemm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, + std::int64_t stride_b, float beta, + cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + void (*dgemm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, double alpha, cl::sycl::buffer &a, + std::int64_t lda, std::int64_t stride_a, + cl::sycl::buffer &b, std::int64_t ldb, + std::int64_t stride_b, double beta, + cl::sycl::buffer &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + void (*cgemm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + void (*zgemm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc, + std::int64_t stride_c, std::int64_t batch_size); + void (*strsm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, cl::sycl::buffer &unit_diag, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size); + void (*dtrsm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, cl::sycl::buffer &unit_diag, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer &alpha, cl::sycl::buffer &a, + cl::sycl::buffer &lda, cl::sycl::buffer &b, + cl::sycl::buffer &ldb, std::int64_t group_count, + cl::sycl::buffer &group_size); + void (*ctrsm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, cl::sycl::buffer &unit_diag, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + void (*ztrsm_batch_group_sycl)( + cl::sycl::queue &queue, cl::sycl::buffer &left_right, + cl::sycl::buffer &upper_lower, + cl::sycl::buffer &trans, cl::sycl::buffer &unit_diag, + cl::sycl::buffer &m, cl::sycl::buffer &n, + cl::sycl::buffer, 1> &alpha, + cl::sycl::buffer, 1> &a, cl::sycl::buffer &lda, + cl::sycl::buffer, 1> &b, cl::sycl::buffer &ldb, + std::int64_t group_count, cl::sycl::buffer &group_size); + void (*strsm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::side left_right, + onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); + void (*dtrsm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::side left_right, + onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + std::int64_t stride_a, cl::sycl::buffer &b, + std::int64_t ldb, std::int64_t stride_b, + std::int64_t batch_size); + void (*ctrsm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::side left_right, + onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); + void (*ztrsm_batch_strided_sycl)(cl::sycl::queue &queue, onemkl::side left_right, + onemkl::uplo upper_lower, onemkl::transpose trans, + onemkl::diag unit_diag, std::int64_t m, std::int64_t n, + std::complex alpha, + cl::sycl::buffer, 1> &a, std::int64_t lda, + std::int64_t stride_a, + cl::sycl::buffer, 1> &b, std::int64_t ldb, + std::int64_t stride_b, std::int64_t batch_size); + void (*sgemmt_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*dgemmt_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, double alpha, + cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*cgemmt_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*zgemmt_sycl)(cl::sycl::queue &queue, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*gemm_f16f16f32_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, + std::int64_t k, float alpha, cl::sycl::buffer &a, + std::int64_t lda, cl::sycl::buffer &b, + std::int64_t ldb, float beta, cl::sycl::buffer &c, + std::int64_t ldc); + void (*gemm_s8u8s32_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, onemkl::offset offsetc, std::int64_t m, + std::int64_t n, std::int64_t k, float alpha, + cl::sycl::buffer &a, std::int64_t lda, int8_t ao, + cl::sycl::buffer &b, std::int64_t ldb, uint8_t bo, + float beta, cl::sycl::buffer &c, std::int64_t ldc, + cl::sycl::buffer &co); + void (*sgemm_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, + float alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, float beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*dgemm_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, + double alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, double beta, + cl::sycl::buffer &c, std::int64_t ldc); + void (*cgemm_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*zgemm_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, + std::complex alpha, cl::sycl::buffer, 1> &a, + std::int64_t lda, cl::sycl::buffer, 1> &b, + std::int64_t ldb, std::complex beta, + cl::sycl::buffer, 1> &c, std::int64_t ldc); + void (*hgemm_ext_sycl)(cl::sycl::queue &queue, onemkl::transpose transa, + onemkl::transpose transb, std::int64_t m, std::int64_t n, std::int64_t k, + half alpha, cl::sycl::buffer &a, std::int64_t lda, + cl::sycl::buffer &b, std::int64_t ldb, half beta, + cl::sycl::buffer &c, std::int64_t ldc); +} function_table_t; + +#endif //_BLAS_FUNCTION_TABLE_HPP_ diff --git a/src/blas/loader.hpp b/src/blas/loader.hpp new file mode 100644 index 000000000..352c29789 --- /dev/null +++ b/src/blas/loader.hpp @@ -0,0 +1,91 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _LOADER_HPP_ +#define _LOADER_HPP_ + +#include +#include +#include "blas/function_table.hpp" + +#define SPEC_VERSION 1 + +#ifdef __linux__ + #include + #define LIB_TYPE void * + #define GET_LIB_HANDLE(libname) dlopen((libname), RTLD_LAZY | RTLD_GLOBAL) + #define GET_FUNC(lib, fn) dlsym(lib, (fn)) + #define FREE_LIB_HANDLE(libname) dlclose(libname) + #define ERROR_MSG dlerror() +#endif + +namespace onemkl { +namespace blas { +namespace detail { + +class table_initializer { + struct handle_deleter { + using pointer = LIB_TYPE; + void operator()(pointer p) const { + ::FREE_LIB_HANDLE(p); + } + }; + using dlhandle = std::unique_ptr; + +public: + function_table_t &operator[](const char *libname) { + auto lib = tables.find(libname); + if (lib != tables.end()) + return lib->second; + return add_table(libname); + } + +private: + function_table_t &add_table(const char *libname) { + auto handle = dlhandle{ ::GET_LIB_HANDLE(libname) }; + if (!handle) { + std::cerr << ERROR_MSG << '\n'; + throw std::runtime_error{ "Couldn't load selected backend" }; + } + + auto t = reinterpret_cast(::GET_FUNC(handle.get(), "mkl_blas_table")); + + if (!t) { + std::cerr << ERROR_MSG << '\n'; + throw std::runtime_error{ "Couldn't load functions from selected backend" }; + } + if (t->version != SPEC_VERSION) + throw std::runtime_error{ "Loaded oneMKL specification version mismatch" }; + + handles[libname] = std::move(handle); + tables[libname] = *t; + return *t; + } + + std::map tables; + std::map handles; +}; + +static table_initializer function_tables; + +} //namespace detail +} // namespace blas +} // namespace onemkl + +#endif //_LOADER_HPP_ diff --git a/src/config.hpp.in b/src/config.hpp.in new file mode 100644 index 000000000..589b5c95d --- /dev/null +++ b/src/config.hpp.in @@ -0,0 +1,27 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef ONEMKL_CONFIG_H +#define ONEMKL_CONFIG_H + +#cmakedefine ENABLE_MKLCPU_BACKEND +#cmakedefine ENABLE_MKLGPU_BACKEND +#cmakedefine BUILD_SHARED_LIBS + +#endif diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt new file mode 100644 index 000000000..ebaa90497 --- /dev/null +++ b/tests/CMakeLists.txt @@ -0,0 +1,24 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# Build GoogleTest first +add_subdirectory(${PROJECT_SOURCE_DIR}/deps/googletest ${CMAKE_BINARY_DIR}/deps/googletest) + +# Build Unit Tests +add_subdirectory(unit_tests) diff --git a/tests/README.md b/tests/README.md new file mode 100644 index 000000000..b12af1e64 --- /dev/null +++ b/tests/README.md @@ -0,0 +1,41 @@ +# oneMKL Interfaces Testing + +## Overview +Inside the `unit_tests` directory, there are domain-level directories which contain domain-specific tests, usually per function or per configuration. + +## Steps +Functional testing is enabled by default via `CMake`, so all relevant functional tests will run automatically after the project is built successfully. + +*Note: A set of `build options` define a `build configuration`. `CMake` builds and runs different set of tests depending on your `build configuration`. This is because `CMake` generates an export header file (config.hpp) for the selected build configuration. Check `/src/config.hpp.in` and `/src/CMakeLists.txt` for details. For details on how `CMake` performs export header generation, refer to [CMake documentation](https://cmake.org/cmake/help/v3.13/module/GenerateExportHeader.html).* + +You can re-run tests without re-building the entire project. + +#### The `CMake` Approach Works for any Generator +```bash +cmake --build . --target test +``` + +#### To use Generator-specific Commands: + +```bash +# For ninja +ninja test +``` + +```bash +# For GNU Makefiles +ctest +# Test filter use case - runs only Gpu specific tests +ctest -R Gpu +# Exclude filtering use case - excludes Cpu tests +ctest -E Cpu +``` + +For more `ctest` options, refer to [ctest manual page](https://cmake.org/cmake/help/v3.13/manual/ctest.1.html). + +## BLAS + +The tests in the level\ directories are for the corresponding level\ BLAS routines. [GoogleTest](https://github.com/google/googletest) is used as the unit-testing framework. + + +*Refer to `/deps/googletest/LICENSE` for GoogleTest license.* diff --git a/tests/unit_tests/CMakeLists.txt b/tests/unit_tests/CMakeLists.txt new file mode 100644 index 000000000..860d05c8d --- /dev/null +++ b/tests/unit_tests/CMakeLists.txt @@ -0,0 +1,98 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +find_package(CBLAS REQUIRED) + +# Build BLAS tests first +add_subdirectory(blas) + +include(GoogleTest) + +# Build final test binaries: test_main_rt is for testing RunTime API, test_main_ct is for testing CompileTime API +get_target_property(GTEST_INCLUDE_DIR gtest INTERFACE_INCLUDE_DIRECTORIES) + +add_executable(test_main_ct main_test.cpp) +target_include_directories(test_main_ct PUBLIC ${GTEST_INCLUDE_DIR}) +target_compile_options(test_main_ct PRIVATE -fsycl) + +if(BUILD_SHARED_LIBS) + add_executable(test_main_rt main_test.cpp) + target_include_directories(test_main_rt PUBLIC ${GTEST_INCLUDE_DIR}) + target_compile_options(test_main_rt PRIVATE -fsycl) + target_link_libraries(test_main_rt PUBLIC + gtest + gtest_main + ${CMAKE_DL_LIBS} + ${CBLAS_LINK} + ONEMKL::SYCL::SYCL + onemkl + blas_level1_rt + blas_level2_rt + blas_level3_rt + ) +endif() + +if(ENABLE_MKLCPU_BACKEND) + add_dependencies(test_main_ct onemkl_blas_mklcpu) + if(BUILD_SHARED_LIBS) + list(APPEND ONEMKL_LIBRARIES onemkl_blas_mklcpu) + else() + list(APPEND ONEMKL_LIBRARIES -foffload-static-lib=${CMAKE_LIBRARY_OUTPUT_DIRECTORY}/libonemkl_blas_mklcpu.a) + find_package(MKL REQUIRED) + list(APPEND ONEMKL_LIBRARIES ${MKL_LINK_C}) + endif() +endif() + +if(ENABLE_MKLGPU_BACKEND) + add_dependencies(test_main_ct onemkl_blas_mklgpu) + if(BUILD_SHARED_LIBS) + list(APPEND ONEMKL_LIBRARIES onemkl_blas_mklgpu) + else() + list(APPEND ONEMKL_LIBRARIES -foffload-static-lib=${CMAKE_LIBRARY_OUTPUT_DIRECTORY}/libonemkl_blas_mklgpu.a) + find_package(MKL REQUIRED) + list(APPEND ONEMKL_LIBRARIES ${MKL_LINK_SYCL}) + endif() +endif() + +target_link_libraries(test_main_ct PUBLIC + gtest + gtest_main + ${CMAKE_DL_LIBS} + ${CBLAS_LINK} + ${ONEMKL_LIBRARIES} + ONEMKL::SYCL::SYCL + blas_level1_ct + blas_level2_ct + blas_level3_ct +) + +if(BUILD_SHARED_LIBS) + set_target_properties(test_main_rt + PROPERTIES BUILD_RPATH $) +# Find individual tests within executable + gtest_discover_tests(test_main_rt + PROPERTIES BUILD_RPATH ${CMAKE_BINARY_DIR}/lib + PROPERTIES ENVIRONMENT LD_LIBRARY_PATH=${CMAKE_BINARY_DIR}/lib:$ENV{LD_LIBRARY_PATH} + ) +endif() + +gtest_discover_tests(test_main_ct + PROPERTIES BUILD_RPATH ${CMAKE_BINARY_DIR}/lib + PROPERTIES ENVIRONMENT LD_LIBRARY_PATH=${CMAKE_BINARY_DIR}/lib:$ENV{LD_LIBRARY_PATH} +) diff --git a/tests/unit_tests/blas/CMakeLists.txt b/tests/unit_tests/blas/CMakeLists.txt new file mode 100644 index 000000000..dce06a451 --- /dev/null +++ b/tests/unit_tests/blas/CMakeLists.txt @@ -0,0 +1,22 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +add_subdirectory(level1) +add_subdirectory(level2) +add_subdirectory(level3) diff --git a/tests/unit_tests/blas/include/allocator_helper.hpp b/tests/unit_tests/blas/include/allocator_helper.hpp new file mode 100644 index 000000000..d2de50438 --- /dev/null +++ b/tests/unit_tests/blas/include/allocator_helper.hpp @@ -0,0 +1,77 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef __ALLOCATOR_HELPER_HPP +#define __ALLOCATOR_HELPER_HPP + +#include +#include +#include +#include + +template +struct allocator_helper { + typedef T* pointer; + typedef const T* const_pointer; + typedef void* void_pointer; + typedef const void* const_void_pointer; + typedef T value_type; + typedef size_t size_type; + typedef ptrdiff_t difference_type; + + template + struct rebind { + typedef allocator_helper other; + }; + + allocator_helper() noexcept {} + template + allocator_helper(allocator_helper& other) noexcept {} + template + allocator_helper(allocator_helper&& other) noexcept {} + + T* allocate(size_t n) { + void* mem = aligned_alloc(align, n * sizeof(T)); + if (!mem) + throw std::bad_alloc(); + + return static_cast(mem); + } + + void deallocate(T* p, size_t n) noexcept { + free(p); + } + + constexpr size_t max_size() const noexcept { + return std::numeric_limits::max() / sizeof(T); + } + + template + constexpr bool operator==(const allocator_helper) const noexcept { + return true; + } + template + constexpr bool operator!=(const allocator_helper) const noexcept { + return false; + } + + typedef std::true_type is_always_equal; +}; + +#endif diff --git a/tests/unit_tests/blas/include/onemkl_blas_helper.hpp b/tests/unit_tests/blas/include/onemkl_blas_helper.hpp new file mode 100644 index 000000000..e79719df4 --- /dev/null +++ b/tests/unit_tests/blas/include/onemkl_blas_helper.hpp @@ -0,0 +1,53 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef ONEMKL_BLAS_HELPER_HPP +#define ONEMKL_BLAS_HELPER_HPP + +#include "cblas.h" + +#include "onemkl/types.hpp" + +/** + * Helper methods for converting between onemkl types and their BLAS + * equivalents. + */ + +inline CBLAS_TRANSPOSE convert_to_cblas_trans(onemkl::transpose trans) { + if (trans == onemkl::transpose::trans) + return CBLAS_TRANSPOSE::CblasTrans; + else if (trans == onemkl::transpose::conjtrans) + return CBLAS_TRANSPOSE::CblasConjTrans; + else + return CBLAS_TRANSPOSE::CblasNoTrans; +} + +inline CBLAS_UPLO convert_to_cblas_uplo(onemkl::uplo is_upper) { + return is_upper == onemkl::uplo::upper ? CBLAS_UPLO::CblasUpper : CBLAS_UPLO::CblasLower; +} + +inline CBLAS_DIAG convert_to_cblas_diag(onemkl::diag is_unit) { + return is_unit == onemkl::diag::unit ? CBLAS_DIAG::CblasUnit : CBLAS_DIAG::CblasNonUnit; +} + +inline CBLAS_SIDE convert_to_cblas_side(onemkl::side is_left) { + return is_left == onemkl::side::left ? CBLAS_SIDE::CblasLeft : CBLAS_SIDE::CblasRight; +} + +#endif // ONEMKL_BLAS_HELPER_HPP diff --git a/tests/unit_tests/blas/include/reference_blas_templates.hpp b/tests/unit_tests/blas/include/reference_blas_templates.hpp new file mode 100644 index 000000000..86a4a0201 --- /dev/null +++ b/tests/unit_tests/blas/include/reference_blas_templates.hpp @@ -0,0 +1,1252 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _REFERENCE_BLAS_TEMPLATES_HPP__ +#define _REFERENCE_BLAS_TEMPLATES_HPP__ + +#include +#include "cblas.h" + +extern "C" { + +void csrot_(const int *N, void *X, const int *incX, void *Y, const int *incY, const float *c, + const float *s); + +void zdrot_(const int *N, void *X, const int *incX, void *Y, const int *incY, const double *c, + const double *s); + +void crotg_(void *a, void *b, const float *c, void *s); + +void zrotg_(void *a, void *b, const double *c, void *s); +} + +// Level 3 + +template +static void gemm(CBLAS_TRANSPOSE transa, CBLAS_TRANSPOSE transb, const int *m, const int *n, + const int *k, const fp *alpha, const fp *a, const int *lda, const fp *b, + const int *ldb, const fp *beta, fp *c, const int *ldc); + +template <> +void gemm(CBLAS_TRANSPOSE transa, CBLAS_TRANSPOSE transb, const int *m, const int *n, const int *k, + const float *alpha, const float *a, const int *lda, const float *b, const int *ldb, + const float *beta, float *c, const int *ldc) { + cblas_sgemm(CblasColMajor, transa, transb, *m, *n, *k, *alpha, a, *lda, b, *ldb, *beta, c, + *ldc); +} + +template <> +void gemm(CBLAS_TRANSPOSE transa, CBLAS_TRANSPOSE transb, const int *m, const int *n, const int *k, + const double *alpha, const double *a, const int *lda, const double *b, const int *ldb, + const double *beta, double *c, const int *ldc) { + cblas_dgemm(CblasColMajor, transa, transb, *m, *n, *k, *alpha, a, *lda, b, *ldb, *beta, c, + *ldc); +} + +template <> +void gemm(CBLAS_TRANSPOSE transa, CBLAS_TRANSPOSE transb, const int *m, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_cgemm(CblasColMajor, transa, transb, *m, *n, *k, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template <> +void gemm(CBLAS_TRANSPOSE transa, CBLAS_TRANSPOSE transb, const int *m, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_zgemm(CblasColMajor, transa, transb, *m, *n, *k, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template +static void symm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, + const fp *alpha, const fp *a, const int *lda, const fp *b, const int *ldb, + const fp *beta, fp *c, const int *ldc); + +template <> +void symm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, const float *alpha, + const float *a, const int *lda, const float *b, const int *ldb, const float *beta, + float *c, const int *ldc) { + cblas_ssymm(CblasColMajor, left_right, uplo, *m, *n, *alpha, a, *lda, b, *ldb, *beta, c, *ldc); +} + +template <> +void symm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, const double *alpha, + const double *a, const int *lda, const double *b, const int *ldb, const double *beta, + double *c, const int *ldc) { + cblas_dsymm(CblasColMajor, left_right, uplo, *m, *n, *alpha, a, *lda, b, *ldb, *beta, c, *ldc); +} + +template <> +void symm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_csymm(CblasColMajor, left_right, uplo, *m, *n, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template <> +void symm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_zsymm(CblasColMajor, left_right, uplo, *m, *n, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template +static void syrk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const fp *alpha, const fp *a, const int *lda, const fp *beta, fp *c, + const int *ldc); + +template <> +void syrk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, const float *alpha, + const float *a, const int *lda, const float *beta, float *c, const int *ldc) { + cblas_ssyrk(CblasColMajor, uplo, trans, *n, *k, *alpha, a, *lda, *beta, c, *ldc); +} + +template <> +void syrk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, const double *alpha, + const double *a, const int *lda, const double *beta, double *c, const int *ldc) { + cblas_dsyrk(CblasColMajor, uplo, trans, *n, *k, *alpha, a, *lda, *beta, c, *ldc); +} + +template <> +void syrk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *beta, std::complex *c, const int *ldc) { + cblas_csyrk(CblasColMajor, uplo, trans, *n, *k, alpha, a, *lda, beta, c, *ldc); +} + +template <> +void syrk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *beta, std::complex *c, const int *ldc) { + cblas_zsyrk(CblasColMajor, uplo, trans, *n, *k, alpha, a, *lda, beta, c, *ldc); +} + +template +static void hemm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, + const fp *alpha, const fp *a, const int *lda, const fp *b, const int *ldb, + const fp *beta, fp *c, const int *ldc); + +template <> +void hemm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_chemm(CblasColMajor, left_right, uplo, *m, *n, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template <> +void hemm(CBLAS_SIDE left_right, CBLAS_UPLO uplo, const int *m, const int *n, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_zhemm(CblasColMajor, left_right, uplo, *m, *n, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template +static void herk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const fp_scalar *alpha, const fp_data *a, const int *lda, const fp_scalar *beta, + fp_data *c, const int *ldc); + +template <> +void herk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, const float *alpha, + const std::complex *a, const int *lda, const float *beta, std::complex *c, + const int *ldc) { + cblas_cherk(CblasColMajor, uplo, trans, *n, *k, *alpha, a, *lda, *beta, c, *ldc); +} + +template <> +void herk(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, const double *alpha, + const std::complex *a, const int *lda, const double *beta, + std::complex *c, const int *ldc) { + cblas_zherk(CblasColMajor, uplo, trans, *n, *k, *alpha, a, *lda, *beta, c, *ldc); +} + +template +static void syr2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const fp *alpha, const fp *a, const int *lda, const fp *b, const int *ldb, + const fp *beta, fp *c, const int *ldc); + +template <> +void syr2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, const float *alpha, + const float *a, const int *lda, const float *b, const int *ldb, const float *beta, + float *c, const int *ldc) { + cblas_ssyr2k(CblasColMajor, uplo, trans, *n, *k, *alpha, a, *lda, b, *ldb, *beta, c, *ldc); +} + +template <> +void syr2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, const double *alpha, + const double *a, const int *lda, const double *b, const int *ldb, const double *beta, + double *c, const int *ldc) { + cblas_dsyr2k(CblasColMajor, uplo, trans, *n, *k, *alpha, a, *lda, b, *ldb, *beta, c, *ldc); +} + +template <> +void syr2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_csyr2k(CblasColMajor, uplo, trans, *n, *k, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template <> +void syr2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const std::complex *beta, + std::complex *c, const int *ldc) { + cblas_zsyr2k(CblasColMajor, uplo, trans, *n, *k, alpha, a, *lda, b, *ldb, beta, c, *ldc); +} + +template +static void her2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const fp_data *alpha, const fp_data *a, const int *lda, const fp_data *b, + const int *ldb, const fp_scalar *beta, fp_data *c, const int *ldc); + +template <> +void her2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const float *beta, std::complex *c, + const int *ldc) { + cblas_cher2k(CblasColMajor, uplo, trans, *n, *k, alpha, a, *lda, b, *ldb, *beta, c, *ldc); +} + +template <> +void her2k(CBLAS_UPLO uplo, CBLAS_TRANSPOSE trans, const int *n, const int *k, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *b, const int *ldb, const double *beta, + std::complex *c, const int *ldc) { + cblas_zher2k(CblasColMajor, uplo, trans, *n, *k, alpha, a, *lda, b, *ldb, *beta, c, *ldc); +} + +template +static void trmm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, + const int *m, const int *n, const fp *alpha, const fp *a, const int *lda, fp *b, + const int *ldb); + +template <> +void trmm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const float *alpha, const float *a, const int *lda, float *b, + const int *ldb) { + cblas_strmm(CblasColMajor, side, uplo, transa, diag, *m, *n, *alpha, a, *lda, b, *ldb); +} + +template <> +void trmm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const double *alpha, const double *a, const int *lda, double *b, + const int *ldb) { + cblas_dtrmm(CblasColMajor, side, uplo, transa, diag, *m, *n, *alpha, a, *lda, b, *ldb); +} + +template <> +void trmm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const std::complex *alpha, const std::complex *a, + const int *lda, std::complex *b, const int *ldb) { + cblas_ctrmm(CblasColMajor, side, uplo, transa, diag, *m, *n, alpha, a, *lda, b, *ldb); +} + +template <> +void trmm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const std::complex *alpha, const std::complex *a, + const int *lda, std::complex *b, const int *ldb) { + cblas_ztrmm(CblasColMajor, side, uplo, transa, diag, *m, *n, alpha, a, *lda, b, *ldb); +} + +template +static void trsm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, + const int *m, const int *n, const fp *alpha, const fp *a, const int *lda, fp *b, + const int *ldb); + +template <> +void trsm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const float *alpha, const float *a, const int *lda, float *b, + const int *ldb) { + cblas_strsm(CblasColMajor, side, uplo, transa, diag, *m, *n, *alpha, a, *lda, b, *ldb); +} + +template <> +void trsm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const double *alpha, const double *a, const int *lda, double *b, + const int *ldb) { + cblas_dtrsm(CblasColMajor, side, uplo, transa, diag, *m, *n, *alpha, a, *lda, b, *ldb); +} + +template <> +void trsm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const std::complex *alpha, const std::complex *a, + const int *lda, std::complex *b, const int *ldb) { + cblas_ctrsm(CblasColMajor, side, uplo, transa, diag, *m, *n, alpha, a, *lda, b, *ldb); +} + +template <> +void trsm(CBLAS_SIDE side, CBLAS_UPLO uplo, CBLAS_TRANSPOSE transa, CBLAS_DIAG diag, const int *m, + const int *n, const std::complex *alpha, const std::complex *a, + const int *lda, std::complex *b, const int *ldb) { + cblas_ztrsm(CblasColMajor, side, uplo, transa, diag, *m, *n, alpha, a, *lda, b, *ldb); +} + +// Level 2 + +template +static void gemv(CBLAS_TRANSPOSE trans, const int *m, const int *n, const fp *alpha, const fp *a, + const int *lda, const fp *x, const int *incx, const fp *beta, fp *y, + const int *incy); + +template <> +void gemv(CBLAS_TRANSPOSE trans, const int *m, const int *n, const float *alpha, const float *a, + const int *lda, const float *x, const int *incx, const float *beta, float *y, + const int *incy) { + cblas_sgemv(CblasColMajor, trans, *m, *n, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template <> +void gemv(CBLAS_TRANSPOSE trans, const int *m, const int *n, const double *alpha, const double *a, + const int *lda, const double *x, const int *incx, const double *beta, double *y, + const int *incy) { + cblas_dgemv(CblasColMajor, trans, *m, *n, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template <> +void gemv(CBLAS_TRANSPOSE trans, const int *m, const int *n, const std::complex *alpha, + const std::complex *a, const int *lda, const std::complex *x, + const int *incx, const std::complex *beta, std::complex *y, + const int *incy) { + cblas_cgemv(CblasColMajor, trans, *m, *n, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template <> +void gemv(CBLAS_TRANSPOSE trans, const int *m, const int *n, const std::complex *alpha, + const std::complex *a, const int *lda, const std::complex *x, + const int *incx, const std::complex *beta, std::complex *y, + const int *incy) { + cblas_zgemv(CblasColMajor, trans, *m, *n, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template +static void gbmv(CBLAS_TRANSPOSE trans, const int *m, const int *n, int *kl, int *ku, + const fp *alpha, const fp *a, const int *lda, const fp *x, const int *incx, + const fp *beta, fp *y, const int *incy); + +template <> +void gbmv(CBLAS_TRANSPOSE trans, const int *m, const int *n, int *kl, int *ku, const float *alpha, + const float *a, const int *lda, const float *x, const int *incx, const float *beta, + float *y, const int *incy) { + cblas_sgbmv(CblasColMajor, trans, *m, *n, *kl, *ku, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template <> +void gbmv(CBLAS_TRANSPOSE trans, const int *m, const int *n, int *kl, int *ku, const double *alpha, + const double *a, const int *lda, const double *x, const int *incx, const double *beta, + double *y, const int *incy) { + cblas_dgbmv(CblasColMajor, trans, *m, *n, *kl, *ku, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template <> +void gbmv(CBLAS_TRANSPOSE trans, const int *m, const int *n, int *kl, int *ku, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *x, const int *incx, const std::complex *beta, + std::complex *y, const int *incy) { + cblas_cgbmv(CblasColMajor, trans, *m, *n, *kl, *ku, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template <> +void gbmv(CBLAS_TRANSPOSE trans, const int *m, const int *n, int *kl, int *ku, + const std::complex *alpha, const std::complex *a, const int *lda, + const std::complex *x, const int *incx, const std::complex *beta, + std::complex *y, const int *incy) { + cblas_zgbmv(CblasColMajor, trans, *m, *n, *kl, *ku, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template +static void ger(const int *m, const int *n, const fp *alpha, const fp *x, const int *incx, + const fp *y, const int *incy, fp *a, const int *lda); + +template <> +void ger(const int *m, const int *n, const float *alpha, const float *x, const int *incx, + const float *y, const int *incy, float *a, const int *lda) { + cblas_sger(CblasColMajor, *m, *n, *alpha, x, *incx, y, *incy, a, *lda); +} + +template <> +void ger(const int *m, const int *n, const double *alpha, const double *x, const int *incx, + const double *y, const int *incy, double *a, const int *lda) { + cblas_dger(CblasColMajor, *m, *n, *alpha, x, *incx, y, *incy, a, *lda); +} + +template +static void gerc(const int *m, const int *n, const fp *alpha, const fp *x, const int *incx, + const fp *y, const int *incy, fp *a, const int *lda); + +template <> +void gerc(const int *m, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a, const int *lda) { + cblas_cgerc(CblasColMajor, *m, *n, alpha, x, *incx, y, *incy, a, *lda); +} + +template <> +void gerc(const int *m, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a, const int *lda) { + cblas_zgerc(CblasColMajor, *m, *n, alpha, x, *incx, y, *incy, a, *lda); +} + +template +static void geru(const int *m, const int *n, const fp *alpha, const fp *x, const int *incx, + const fp *y, const int *incy, fp *a, const int *lda); + +template <> +void geru(const int *m, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a, const int *lda) { + cblas_cgeru(CblasColMajor, *m, *n, alpha, x, *incx, y, *incy, a, *lda); +} + +template <> +void geru(const int *m, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a, const int *lda) { + cblas_zgeru(CblasColMajor, *m, *n, alpha, x, *incx, y, *incy, a, *lda); +} + +template +static void hbmv(CBLAS_UPLO upper_lower, const int *n, const int *k, const fp *alpha, const fp *a, + const int *lda, const fp *x, const int *incx, const fp *beta, fp *y, + const int *incy); + +template <> +void hbmv(CBLAS_UPLO upper_lower, const int *n, const int *k, const std::complex *alpha, + const std::complex *a, const int *lda, const std::complex *x, + const int *incx, const std::complex *beta, std::complex *y, + const int *incy) { + cblas_chbmv(CblasColMajor, upper_lower, *n, *k, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template <> +void hbmv(CBLAS_UPLO upper_lower, const int *n, const int *k, const std::complex *alpha, + const std::complex *a, const int *lda, const std::complex *x, + const int *incx, const std::complex *beta, std::complex *y, + const int *incy) { + cblas_zhbmv(CblasColMajor, upper_lower, *n, *k, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template +static void hemv(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *a, const int *lda, + const fp *x, const int *incx, const fp *beta, fp *y, const int *incy); + +template <> +void hemv(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *a, const int *lda, const std::complex *x, + const int *incx, const std::complex *beta, std::complex *y, + const int *incy) { + cblas_chemv(CblasColMajor, upper_lower, *n, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template <> +void hemv(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *a, const int *lda, const std::complex *x, + const int *incx, const std::complex *beta, std::complex *y, + const int *incy) { + cblas_zhemv(CblasColMajor, upper_lower, *n, alpha, a, *lda, x, *incx, beta, y, *incy); +} + +template +static void her(CBLAS_UPLO upper_lower, const int *n, const fp_scalar *alpha, const fp_data *x, + const int *incx, fp_data *a, const int *lda); + +template <> +void her(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const std::complex *x, + const int *incx, std::complex *a, const int *lda) { + cblas_cher(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a, *lda); +} + +template <> +void her(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const std::complex *x, + const int *incx, std::complex *a, const int *lda) { + cblas_zher(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a, *lda); +} + +template +static void her2(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *x, + const int *incx, const fp *y, const int *incy, fp *a, const int *lda); + +template <> +void her2(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a, const int *lda) { + cblas_cher2(CblasColMajor, upper_lower, *n, alpha, x, *incx, y, *incy, a, *lda); +} + +template <> +void her2(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a, const int *lda) { + cblas_zher2(CblasColMajor, upper_lower, *n, alpha, x, *incx, y, *incy, a, *lda); +} + +template +static void hpmv(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *a, const fp *x, + const int *incx, const fp *beta, fp *y, const int *incy); + +template <> +void hpmv(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *a, const std::complex *x, const int *incx, + const std::complex *beta, std::complex *y, const int *incy) { + cblas_chpmv(CblasColMajor, upper_lower, *n, alpha, a, x, *incx, beta, y, *incy); +} + +template <> +void hpmv(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *a, const std::complex *x, const int *incx, + const std::complex *beta, std::complex *y, const int *incy) { + cblas_zhpmv(CblasColMajor, upper_lower, *n, alpha, a, x, *incx, beta, y, *incy); +} + +template +static void hpr(CBLAS_UPLO upper_lower, const int *n, const fp_scalar *alpha, const fp_data *x, + const int *incx, fp_data *a); + +template <> +void hpr(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const std::complex *x, + const int *incx, std::complex *a) { + cblas_chpr(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a); +} + +template <> +void hpr(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const std::complex *x, + const int *incx, std::complex *a) { + cblas_zhpr(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a); +} + +template +static void hpr2(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *x, + const int *incx, const fp *y, const int *incy, fp *a); + +template <> +void hpr2(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a) { + cblas_chpr2(CblasColMajor, upper_lower, *n, alpha, x, *incx, y, *incy, a); +} + +template <> +void hpr2(CBLAS_UPLO upper_lower, const int *n, const std::complex *alpha, + const std::complex *x, const int *incx, const std::complex *y, + const int *incy, std::complex *a) { + cblas_zhpr2(CblasColMajor, upper_lower, *n, alpha, x, *incx, y, *incy, a); +} + +template +static void sbmv(CBLAS_UPLO upper_lower, const int *n, const int *k, const fp *alpha, const fp *a, + const int *lda, const fp *x, const int *incx, const fp *beta, fp *y, + const int *incy); + +template <> +void sbmv(CBLAS_UPLO upper_lower, const int *n, const int *k, const float *alpha, const float *a, + const int *lda, const float *x, const int *incx, const float *beta, float *y, + const int *incy) { + cblas_ssbmv(CblasColMajor, upper_lower, *n, *k, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template <> +void sbmv(CBLAS_UPLO upper_lower, const int *n, const int *k, const double *alpha, const double *a, + const int *lda, const double *x, const int *incx, const double *beta, double *y, + const int *incy) { + cblas_dsbmv(CblasColMajor, upper_lower, *n, *k, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template +static void symv(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *a, const int *lda, + const fp *x, const int *incx, const fp *beta, fp *y, const int *incy); + +template <> +void symv(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const float *a, const int *lda, + const float *x, const int *incx, const float *beta, float *y, const int *incy) { + cblas_ssymv(CblasColMajor, upper_lower, *n, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template <> +void symv(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const double *a, + const int *lda, const double *x, const int *incx, const double *beta, double *y, + const int *incy) { + cblas_dsymv(CblasColMajor, upper_lower, *n, *alpha, a, *lda, x, *incx, *beta, y, *incy); +} + +template +static void syr(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *x, const int *incx, + fp *a, const int *lda); + +template <> +void syr(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const float *x, const int *incx, + float *a, const int *lda) { + cblas_ssyr(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a, *lda); +} + +template <> +void syr(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const double *x, + const int *incx, double *a, const int *lda) { + cblas_dsyr(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a, *lda); +} + +template +static void syr2(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *x, + const int *incx, const fp *y, const int *incy, fp *a, const int *lda); + +template <> +void syr2(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const float *x, const int *incx, + const float *y, const int *incy, float *a, const int *lda) { + cblas_ssyr2(CblasColMajor, upper_lower, *n, *alpha, x, *incx, y, *incy, a, *lda); +} + +template <> +void syr2(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const double *x, + const int *incx, const double *y, const int *incy, double *a, const int *lda) { + cblas_dsyr2(CblasColMajor, upper_lower, *n, *alpha, x, *incx, y, *incy, a, *lda); +} + +template +static void spmv(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *a, const fp *x, + const int *incx, const fp *beta, fp *y, const int *incy); + +template <> +void spmv(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const float *a, const float *x, + const int *incx, const float *beta, float *y, const int *incy) { + cblas_sspmv(CblasColMajor, upper_lower, *n, *alpha, a, x, *incx, *beta, y, *incy); +} + +template <> +void spmv(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const double *a, + const double *x, const int *incx, const double *beta, double *y, const int *incy) { + cblas_dspmv(CblasColMajor, upper_lower, *n, *alpha, a, x, *incx, *beta, y, *incy); +} + +template +static void spr(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *x, const int *incx, + fp *a); + +template <> +void spr(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const float *x, const int *incx, + float *a) { + cblas_sspr(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a); +} + +template <> +void spr(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const double *x, + const int *incx, double *a) { + cblas_dspr(CblasColMajor, upper_lower, *n, *alpha, x, *incx, a); +} + +template +static void spr2(CBLAS_UPLO upper_lower, const int *n, const fp *alpha, const fp *x, + const int *incx, const fp *y, const int *incy, fp *a); + +template <> +void spr2(CBLAS_UPLO upper_lower, const int *n, const float *alpha, const float *x, const int *incx, + const float *y, const int *incy, float *a) { + cblas_sspr2(CblasColMajor, upper_lower, *n, *alpha, x, *incx, y, *incy, a); +} + +template <> +void spr2(CBLAS_UPLO upper_lower, const int *n, const double *alpha, const double *x, + const int *incx, const double *y, const int *incy, double *a) { + cblas_dspr2(CblasColMajor, upper_lower, *n, *alpha, x, *incx, y, *incy, a); +} + +template +static void tbmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const fp *a, const int *lda, fp *x, const int *incx); + +template <> +void tbmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const float *a, const int *lda, float *x, const int *incx) { + cblas_stbmv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template <> +void tbmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const double *a, const int *lda, double *x, const int *incx) { + cblas_dtbmv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template <> +void tbmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const std::complex *a, const int *lda, std::complex *x, + const int *incx) { + cblas_ctbmv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template <> +void tbmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const std::complex *a, const int *lda, std::complex *x, + const int *incx) { + cblas_ztbmv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template +static void tbsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const fp *a, const int *lda, fp *x, const int *incx); + +template <> +void tbsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const float *a, const int *lda, float *x, const int *incx) { + cblas_stbsv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template <> +void tbsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const double *a, const int *lda, double *x, const int *incx) { + cblas_dtbsv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template <> +void tbsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const std::complex *a, const int *lda, std::complex *x, + const int *incx) { + cblas_ctbsv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template <> +void tbsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const int *k, const std::complex *a, const int *lda, std::complex *x, + const int *incx) { + cblas_ztbsv(CblasColMajor, upper_lower, trans, unit_diag, *n, *k, a, *lda, x, *incx); +} + +template +static void tpmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const fp *a, fp *x, const int *incx); + +template <> +void tpmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const float *a, float *x, const int *incx) { + cblas_stpmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template <> +void tpmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const double *a, double *x, const int *incx) { + cblas_dtpmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template <> +void tpmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, std::complex *x, const int *incx) { + cblas_ctpmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template <> +void tpmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, std::complex *x, const int *incx) { + cblas_ztpmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template +static void tpsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const fp *a, fp *x, const int *incx); + +template <> +void tpsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const float *a, float *x, const int *incx) { + cblas_stpsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template <> +void tpsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const double *a, double *x, const int *incx) { + cblas_dtpsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template <> +void tpsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, std::complex *x, const int *incx) { + cblas_ctpsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template <> +void tpsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, std::complex *x, const int *incx) { + cblas_ztpsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, x, *incx); +} + +template +static void trmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const fp *a, const int *lda, fp *x, const int *incx); + +template <> +void trmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const float *a, const int *lda, float *x, const int *incx) { + cblas_strmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template <> +void trmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const double *a, const int *lda, double *x, const int *incx) { + cblas_dtrmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template <> +void trmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, const int *lda, std::complex *x, const int *incx) { + cblas_ctrmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template <> +void trmv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, const int *lda, std::complex *x, const int *incx) { + cblas_ztrmv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template +static void trsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const fp *a, const int *lda, fp *x, const int *incx); + +template <> +void trsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const float *a, const int *lda, float *x, const int *incx) { + cblas_strsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template <> +void trsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const double *a, const int *lda, double *x, const int *incx) { + cblas_dtrsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template <> +void trsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, const int *lda, std::complex *x, const int *incx) { + cblas_ctrsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +template <> +void trsv(CBLAS_UPLO upper_lower, CBLAS_TRANSPOSE trans, CBLAS_DIAG unit_diag, const int *n, + const std::complex *a, const int *lda, std::complex *x, const int *incx) { + cblas_ztrsv(CblasColMajor, upper_lower, trans, unit_diag, *n, a, *lda, x, *incx); +} + +// Level 1 + +template +static fp_res asum(const int *n, const fp_data *x, const int *incx); + +template <> +float asum(const int *n, const float *x, const int *incx) { + return cblas_sasum(*n, x, *incx); +} + +template <> +double asum(const int *n, const double *x, const int *incx) { + return cblas_dasum(*n, x, *incx); +} + +template <> +float asum(const int *n, const std::complex *x, const int *incx) { + return cblas_scasum(*n, x, *incx); +} + +template <> +double asum(const int *n, const std::complex *x, const int *incx) { + return cblas_dzasum(*n, x, *incx); +} + +template +static void axpy(const int *n, const fp *alpha, const fp *x, const int *incx, fp *y, + const int *incy); + +template <> +void axpy(const int *n, const float *alpha, const float *x, const int *incx, float *y, + const int *incy) { + cblas_saxpy(*n, *alpha, x, *incx, y, *incy); +} + +template <> +void axpy(const int *n, const double *alpha, const double *x, const int *incx, double *y, + const int *incy) { + cblas_daxpy(*n, *alpha, x, *incx, y, *incy); +} + +template <> +void axpy(const int *n, const std::complex *alpha, const std::complex *x, + const int *incx, std::complex *y, const int *incy) { + cblas_caxpy(*n, alpha, x, *incx, y, *incy); +} + +template <> +void axpy(const int *n, const std::complex *alpha, const std::complex *x, + const int *incx, std::complex *y, const int *incy) { + cblas_zaxpy(*n, alpha, x, *incx, y, *incy); +} + +template +static void copy(const int *n, const fp *x, const int *incx, fp *y, const int *incy); + +template <> +void copy(const int *n, const float *x, const int *incx, float *y, const int *incy) { + cblas_scopy(*n, x, *incx, y, *incy); +} +template <> +void copy(const int *n, const double *x, const int *incx, double *y, const int *incy) { + cblas_dcopy(*n, x, *incx, y, *incy); +} +template <> +void copy(const int *n, const std::complex *x, const int *incx, std::complex *y, + const int *incy) { + cblas_ccopy(*n, x, *incx, y, *incy); +} +template <> +void copy(const int *n, const std::complex *x, const int *incx, std::complex *y, + const int *incy) { + cblas_zcopy(*n, x, *incx, y, *incy); +} + +template +static fp_res dot(const int *n, const fp *x, const int *incx, const fp *y, const int *incy); + +template <> +float dot(const int *n, const float *x, const int *incx, const float *y, const int *incy) { + return cblas_sdot(*n, x, *incx, y, *incy); +} + +template <> +double dot(const int *n, const double *x, const int *incx, const double *y, const int *incy) { + return cblas_ddot(*n, x, *incx, y, *incy); +} + +template <> +double dot(const int *n, const float *x, const int *incx, const float *y, const int *incy) { + return cblas_dsdot(*n, x, *incx, y, *incy); +} + +static float sdsdot(const int *n, const float *sb, const float *x, const int *incx, const float *y, + const int *incy) { + return cblas_sdsdot(*n, *sb, x, *incx, y, *incy); +} + +template +static fp_res nrm2(const int *n, const fp *x, const int *incx); + +template <> +float nrm2(const int *n, const float *x, const int *incx) { + return cblas_snrm2(*n, x, *incx); +} + +template <> +double nrm2(const int *n, const double *x, const int *incx) { + return cblas_dnrm2(*n, x, *incx); +} + +template <> +float nrm2(const int *n, const std::complex *x, const int *incx) { + return cblas_scnrm2(*n, x, *incx); +} + +template <> +double nrm2(const int *n, const std::complex *x, const int *incx) { + return cblas_dznrm2(*n, x, *incx); +} + +template +static void rot(const int *n, fp *x, const int *incx, fp *y, const int *incy, const fp_scalar *c, + const fp_scalar *s); + +template <> +void rot(const int *n, float *x, const int *incx, float *y, const int *incy, const float *c, + const float *s) { + cblas_srot(*n, x, *incx, y, *incy, *c, *s); +} + +template <> +void rot(const int *n, double *x, const int *incx, double *y, const int *incy, const double *c, + const double *s) { + cblas_drot(*n, x, *incx, y, *incy, *c, *s); +} + +template <> +void rot(const int *n, std::complex *x, const int *incx, std::complex *y, + const int *incy, const float *c, const float *s) { + csrot_(n, (void *)x, incx, (void *)y, incy, c, s); +} + +template <> +void rot(const int *n, std::complex *x, const int *incx, std::complex *y, + const int *incy, const double *c, const double *s) { + zdrot_(n, (void *)x, incx, (void *)y, incy, c, s); +} + +template +static void rotg(fp *a, fp *b, fp_c *c, fp *s); + +template <> +void rotg(float *a, float *b, float *c, float *s) { + cblas_srotg(a, b, c, s); +} + +template <> +void rotg(double *a, double *b, double *c, double *s) { + cblas_drotg(a, b, c, s); +} + +template <> +void rotg(std::complex *a, std::complex *b, float *c, std::complex *s) { + crotg_((void *)a, (void *)b, c, (void *)s); +} + +template <> +void rotg(std::complex *a, std::complex *b, double *c, std::complex *s) { + zrotg_((void *)a, (void *)b, c, (void *)s); +} + +template +static void rotm(const int *n, fp *x, const int *incx, fp *y, const int *incy, const fp *param); + +template <> +void rotm(const int *n, float *x, const int *incx, float *y, const int *incy, const float *param) { + cblas_srotm(*n, x, *incx, y, *incy, param); +} + +template <> +void rotm(const int *n, double *x, const int *incx, double *y, const int *incy, + const double *param) { + cblas_drotm(*n, x, *incx, y, *incy, param); +} + +template +static void rotmg(fp *d1, fp *d2, fp *x1, fp *y1, fp *param); + +template <> +void rotmg(float *d1, float *d2, float *x1, float *y1, float *param) { + cblas_srotmg(d1, d2, x1, *y1, param); +} + +template <> +void rotmg(double *d1, double *d2, double *x1, double *y1, double *param) { + cblas_drotmg(d1, d2, x1, *y1, param); +} + +template +static void scal(const int *n, const fp_scalar *alpha, fp_data *x, const int *incx); + +template <> +void scal(const int *n, const float *alpha, float *x, const int *incx) { + cblas_sscal(*n, *alpha, x, *incx); +} +template <> +void scal(const int *n, const double *alpha, double *x, const int *incx) { + cblas_dscal(*n, *alpha, x, *incx); +} +template <> +void scal(const int *n, const std::complex *alpha, std::complex *x, const int *incx) { + cblas_cscal(*n, alpha, x, *incx); +} +template <> +void scal(const int *n, const std::complex *alpha, std::complex *x, + const int *incx) { + cblas_zscal(*n, alpha, x, *incx); +} +template <> +void scal(const int *n, const float *alpha, std::complex *x, const int *incx) { + cblas_csscal(*n, *alpha, x, *incx); +} +template <> +void scal(const int *n, const double *alpha, std::complex *x, const int *incx) { + cblas_zdscal(*n, *alpha, x, *incx); +} + +template +static void swap(const int *n, fp *x, const int *incx, fp *y, const int *incy); + +template <> +void swap(const int *n, float *x, const int *incx, float *y, const int *incy) { + cblas_sswap(*n, x, *incx, y, *incy); +} + +template <> +void swap(const int *n, double *x, const int *incx, double *y, const int *incy) { + cblas_dswap(*n, x, *incx, y, *incy); +} + +template <> +void swap(const int *n, std::complex *x, const int *incx, std::complex *y, + const int *incy) { + cblas_cswap(*n, x, *incx, y, *incy); +} + +template <> +void swap(const int *n, std::complex *x, const int *incx, std::complex *y, + const int *incy) { + cblas_zswap(*n, x, *incx, y, *incy); +} + +template +static void dotc(fp *pres, const int *n, const fp *x, const int *incx, const fp *y, + const int *incy); + +template <> +void dotc(std::complex *pres, const int *n, const std::complex *x, const int *incx, + const std::complex *y, const int *incy) { + cblas_cdotc_sub(*n, x, *incx, y, *incy, pres); +} + +template <> +void dotc(std::complex *pres, const int *n, const std::complex *x, const int *incx, + const std::complex *y, const int *incy) { + cblas_zdotc_sub(*n, x, *incx, y, *incy, pres); +} + +template +static void dotu(fp *pres, const int *n, const fp *x, const int *incx, const fp *y, + const int *incy); + +template <> +void dotu(std::complex *pres, const int *n, const std::complex *x, const int *incx, + const std::complex *y, const int *incy) { + cblas_cdotu_sub(*n, x, *incx, y, *incy, pres); +} + +template <> +void dotu(std::complex *pres, const int *n, const std::complex *x, const int *incx, + const std::complex *y, const int *incy) { + cblas_zdotu_sub(*n, x, *incx, y, *incy, pres); +} + +template +static int iamax(const int *n, const fp *x, const int *incx); + +template <> +int iamax(const int *n, const float *x, const int *incx) { + return cblas_isamax(*n, x, *incx); +} + +template <> +int iamax(const int *n, const double *x, const int *incx) { + return cblas_idamax(*n, x, *incx); +} + +template <> +int iamax(const int *n, const std::complex *x, const int *incx) { + return cblas_icamax(*n, x, *incx); +} + +template <> +int iamax(const int *n, const std::complex *x, const int *incx) { + return cblas_izamax(*n, x, *incx); +} + +inline float abs_val(float val) { + return std::abs(val); +} + +inline double abs_val(double val) { + return std::abs(val); +} + +inline float abs_val(std::complex val) { + return std::abs(val.real()) + std::abs(val.imag()); +} + +inline double abs_val(std::complex val) { + return std::abs(val.real()) + std::abs(val.imag()); +} + +template +static int iamin(const int *n, const fp *x, const int *incx); + +template <> +int iamin(const int *n, const float *x, const int *incx) { + if (*n < 1 || *incx < 1) { + return 0; + } + int min_idx = 0; + auto min_val = abs_val(x[0]); + + for (int logical_i = 0; logical_i < *n; ++logical_i) { + int i = logical_i * std::abs(*incx); + auto curr_val = abs_val(x[i]); + bool is_first_nan = std::isnan(curr_val) && !std::isnan(min_val); + if (is_first_nan || curr_val < min_val) { + min_idx = logical_i; + min_val = curr_val; + } + } + return min_idx; +} + +template <> +int iamin(const int *n, const double *x, const int *incx) { + if (*n < 1 || *incx < 1) { + return 0; + } + int min_idx = 0; + auto min_val = abs_val(x[0]); + + for (int logical_i = 0; logical_i < *n; ++logical_i) { + int i = logical_i * std::abs(*incx); + auto curr_val = abs_val(x[i]); + bool is_first_nan = std::isnan(curr_val) && !std::isnan(min_val); + if (is_first_nan || curr_val < min_val) { + min_idx = logical_i; + min_val = curr_val; + } + } + return min_idx; +} + +template <> +int iamin(const int *n, const std::complex *x, const int *incx) { + if (*n < 1 || *incx < 1) { + return 0; + } + int min_idx = 0; + auto min_val = abs_val(x[0]); + + for (int logical_i = 0; logical_i < *n; ++logical_i) { + int i = logical_i * std::abs(*incx); + auto curr_val = abs_val(x[i]); + bool is_first_nan = std::isnan(curr_val) && !std::isnan(min_val); + if (is_first_nan || curr_val < min_val) { + min_idx = logical_i; + min_val = curr_val; + } + } + return min_idx; +} + +template <> +int iamin(const int *n, const std::complex *x, const int *incx) { + if (*n < 1 || *incx < 1) { + return 0; + } + int min_idx = 0; + auto min_val = abs_val(x[0]); + + for (int logical_i = 0; logical_i < *n; ++logical_i) { + int i = logical_i * std::abs(*incx); + auto curr_val = abs_val(x[i]); + bool is_first_nan = std::isnan(curr_val) && !std::isnan(min_val); + if (is_first_nan || curr_val < min_val) { + min_idx = logical_i; + min_val = curr_val; + } + } + return min_idx; +} + +#endif /* header guard */ diff --git a/tests/unit_tests/blas/include/test_common.hpp b/tests/unit_tests/blas/include/test_common.hpp new file mode 100644 index 000000000..9b79b11af --- /dev/null +++ b/tests/unit_tests/blas/include/test_common.hpp @@ -0,0 +1,408 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _TEST_COMMON_HPP__ +#define _TEST_COMMON_HPP__ + +#include +#include + +#include + +// Complex helpers. +template +struct complex_info { + using real_type = T; + static const bool is_complex = false; +}; + +template +struct complex_info> { + using real_type = T; + static const bool is_complex = true; +}; + +template +constexpr bool is_complex() { + return complex_info::is_complex; +} +template +constexpr int num_components() { + return is_complex() ? 2 : 1; +} + +// Matrix helpers. +template +constexpr T inner_dimension(onemkl::transpose trans, T m, T n) { + return (trans == onemkl::transpose::nontrans) ? m : n; +} +template +constexpr T outer_dimension(onemkl::transpose trans, T m, T n) { + return (trans == onemkl::transpose::nontrans) ? n : m; +} +template +constexpr T matrix_size(onemkl::transpose trans, T m, T n, T ldm) { + return outer_dimension(trans, m, n) * ldm; +} + +// SYCL buffer creation helper. +template +cl::sycl::buffer make_buffer(const vec &v) { + cl::sycl::buffer buf(v.data(), cl::sycl::range<1>(v.size())); + return buf; +} + +// Reference helpers. +template +struct ref_type_info { + using type = T; +}; + +template <> +struct ref_type_info> { + using type = std::complex; +}; +template <> +struct ref_type_info> { + using type = std::complex; +}; +template <> +struct ref_type_info { + using type = int8_t; +}; +template <> +struct ref_type_info { + using type = uint8_t; +}; +template <> +struct ref_type_info { + using type = int32_t; +}; + +// Random initialization. +template +static fp rand_scalar() { + return fp(std::rand()) / fp(RAND_MAX) - fp(0.5); +} +template +static std::complex rand_complex_scalar() { + return std::complex(rand_scalar(), rand_scalar()); +} +template <> +std::complex rand_scalar() { + return rand_complex_scalar(); +} +template <> +std::complex rand_scalar() { + return rand_complex_scalar(); +} +template <> +int8_t rand_scalar() { + return std::rand() % 254 - 127; +} +template <> +int32_t rand_scalar() { + return std::rand() % 256 - 128; +} +template <> +uint8_t rand_scalar() { + return std::rand() % 128; +} + +template +static fp rand_scalar(int mag) { + fp tmp = fp(mag) + fp(std::rand()) / fp(RAND_MAX) - fp(0.5); + if (std::rand() % 2) + return tmp; + else + return -tmp; +} +template +static std::complex rand_complex_scalar(int mag) { + return std::complex(rand_scalar(mag), rand_scalar(mag)); +} +template <> +std::complex rand_scalar(int mag) { + return rand_complex_scalar(mag); +} +template <> +std::complex rand_scalar(int mag) { + return rand_complex_scalar(mag); +} + +template +void rand_vector(vec &v, int n, int inc) { + using fp = typename vec::value_type; + int abs_inc = std::abs(inc); + + v.resize(n * abs_inc); + + for (int i = 0; i < n; i++) + v[i * abs_inc] = rand_scalar(); +} + +template +void print_matrix(vec &M, onemkl::transpose trans, int m, int n, int ld, char *name) { + std::cout << "Matrix " << name << ":\n"; + for (int i = 0; i < m; i++) { + for (int j = 0; j < n; j++) { + if (trans == onemkl::transpose::nontrans) + std::cout << (double)M[i + j * ld] << " "; + else + std::cout << (double)M[j + i * ld] << " "; + } + std::cout << std::endl; + } +} + +template +void copy_matrix(vec_src &src, onemkl::transpose trans, int m, int n, int ld, vec_dest &dest) { + using T_data = typename vec_dest::value_type; + dest.resize(matrix_size(trans, m, n, ld)); + if (trans == onemkl::transpose::nontrans) { + for (int j = 0; j < n; j++) + for (int i = 0; i < m; i++) + dest[i + j * ld] = (T_data)src[i + j * ld]; + } + else { + for (int i = 0; i < m; i++) + for (int j = 0; j < n; j++) + dest[j + i * ld] = (T_data)src[j + i * ld]; + } +} + +template +void rand_matrix(vec &M, onemkl::transpose trans, int m, int n, int ld) { + using fp = typename vec::value_type; + + M.resize(matrix_size(trans, m, n, ld)); + + if (trans == onemkl::transpose::nontrans) { + for (int j = 0; j < n; j++) + for (int i = 0; i < m; i++) + M[i + j * ld] = rand_scalar(); + } + else { + for (int i = 0; i < m; i++) + for (int j = 0; j < n; j++) + M[j + i * ld] = rand_scalar(); + } +} + +template +void rand_matrix(fp *M, onemkl::transpose trans, int m, int n, int ld) { + if (trans == onemkl::transpose::nontrans) { + for (int j = 0; j < n; j++) + for (int i = 0; i < m; i++) + M[i + j * ld] = rand_scalar(); + } + else { + for (int i = 0; i < m; i++) + for (int j = 0; j < n; j++) + M[j + i * ld] = rand_scalar(); + } +} + +template +void rand_trsm_matrix(vec &M, onemkl::transpose trans, int m, int n, int ld) { + using fp = typename vec::value_type; + + M.resize(matrix_size(trans, m, n, ld)); + + if (trans == onemkl::transpose::nontrans) { + for (int j = 0; j < n; j++) + for (int i = 0; i < m; i++) { + if (i == j) + M[i + j * ld] = rand_scalar(10); + else + M[i + j * ld] = rand_scalar(); + } + } + else { + for (int i = 0; i < m; i++) + for (int j = 0; j < n; j++) { + if (i == j) + M[j + i * ld] = rand_scalar(10); + else + M[j + i * ld] = rand_scalar(); + } + } +} + +template +void rand_trsm_matrix(fp *M, onemkl::transpose trans, int m, int n, int ld) { + if (trans == onemkl::transpose::nontrans) { + for (int j = 0; j < n; j++) + for (int i = 0; i < m; i++) { + if (i == j) + M[i + j * ld] = rand_scalar(10); + else + M[i + j * ld] = rand_scalar(); + } + } + else { + for (int i = 0; i < m; i++) + for (int j = 0; j < n; j++) { + if (i == j) + M[j + i * ld] = rand_scalar(10); + else + M[j + i * ld] = rand_scalar(); + } + } +} + +// Correctness checking. +template +typename std::enable_if::value, bool>::type check_equal(fp x, fp x_ref, + int error_mag) { + using fp_real = typename complex_info::real_type; + fp_real bound = (error_mag * num_components() * std::numeric_limits::epsilon()); + + bool ok; + + fp_real aerr = std::abs(x - x_ref); + fp_real rerr = aerr / std::abs(x_ref); + ok = (rerr <= bound) || (aerr <= bound); + if (!ok) + std::cout << "relative error = " << rerr << " absolute error = " << aerr + << " limit = " << bound << std::endl; + return ok; +} + +template +typename std::enable_if::value, bool>::type check_equal(fp x, fp x_ref, + int error_mag) { + return (x == x_ref); +} + +template +bool check_equal_trsm(fp x, fp x_ref, int error_mag) { + using fp_real = typename complex_info::real_type; + fp_real bound = std::max(fp_real(5e-5), (error_mag * num_components() * + std::numeric_limits::epsilon())); + bool ok; + + fp_real aerr = std::abs(x - x_ref); + fp_real rerr = aerr / std::abs(x_ref); + ok = (rerr <= bound) || (aerr <= bound); + if (!ok) + std::cout << "relative error = " << rerr << " absolute error = " << aerr + << " limit = " << bound << std::endl; + return ok; +} + +template +bool check_equal(fp x, fp x_ref, int error_mag, std::ostream &out) { + bool good = check_equal(x, x_ref, error_mag); + + if (!good) { + out << "Difference in result: DPC++ " << x << " vs. Reference " << x_ref << std::endl; + } + return good; +} + +template +bool check_equal_vector(vec1 &v, vec2 &v_ref, int n, int inc, int error_mag, std::ostream &out) { + int abs_inc = std::abs(inc); + bool good = true; + + for (int i = 0; i < n; i++) { + if (!check_equal(v[i * abs_inc], v_ref[i * abs_inc], error_mag)) { + int i_actual = (inc > 0) ? i : n - i; + std::cout << "Difference in entry " << i_actual << ": DPC++ " << v[i * abs_inc] + << " vs. Reference " << v_ref[i * abs_inc] << std::endl; + good = false; + } + } + + return good; +} + +template +bool check_equal_trsv_vector(vec1 &v, vec2 &v_ref, int n, int inc, int error_mag, + std::ostream &out) { + int abs_inc = std::abs(inc); + bool good = true; + + for (int i = 0; i < n; i++) { + if (!check_equal_trsm(v[i * abs_inc], v_ref[i * abs_inc], error_mag)) { + int i_actual = (inc > 0) ? i : n - i; + std::cout << "Difference in entry " << i_actual << ": DPC++ " << v[i * abs_inc] + << " vs. Reference " << v_ref[i * abs_inc] << std::endl; + good = false; + } + } + + return good; +} + +template +bool check_equal_matrix(acc1 &M, acc2 &M_ref, int m, int n, int ld, int error_mag, + std::ostream &out) { + bool good = true; + + for (int j = 0; j < n; j++) { + for (int i = 0; i < m; i++) { + if (!check_equal(M[i + j * ld], M_ref[i + j * ld], error_mag)) { + out << "Difference in entry (" << i << ',' << j << "): DPC++ " << M[i + j * ld] + << " vs. Reference " << M_ref[i + j * ld] << std::endl; + good = false; + } + } + } + + return good; +} + +template +bool check_equal_matrix(fp *M, fp *M_ref, int m, int n, int ld, int error_mag, std::ostream &out) { + bool good = true; + + for (int j = 0; j < n; j++) { + for (int i = 0; i < m; i++) { + if (!check_equal(M[i + j * ld], M_ref[i + j * ld], error_mag)) { + out << "Difference in entry (" << i << ',' << j << "): DPC++ " << M[i + j * ld] + << " vs. Reference " << M_ref[i + j * ld] << std::endl; + good = false; + } + } + } + + return good; +} + +template +bool check_equal_trsm_matrix(acc1 &M, acc2 &M_ref, int m, int n, int ld, int error_mag, + std::ostream &out) { + bool good = true; + + for (int j = 0; j < n; j++) { + for (int i = 0; i < m; i++) { + if (!check_equal_trsm(M[i + j * ld], M_ref[i + j * ld], error_mag)) { + out << "Difference in entry (" << i << ',' << j << "): DPC++ " << M[i + j * ld] + << " vs. Reference " << M_ref[i + j * ld] << std::endl; + good = false; + } + } + } + + return good; +} + +#endif /* header guard */ diff --git a/tests/unit_tests/blas/level1/CMakeLists.txt b/tests/unit_tests/blas/level1/CMakeLists.txt new file mode 100644 index 000000000..4b12b33eb --- /dev/null +++ b/tests/unit_tests/blas/level1/CMakeLists.txt @@ -0,0 +1,47 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# Build object from all test sources +set(L1_SOURCES "nrm2.cpp" "iamin.cpp" "iamax.cpp" "dotu.cpp" "dot.cpp" "dotc.cpp" "copy.cpp" "axpy.cpp" "asum.cpp" "swap.cpp" "sdsdot.cpp" "scal.cpp" "rotmg.cpp" "rotm.cpp" "rotg.cpp" "rot.cpp") + +if(BUILD_SHARED_LIBS) + add_library(blas_level1_rt OBJECT ${L1_SOURCES}) + target_compile_options(blas_level1_rt PRIVATE -DCALL_RT_API) + target_include_directories(blas_level1_rt + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../include + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../include + PUBLIC ${PROJECT_SOURCE_DIR}/include + PUBLIC ${PROJECT_SOURCE_DIR}/deps/googletest/include + PUBLIC ${CMAKE_BINARY_DIR}/bin/onemkl + PUBLIC ${CBLAS_INCLUDE} + ) + target_link_libraries(blas_level1_rt PUBLIC ONEMKL::SYCL::SYCL) +endif() + +add_library(blas_level1_ct OBJECT ${L1_SOURCES}) +target_compile_options(blas_level1_ct PRIVATE) +target_include_directories(blas_level1_ct + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../include + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../include + PUBLIC ${PROJECT_SOURCE_DIR}/include + PUBLIC ${PROJECT_SOURCE_DIR}/deps/googletest/include + PUBLIC ${CMAKE_BINARY_DIR}/bin/onemkl + PUBLIC ${CBLAS_INCLUDE} +) +target_link_libraries(blas_level1_ct PUBLIC ONEMKL::SYCL::SYCL) diff --git a/tests/unit_tests/blas/level1/asum.cpp b/tests/unit_tests/blas/level1/asum.cpp new file mode 100644 index 000000000..c180396da --- /dev/null +++ b/tests/unit_tests/blas/level1/asum.cpp @@ -0,0 +1,130 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx) { + // Prepare data. + vector x; + fp_res result = fp_res(-1), result_ref = fp_res(-1); + + rand_vector(x, N, incx); + + // Call Reference ASUM. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = std::abs(incx); + + result_ref = ::asum(&N_ref, (fp_ref*)x.data(), &incx_ref); + // Call DPC++ ASUM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during ASUM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::asum(main_queue, N, x_buffer, incx, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::asum, (main_queue, N, x_buffer, incx, result_buffer)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during ASUM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_ref, N, std::cout); + } + + return good; +} + +class AsumTests : public ::testing::TestWithParam {}; + +TEST_P(AsumTests, RealSinglePrecision) { + EXPECT_TRUE((::test(GetParam(), 1357, 2))); + EXPECT_TRUE((::test(GetParam(), 1357, 1))); + EXPECT_TRUE((::test(GetParam(), 1357, -3))); +} + +TEST_P(AsumTests, RealDoublePrecision) { + EXPECT_TRUE((::test(GetParam(), 1357, 2))); + EXPECT_TRUE((::test(GetParam(), 1357, 1))); + EXPECT_TRUE((::test(GetParam(), 1357, -3))); +} + +TEST_P(AsumTests, ComplexSinglePrecision) { + EXPECT_TRUE((::test, float>(GetParam(), 1357, 2))); + EXPECT_TRUE((::test, float>(GetParam(), 1357, 1))); + EXPECT_TRUE((::test, float>(GetParam(), 1357, -3))); +} + +TEST_P(AsumTests, ComplexDoublePrecision) { + EXPECT_TRUE((test, double>(GetParam(), 1357, 2))); + EXPECT_TRUE((test, double>(GetParam(), 1357, 1))); + EXPECT_TRUE((test, double>(GetParam(), 1357, -3))); +} + +INSTANTIATE_TEST_SUITE_P(AsumTestSuite, AsumTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/axpy.cpp b/tests/unit_tests/blas/level1/axpy.cpp new file mode 100644 index 000000000..e289aabfa --- /dev/null +++ b/tests/unit_tests/blas/level1/axpy.cpp @@ -0,0 +1,135 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int N, int incx, int incy, fp alpha) { + // Prepare data. + vector x, y, y_ref; + + rand_vector(x, N, incx); + rand_vector(y, N, incy); + y_ref = y; + + // Call Reference AXPY. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::axpy(&N_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), &incx_ref, (fp_ref *)y_ref.data(), + &incy_ref); + + // Call DPC++ AXPY. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during AXPY:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + + try { +#ifdef CALL_RT_API + onemkl::blas::axpy(main_queue, N, alpha, x_buffer, incx, y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::axpy, + (main_queue, N, alpha, x_buffer, incx, y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during AXPY:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, N, incy, N, std::cout); + } + + return good; +} + +class AxpyTests : public ::testing::TestWithParam {}; + +TEST_P(AxpyTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, alpha)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, alpha)); + EXPECT_TRUE(test(GetParam(), 1357, -3, -2, alpha)); +} +TEST_P(AxpyTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, alpha)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, alpha)); + EXPECT_TRUE(test(GetParam(), 1357, -3, -2, alpha)); +} +TEST_P(AxpyTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3, alpha)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1, alpha)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2, alpha)); +} +TEST_P(AxpyTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3, alpha)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1, alpha)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2, alpha)); +} + +INSTANTIATE_TEST_SUITE_P(AxpyTestSuite, AxpyTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/copy.cpp b/tests/unit_tests/blas/level1/copy.cpp new file mode 100644 index 000000000..33eafa26f --- /dev/null +++ b/tests/unit_tests/blas/level1/copy.cpp @@ -0,0 +1,130 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx, int incy) { + // Prepare data. + vector x, y, y_ref; + + rand_vector(x, N, incx); + rand_vector(y, N, incy); + y_ref = y; + + // Call Reference COPY. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::copy(&N_ref, (fp_ref*)x.data(), &incx_ref, (fp_ref*)y_ref.data(), &incy_ref); + + // Call DPC++ COPY. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during COPY:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + + try { +#ifdef CALL_RT_API + onemkl::blas::copy(main_queue, N, x_buffer, incx, y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::copy, + (main_queue, N, x_buffer, incx, y_buffer, incy)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during COPY:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, N, incy, N, std::cout); + } + + return good; +} + +class CopyTests : public ::testing::TestWithParam {}; + +TEST_P(CopyTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test(GetParam(), 1357, -3, -2)); +} +TEST_P(CopyTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test(GetParam(), 1357, -3, -2)); +} +TEST_P(CopyTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2)); +} +TEST_P(CopyTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2)); +} + +INSTANTIATE_TEST_SUITE_P(CopyTestSuite, CopyTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/dot.cpp b/tests/unit_tests/blas/level1/dot.cpp new file mode 100644 index 000000000..46d3940d0 --- /dev/null +++ b/tests/unit_tests/blas/level1/dot.cpp @@ -0,0 +1,124 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx, int incy) { + // Prepare data. + vector x, y; + fp_res result = fp_res(-1), result_ref = fp_res(-1); + + rand_vector(x, N, incx); + rand_vector(y, N, incy); + + // Call Reference DOT. + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + result_ref = ::dot(&N_ref, (fp*)x.data(), &incx_ref, (fp*)y.data(), &incy_ref); + + // Call DPC++ DOT. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during DOT:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::dot(main_queue, N, x_buffer, incx, y_buffer, incy, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::dot, + (main_queue, N, x_buffer, incx, y_buffer, incy, result_buffer)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during DOT:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_ref, N, std::cout); + } + + return good; +} + +class DotTests : public ::testing::TestWithParam {}; + +TEST_P(DotTests, RealSinglePrecision) { + EXPECT_TRUE((test(GetParam(), 1357, 2, 3))); + EXPECT_TRUE((test(GetParam(), 1357, 1, 1))); + EXPECT_TRUE((test(GetParam(), 1357, -3, -2))); +} +TEST_P(DotTests, RealDoublePrecision) { + EXPECT_TRUE((test(GetParam(), 1357, 2, 3))); + EXPECT_TRUE((test(GetParam(), 1357, 1, 1))); + EXPECT_TRUE((test(GetParam(), 1357, -3, -2))); +} +//TEST_P(DotTests, RealDoubleSinglePrecision) { +// EXPECT_TRUE((test(GetParam(), 1357, 2, 3))); +// EXPECT_TRUE((test(GetParam(), 1357, 1, 1))); +// EXPECT_TRUE((test(GetParam(), 1357, -3, -2))); +//} + +INSTANTIATE_TEST_SUITE_P(DotTestSuite, DotTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/dotc.cpp b/tests/unit_tests/blas/level1/dotc.cpp new file mode 100644 index 000000000..97f033cc4 --- /dev/null +++ b/tests/unit_tests/blas/level1/dotc.cpp @@ -0,0 +1,122 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int N, int incx, int incy) { + // Prepare data. + vector x, y; + fp result = 0.0, result_reference = 0.0; + + rand_vector(x, N, incx); + rand_vector(y, N, incy); + + // Call Reference DOTC. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::dotc((fp_ref *)&result_reference, &N_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)y.data(), + &incy_ref); + + // Call DPC++ DOTC. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during DOTC:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::dotc(main_queue, N, x_buffer, incx, y_buffer, incy, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::dotc, + (main_queue, N, x_buffer, incx, y_buffer, incy, result_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during DOTC:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_reference, N, std::cout); + } + + return good; +} + +class DotcTests : public ::testing::TestWithParam {}; + +TEST_P(DotcTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2)); +} +TEST_P(DotcTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2)); +} + +INSTANTIATE_TEST_SUITE_P(DotcTestSuite, DotcTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/dotu.cpp b/tests/unit_tests/blas/level1/dotu.cpp new file mode 100644 index 000000000..19eb49cd1 --- /dev/null +++ b/tests/unit_tests/blas/level1/dotu.cpp @@ -0,0 +1,122 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int N, int incx, int incy) { + // Prepare data. + vector x, y; + fp result = 0.0, result_reference = 0.0; + + rand_vector(x, N, incx); + rand_vector(y, N, incy); + + // Call Reference DOTU. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::dotu((fp_ref *)&result_reference, &N_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)y.data(), + &incy_ref); + + // Call DPC++ DOTU. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during DOTU:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::dotu(main_queue, N, x_buffer, incx, y_buffer, incy, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::dotu, + (main_queue, N, x_buffer, incx, y_buffer, incy, result_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during DOTU:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_reference, N, std::cout); + } + + return good; +} + +class DotuTests : public ::testing::TestWithParam {}; + +TEST_P(DotuTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2)); +} +TEST_P(DotuTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3, -2)); +} + +INSTANTIATE_TEST_SUITE_P(DotuTestSuite, DotuTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/iamax.cpp b/tests/unit_tests/blas/level1/iamax.cpp new file mode 100644 index 000000000..985beced4 --- /dev/null +++ b/tests/unit_tests/blas/level1/iamax.cpp @@ -0,0 +1,128 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx) { + // Prepare data. + vector x; + int64_t result = -1, result_ref = -1; + rand_vector(x, N, incx); + + // Call Reference IAMAX. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx; + + result_ref = ::iamax(&N_ref, (fp_ref*)x.data(), &incx_ref); + + // Call DPC++ IAMAX. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during IAMAX:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::iamax(main_queue, N, x_buffer, incx, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::iamax, + (main_queue, N, x_buffer, incx, result_buffer)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during IAMAX:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_ref, 0, std::cout); + } + + return good; +} + +class IamaxTests : public ::testing::TestWithParam {}; + +TEST_P(IamaxTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2)); + EXPECT_TRUE(test(GetParam(), 1357, 1)); + EXPECT_TRUE(test(GetParam(), 1357, -3)); +} +TEST_P(IamaxTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2)); + EXPECT_TRUE(test(GetParam(), 1357, 1)); + EXPECT_TRUE(test(GetParam(), 1357, -3)); +} +TEST_P(IamaxTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2)); + EXPECT_TRUE(test>(GetParam(), 1357, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3)); +} +TEST_P(IamaxTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2)); + EXPECT_TRUE(test>(GetParam(), 1357, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3)); +} + +INSTANTIATE_TEST_SUITE_P(IamaxTestSuite, IamaxTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/iamin.cpp b/tests/unit_tests/blas/level1/iamin.cpp new file mode 100644 index 000000000..5a5ad9071 --- /dev/null +++ b/tests/unit_tests/blas/level1/iamin.cpp @@ -0,0 +1,128 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx) { + // Prepare data. + vector x; + int64_t result = -1, result_ref = -1; + rand_vector(x, N, incx); + + // Call Reference IAMIN. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx; + + result_ref = ::iamin(&N_ref, (fp_ref*)x.data(), &incx_ref); + + // Call DPC++ IAMIN. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during IAMIN:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::iamin(main_queue, N, x_buffer, incx, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::iamin, + (main_queue, N, x_buffer, incx, result_buffer)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during IAMIN:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_ref, 0, std::cout); + } + + return good; +} + +class IaminTests : public ::testing::TestWithParam {}; + +TEST_P(IaminTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2)); + EXPECT_TRUE(test(GetParam(), 1357, 1)); + EXPECT_TRUE(test(GetParam(), 1357, -3)); +} +TEST_P(IaminTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2)); + EXPECT_TRUE(test(GetParam(), 1357, 1)); + EXPECT_TRUE(test(GetParam(), 1357, -3)); +} +TEST_P(IaminTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2)); + EXPECT_TRUE(test>(GetParam(), 1357, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3)); +} +TEST_P(IaminTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2)); + EXPECT_TRUE(test>(GetParam(), 1357, 1)); + EXPECT_TRUE(test>(GetParam(), 1357, -3)); +} + +INSTANTIATE_TEST_SUITE_P(IaminTestSuite, IaminTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/nrm2.cpp b/tests/unit_tests/blas/level1/nrm2.cpp new file mode 100644 index 000000000..8f07953f5 --- /dev/null +++ b/tests/unit_tests/blas/level1/nrm2.cpp @@ -0,0 +1,128 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx) { + // Prepare data. + vector x; + fp_res result = fp_res(-1), result_ref = fp_res(-1); + + rand_vector(x, N, incx); + + // Call Reference NRM2. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = std::abs(incx); + + result_ref = ::nrm2(&N_ref, (fp_ref*)x.data(), &incx_ref); + + // Call DPC++ NRM2. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during NRM2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::nrm2(main_queue, N, x_buffer, incx, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::nrm2, (main_queue, N, x_buffer, incx, result_buffer)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during NRM2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_ref, N, std::cout); + } + + return good; +} + +class Nrm2Tests : public ::testing::TestWithParam {}; + +TEST_P(Nrm2Tests, RealSinglePrecision) { + EXPECT_TRUE((test(GetParam(), 1357, 2))); + EXPECT_TRUE((test(GetParam(), 1357, 1))); + EXPECT_TRUE((test(GetParam(), 1357, -3))); +} +TEST_P(Nrm2Tests, RealDoublePrecision) { + EXPECT_TRUE((test(GetParam(), 1357, 2))); + EXPECT_TRUE((test(GetParam(), 1357, 1))); + EXPECT_TRUE((test(GetParam(), 1357, -3))); +} +TEST_P(Nrm2Tests, ComplexSinglePrecision) { + EXPECT_TRUE((test, float>(GetParam(), 1357, 2))); + EXPECT_TRUE((test, float>(GetParam(), 1357, 1))); + EXPECT_TRUE((test, float>(GetParam(), 1357, -3))); +} +TEST_P(Nrm2Tests, ComplexDoublePrecision) { + EXPECT_TRUE((test, double>(GetParam(), 1357, 2))); + EXPECT_TRUE((test, double>(GetParam(), 1357, 1))); + EXPECT_TRUE((test, double>(GetParam(), 1357, -3))); +} + +INSTANTIATE_TEST_SUITE_P(Nrm2TestSuite, Nrm2Tests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/rot.cpp b/tests/unit_tests/blas/level1/rot.cpp new file mode 100644 index 000000000..a00463d9d --- /dev/null +++ b/tests/unit_tests/blas/level1/rot.cpp @@ -0,0 +1,141 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int N, int incx, int incy, fp_scalar c, fp_scalar s) { + // Prepare data. + vector x, x_ref, y, y_ref; + rand_vector(x, N, incx); + rand_vector(y, N, incy); + y_ref = y; + x_ref = x; + + // Call Reference ROT. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::rot(&N_ref, (fp_ref *)x_ref.data(), &incx_ref, (fp_ref *)y_ref.data(), &incy_ref, + (fp_scalar *)&c, (fp_scalar *)&s); + + // Call DPC++ ROT. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during ROT:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + + try { +#ifdef CALL_RT_API + onemkl::blas::rot(main_queue, N, x_buffer, incx, y_buffer, incy, c, s); +#else + TEST_RUN_CT(main_queue, onemkl::blas::rot, + (main_queue, N, x_buffer, incx, y_buffer, incy, c, s)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during ROT:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + bool good_x = check_equal_vector(x_accessor, x_ref, N, incx, N, std::cout); + auto y_accessor = y_buffer.template get_access(); + bool good_y = check_equal_vector(y_accessor, y_ref, N, incy, N, std::cout); + good = good_x && good_y; + } + + return good; +} + +class RotTests : public ::testing::TestWithParam {}; + +TEST_P(RotTests, RealSinglePrecision) { + float c(2.0); + float s(-0.5); + EXPECT_TRUE((test(GetParam(), 1357, 2, 3, c, s))); + EXPECT_TRUE((test(GetParam(), 1357, 1, 1, c, s))); + EXPECT_TRUE((test(GetParam(), 1357, -2, -3, c, s))); +} +TEST_P(RotTests, RealDoublePrecision) { + double c(2.0); + double s(-0.5); + EXPECT_TRUE((test(GetParam(), 1357, 2, 3, c, s))); + EXPECT_TRUE((test(GetParam(), 1357, 1, 1, c, s))); + EXPECT_TRUE((test(GetParam(), 1357, -2, -3, c, s))); +} +TEST_P(RotTests, ComplexSinglePrecision) { + float c = 2.0; + float s = -0.5; + EXPECT_TRUE((test, float>(GetParam(), 1357, 2, 3, c, s))); + EXPECT_TRUE((test, float>(GetParam(), 1357, 1, 1, c, s))); + EXPECT_TRUE((test, float>(GetParam(), 1357, -2, -3, c, s))); +} +TEST_P(RotTests, ComplexDoublePrecision) { + double c = 2.0; + double s = -0.5; + EXPECT_TRUE((test, double>(GetParam(), 1357, 2, 3, c, s))); + EXPECT_TRUE((test, double>(GetParam(), 1357, 1, 1, c, s))); + EXPECT_TRUE((test, double>(GetParam(), 1357, -2, -3, c, s))); +} + +INSTANTIATE_TEST_SUITE_P(RotTestSuite, RotTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/rotg.cpp b/tests/unit_tests/blas/level1/rotg.cpp new file mode 100644 index 000000000..ae6cc1a10 --- /dev/null +++ b/tests/unit_tests/blas/level1/rotg.cpp @@ -0,0 +1,149 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev) { + // Prepare data. + fp a, b, s, a_ref, b_ref, s_ref; + fp_scalar c, c_ref; + + a = rand_scalar(); + b = rand_scalar(); + a_ref = a; + b_ref = b; + + // Call Reference ROTG. + using fp_ref = typename ref_type_info::type; + + ::rotg((fp_ref *)&a_ref, (fp_ref *)&b_ref, (fp_scalar *)&c_ref, (fp_ref *)&s_ref); + + // Call DPC++ ROTG. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during ROTG:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer a_buffer(&a, range<1>(1)); + buffer b_buffer(&b, range<1>(1)); + buffer c_buffer(&c, range<1>(1)); + buffer s_buffer(&s, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::rotg(main_queue, a_buffer, b_buffer, c_buffer, s_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::rotg, + (main_queue, a_buffer, b_buffer, c_buffer, s_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during ROTG:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto a_accessor = a_buffer.template get_access(); + bool good_a = check_equal(a_accessor[0], a_ref, 4, std::cout); + auto b_accessor = b_buffer.template get_access(); + bool good_b = check_equal(b_accessor[0], b_ref, 4, std::cout); + auto s_accessor = s_buffer.template get_access(); + bool good_s = check_equal(s_accessor[0], s_ref, 4, std::cout); + auto c_accessor = c_buffer.template get_access(); + bool good_c = check_equal(c_accessor[0], c_ref, 4, std::cout); + + good = good_a && good_b && good_c && good_s; + } + + return good; +} + +class RotgTests : public ::testing::TestWithParam {}; + +TEST_P(RotgTests, RealSinglePrecision) { + float c(2.0); + float s(-0.5); + EXPECT_TRUE((test(GetParam()))); + EXPECT_TRUE((test(GetParam()))); + EXPECT_TRUE((test(GetParam()))); +} +TEST_P(RotgTests, RealDoublePrecision) { + double c(2.0); + double s(-0.5); + EXPECT_TRUE((test(GetParam()))); + EXPECT_TRUE((test(GetParam()))); + EXPECT_TRUE((test(GetParam()))); +} +TEST_P(RotgTests, ComplexSinglePrecision) { + float c = 2.0; + float s = -0.5; + EXPECT_TRUE((test, float>(GetParam()))); + EXPECT_TRUE((test, float>(GetParam()))); + EXPECT_TRUE((test, float>(GetParam()))); +} +TEST_P(RotgTests, ComplexDoublePrecision) { + double c = 2.0; + double s = -0.5; + EXPECT_TRUE((test, double>(GetParam()))); + EXPECT_TRUE((test, double>(GetParam()))); + EXPECT_TRUE((test, double>(GetParam()))); +} + +INSTANTIATE_TEST_SUITE_P(RotgTestSuite, RotgTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/rotm.cpp b/tests/unit_tests/blas/level1/rotm.cpp new file mode 100644 index 000000000..08c791769 --- /dev/null +++ b/tests/unit_tests/blas/level1/rotm.cpp @@ -0,0 +1,154 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int N, int incx, int incy, fp flag) { + // Prepare data. + vector x, x_ref, y, y_ref; + vector param; + rand_vector(x, N, incx); + rand_vector(y, N, incy); + rand_vector(param, 5, 1); + param[0] = flag; + y_ref = y; + x_ref = x; + + // Call Reference ROTM. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::rotm(&N_ref, (fp_ref *)x_ref.data(), &incx_ref, (fp_ref *)y_ref.data(), &incy_ref, + (fp_ref *)param.data()); + + // Call DPC++ ROTM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during ROTM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer param_buffer = make_buffer(param); + + try { +#ifdef CALL_RT_API + onemkl::blas::rotm(main_queue, N, x_buffer, incx, y_buffer, incy, param_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::rotm, + (main_queue, N, x_buffer, incx, y_buffer, incy, param_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during ROTM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + bool good_x = check_equal_vector(x_accessor, x_ref, N, incx, N, std::cout); + auto y_accessor = y_buffer.template get_access(); + bool good_y = check_equal_vector(y_accessor, y_ref, N, incy, N, std::cout); + good = good_x && good_y; + } + + return good; +} + +class RotmTests : public ::testing::TestWithParam {}; + +TEST_P(RotmTests, RealSinglePrecision) { + float flag(-1.0); + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); + flag = 0.0; + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); + flag = 1.0; + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); + flag = -2.0; + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); +} +TEST_P(RotmTests, RealDoublePrecision) { + double flag(-1.0); + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); + flag = 0.0; + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); + flag = 1.0; + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); + flag = -2.0; + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, flag)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, flag)); +} + +INSTANTIATE_TEST_SUITE_P(RotmTestSuite, RotmTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/rotmg.cpp b/tests/unit_tests/blas/level1/rotmg.cpp new file mode 100644 index 000000000..82a61eb2e --- /dev/null +++ b/tests/unit_tests/blas/level1/rotmg.cpp @@ -0,0 +1,128 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev) { + // Prepare data. + fp d1, d2, x1, y1, d1_ref, d2_ref, x1_ref; + vector param(5, fp(0)), param_ref(5, fp(0)); + + d1 = rand_scalar(); + d1 = abs(d1); + d2 = rand_scalar(); + x1 = rand_scalar(); + y1 = rand_scalar(); + d1_ref = d1; + d2_ref = d2; + x1_ref = x1; + + // Call Reference ROTMG. + + ::rotmg(&d1_ref, &d2_ref, &x1_ref, &y1, (fp*)param_ref.data()); + + // Call DPC++ ROTMG. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during ROTMG:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer d1_buffer(&d1, range<1>(1)); + buffer d2_buffer(&d2, range<1>(1)); + buffer x1_buffer(&x1, range<1>(1)); + buffer param_buffer = make_buffer(param); + try { +#ifdef CALL_RT_API + onemkl::blas::rotmg(main_queue, d1_buffer, d2_buffer, x1_buffer, y1, param_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::rotmg, + (main_queue, d1_buffer, d2_buffer, x1_buffer, y1, param_buffer)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during ROTMG:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto d1_accessor = d1_buffer.template get_access(); + bool good_d1 = check_equal(d1_accessor[0], d1_ref, 1, std::cout); + auto d2_accessor = d2_buffer.template get_access(); + bool good_d2 = check_equal(d2_accessor[0], d2_ref, 1, std::cout); + auto x1_accessor = x1_buffer.template get_access(); + bool good_x1 = check_equal(x1_accessor[0], x1_ref, 1, std::cout); + auto param_accessor = param_buffer.template get_access(); + bool good_param = check_equal_vector(param_accessor, param_ref, 5, 1, 1, std::cout); + good = good_d1 && good_d2 && good_x1 && good_param; + } + + return good; +} + +class RotmgTests : public ::testing::TestWithParam {}; + +TEST_P(RotmgTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam())); +} +TEST_P(RotmgTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam())); +} + +INSTANTIATE_TEST_SUITE_P(RotmgTestSuite, RotmgTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/scal.cpp b/tests/unit_tests/blas/level1/scal.cpp new file mode 100644 index 000000000..238bb33da --- /dev/null +++ b/tests/unit_tests/blas/level1/scal.cpp @@ -0,0 +1,139 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx, fp_scalar alpha) { + // Prepare data. + vector x, x_ref; + + rand_vector(x, N, incx); + x_ref = x; + + // Call Reference SCAL. + using fp_ref = typename ref_type_info::type; + using fp_scalar_mkl = typename ref_type_info::type; + + const int N_ref = N, incx_ref = std::abs(incx); + + ::scal(&N_ref, (fp_scalar_mkl*)&alpha, (fp_ref*)x_ref.data(), &incx_ref); + + // Call DPC++ SCAL. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during SCAL:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + + try { +#ifdef CALL_RT_API + onemkl::blas::scal(main_queue, N, alpha, x_buffer, incx); +#else + TEST_RUN_CT(main_queue, onemkl::blas::scal, (main_queue, N, alpha, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during SCAL:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_vector(x_accessor, x_ref, N, incx, N, std::cout); + } + + return good; +} + +class ScalTests : public ::testing::TestWithParam {}; + +TEST_P(ScalTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE((test(GetParam(), 1357, 2, alpha))); + EXPECT_TRUE((test(GetParam(), 1357, -3, alpha))); +} +TEST_P(ScalTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE((test(GetParam(), 1357, 2, alpha))); + EXPECT_TRUE((test(GetParam(), 1357, -3, alpha))); +} +TEST_P(ScalTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE((test, std::complex>(GetParam(), 1357, 2, alpha))); + EXPECT_TRUE((test, std::complex>(GetParam(), 1357, -3, alpha))); +} +TEST_P(ScalTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE((test, std::complex>(GetParam(), 1357, 2, alpha))); + EXPECT_TRUE((test, std::complex>(GetParam(), 1357, -3, alpha))); +} +TEST_P(ScalTests, ComplexRealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE((test, float>(GetParam(), 1357, 2, alpha))); + EXPECT_TRUE((test, float>(GetParam(), 1357, -3, alpha))); +} +TEST_P(ScalTests, ComplexRealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE((test, double>(GetParam(), 1357, 2, alpha))); + EXPECT_TRUE((test, double>(GetParam(), 1357, -3, alpha))); +} + +INSTANTIATE_TEST_SUITE_P(ScalTestSuite, ScalTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/sdsdot.cpp b/tests/unit_tests/blas/level1/sdsdot.cpp new file mode 100644 index 000000000..1ad37b0a1 --- /dev/null +++ b/tests/unit_tests/blas/level1/sdsdot.cpp @@ -0,0 +1,115 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +bool test(const device &dev, int N, int incx, int incy, float alpha) { + // Prepare data. + vector x, y; + float result = float(-1), result_ref = float(-1); + + rand_vector(x, N, incx); + rand_vector(y, N, incy); + + // Call Reference SDSDOT. + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + result_ref = ::sdsdot(&N_ref, (float *)&alpha, (float *)x.data(), &incx_ref, (float *)y.data(), + &incy_ref); + + // Call DPC++ SDSDOT. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SDSDOT:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer result_buffer(&result, range<1>(1)); + + try { +#ifdef CALL_RT_API + onemkl::blas::sdsdot(main_queue, N, alpha, x_buffer, incx, y_buffer, incy, result_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::sdsdot, + (main_queue, N, alpha, x_buffer, incx, y_buffer, incy, result_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SDSDOT:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto result_accessor = result_buffer.template get_access(); + good = check_equal(result_accessor[0], result_ref, N, std::cout); + } + + return good; +} + +class SdsdotTests : public ::testing::TestWithParam {}; + +TEST_P(SdsdotTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2, 3, 2.0)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3, 2.0)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1, 2.0)); +} + +INSTANTIATE_TEST_SUITE_P(SdsdotTestSuite, SdsdotTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level1/swap.cpp b/tests/unit_tests/blas/level1/swap.cpp new file mode 100644 index 000000000..8e47ee819 --- /dev/null +++ b/tests/unit_tests/blas/level1/swap.cpp @@ -0,0 +1,133 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, int N, int incx, int incy) { + // Prepare data. + vector x, x_ref, y, y_ref; + rand_vector(x, N, incx); + rand_vector(y, N, incy); + y_ref = y; + x_ref = x; + + // Call Reference SWAP. + using fp_ref = typename ref_type_info::type; + const int N_ref = N, incx_ref = incx, incy_ref = incy; + + ::swap(&N_ref, (fp_ref*)x_ref.data(), &incx_ref, (fp_ref*)y_ref.data(), &incy_ref); + + // Call DPC++ SWAP. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during SWAP:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + + try { +#ifdef CALL_RT_API + onemkl::blas::swap(main_queue, N, x_buffer, incx, y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::swap, + (main_queue, N, x_buffer, incx, y_buffer, incy)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during SWAP:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + auto x_accessor = x_buffer.template get_access(); + bool good_y = check_equal_vector(y_accessor, y_ref, N, incy, N, std::cout); + bool good_x = check_equal_vector(x_accessor, x_ref, N, incx, N, std::cout); + good = good_x && good_y; + } + + return good; +} + +class SwapTests : public ::testing::TestWithParam {}; + +TEST_P(SwapTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1)); +} +TEST_P(SwapTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test(GetParam(), 1357, -2, -3)); + EXPECT_TRUE(test(GetParam(), 1357, 1, 1)); +} +TEST_P(SwapTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, -2, -3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); +} +TEST_P(SwapTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), 1357, 2, 3)); + EXPECT_TRUE(test>(GetParam(), 1357, -2, -3)); + EXPECT_TRUE(test>(GetParam(), 1357, 1, 1)); +} + +INSTANTIATE_TEST_SUITE_P(SwapTestSuite, SwapTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/CMakeLists.txt b/tests/unit_tests/blas/level2/CMakeLists.txt new file mode 100644 index 000000000..271fbb7de --- /dev/null +++ b/tests/unit_tests/blas/level2/CMakeLists.txt @@ -0,0 +1,48 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# Build object from all test sources +set(L2_SOURCES "hpr2.cpp" "hpmv.cpp" "her.cpp" "her2.cpp" "hemv.cpp" "hbmv.cpp" "geru.cpp" "ger.cpp" "gerc.cpp" "gemv.cpp" "gbmv.cpp" "trsv.cpp" "trmv.cpp" "tpsv.cpp" "tpmv.cpp" "tbsv.cpp" "tbmv.cpp" "syr.cpp" "syr2.cpp" "symv.cpp" "spr.cpp" "spr2.cpp" "spmv.cpp" "sbmv.cpp" "hpr.cpp") + +if(BUILD_SHARED_LIBS) + add_library(blas_level2_rt OBJECT ${L2_SOURCES}) + target_compile_options(blas_level2_rt PRIVATE -DCALL_RT_API) + target_include_directories(blas_level2_rt + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../include + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../include + PUBLIC ${PROJECT_SOURCE_DIR}/include + PUBLIC ${PROJECT_SOURCE_DIR}/deps/googletest/include + PUBLIC ${CMAKE_BINARY_DIR}/bin/onemkl + PUBLIC ${CBLAS_INCLUDE} + ) + target_link_libraries(blas_level2_rt PUBLIC ONEMKL::SYCL::SYCL) +endif() + +add_library(blas_level2_ct OBJECT ${L2_SOURCES}) +target_compile_options(blas_level2_ct PRIVATE) +target_include_directories(blas_level2_ct + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../include + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../include + PUBLIC ${PROJECT_SOURCE_DIR}/include + PUBLIC ${PROJECT_SOURCE_DIR}/deps/googletest/include + PUBLIC ${CMAKE_BINARY_DIR}/bin/onemkl + PUBLIC ${CBLAS_INCLUDE} +) +target_link_libraries(blas_level2_ct PUBLIC ONEMKL::SYCL::SYCL) + diff --git a/tests/unit_tests/blas/level2/gbmv.cpp b/tests/unit_tests/blas/level2/gbmv.cpp new file mode 100644 index 000000000..d0a7c0a19 --- /dev/null +++ b/tests/unit_tests/blas/level2/gbmv.cpp @@ -0,0 +1,198 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::transpose transa, int m, int n, int kl, int ku, fp alpha, + fp beta, int incx, int incy, int lda) { + // Prepare data. + int x_len = outer_dimension(transa, m, n); + int y_len = inner_dimension(transa, m, n); + + vector x, y, y_ref, A; + + rand_vector(x, x_len, incx); + rand_vector(y, y_len, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, m, n, lda); + + // Call Reference GBMV. + const int m_ref = m, n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + int kl_ref = kl, ku_ref = ku; + using fp_ref = typename ref_type_info::type; + + ::gbmv(convert_to_cblas_trans(transa), &m_ref, &n_ref, &kl_ref, &ku_ref, (fp_ref *)&alpha, + (fp_ref *)A.data(), &lda_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, + (fp_ref *)y_ref.data(), &incy_ref); + + // Call DPC++ GBMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during GBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::gbmv(main_queue, transa, m, n, kl, ku, alpha, A_buffer, lda, x_buffer, incx, + beta, y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::gbmv, + (main_queue, transa, m, n, kl, ku, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during GBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, y_len, incy, std::max(m, n), std::cout); + } + + return good; +} + +class GbmvTests : public ::testing::TestWithParam {}; + +TEST_P(GbmvTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, alpha, beta, -2, + -3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, alpha, beta, 1, 1, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, beta, 1, 1, 42)); +} +TEST_P(GbmvTests, RealDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, alpha, beta, -2, + -3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, alpha, beta, 1, 1, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, beta, 1, 1, 42)); +} +TEST_P(GbmvTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, + alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, + alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, + alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, + beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, + beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, alpha, + beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, 5, 7, + alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, 5, 7, + alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, 5, 7, + alpha, beta, 1, 1, 42)); +} +TEST_P(GbmvTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, + alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, + alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, 5, 7, + alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, + alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, + alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, 5, 7, + alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, 5, 7, + alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, 5, 7, + alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, 5, 7, + alpha, beta, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(GbmvTestSuite, GbmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/gemv.cpp b/tests/unit_tests/blas/level2/gemv.cpp new file mode 100644 index 000000000..48ab54fcc --- /dev/null +++ b/tests/unit_tests/blas/level2/gemv.cpp @@ -0,0 +1,192 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::transpose transa, int m, int n, fp alpha, fp beta, int incx, + int incy, int lda) { + // Prepare data. + int x_len = outer_dimension(transa, m, n); + int y_len = inner_dimension(transa, m, n); + + vector x, y, y_ref, A; + + rand_vector(x, x_len, incx); + rand_vector(y, y_len, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, m, n, lda); + + // Call Reference GEMV. + const int m_ref = m, n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::gemv(convert_to_cblas_trans(transa), &m_ref, &n_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + &lda_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), + &incy_ref); + + // Call DPC++ GEMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during GEMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::gemv(main_queue, transa, m, n, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::gemv, + (main_queue, transa, m, n, alpha, A_buffer, lda, x_buffer, incx, beta, y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during GEMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, y_len, incy, std::max(m, n), std::cout); + } + + return good; +} + +class GemvTests : public ::testing::TestWithParam {}; + +TEST_P(GemvTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, 1, 1, 42)); +} +TEST_P(GemvTests, RealDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, 1, 1, 42)); +} +TEST_P(GemvTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, + beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, + beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, + beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, + 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, + -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, alpha, beta, + 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, alpha, + beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, alpha, + beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, alpha, + beta, 1, 1, 42)); +} +TEST_P(GemvTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, + beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, + beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, 25, 30, alpha, + beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, alpha, + beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, alpha, + beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, 25, 30, alpha, + beta, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, alpha, + beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, alpha, + beta, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, 25, 30, alpha, + beta, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(GemvTestSuite, GemvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/ger.cpp b/tests/unit_tests/blas/level2/ger.cpp new file mode 100644 index 000000000..0572ecf36 --- /dev/null +++ b/tests/unit_tests/blas/level2/ger.cpp @@ -0,0 +1,126 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int m, int n, fp alpha, int incx, int incy, int lda) { + // Prepare data. + + vector x, y, A_ref, A; + + rand_vector(x, m, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, m, n, lda); + A_ref = A; + + // Call Reference GER. + const int m_ref = m, n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::ger(&m_ref, &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), &incx_ref, (fp_ref *)y.data(), + &incy_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ GER. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during GER:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::ger(main_queue, m, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda); +#else + TEST_RUN_CT(main_queue, onemkl::blas::ger, + (main_queue, m, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during GER:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, m, n, lda, std::max(m, n), std::cout); + } + + return good; +} + +class GerTests : public ::testing::TestWithParam {}; + +TEST_P(GerTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), 25, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), 25, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), 25, 30, alpha, 1, 1, 42)); +} +TEST_P(GerTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), 25, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), 25, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), 25, 30, alpha, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(GerTestSuite, GerTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/gerc.cpp b/tests/unit_tests/blas/level2/gerc.cpp new file mode 100644 index 000000000..6951974c7 --- /dev/null +++ b/tests/unit_tests/blas/level2/gerc.cpp @@ -0,0 +1,127 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int m, int n, fp alpha, int incx, int incy, int lda) { + // Prepare data. + + vector x, y, A_ref, A; + + rand_vector(x, m, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, m, n, lda); + A_ref = A; + + // Call Reference GERC. + const int m_ref = m, n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::gerc(&m_ref, &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), &incx_ref, (fp_ref *)y.data(), + &incy_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ GERC. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during GERC:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::gerc(main_queue, m, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda); +#else + TEST_RUN_CT(main_queue, onemkl::blas::gerc, + (main_queue, m, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during GERC:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, m, n, lda, std::max(m, n), std::cout); + } + + return good; +} + +class GercTests : public ::testing::TestWithParam {}; + +TEST_P(GercTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 1, 1, 42)); +} +TEST_P(GercTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(GercTestSuite, GercTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/geru.cpp b/tests/unit_tests/blas/level2/geru.cpp new file mode 100644 index 000000000..c169c474e --- /dev/null +++ b/tests/unit_tests/blas/level2/geru.cpp @@ -0,0 +1,127 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, int m, int n, fp alpha, int incx, int incy, int lda) { + // Prepare data. + + vector x, y, A_ref, A; + + rand_vector(x, m, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, m, n, lda); + A_ref = A; + + // Call Reference GERU. + const int m_ref = m, n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::geru(&m_ref, &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), &incx_ref, (fp_ref *)y.data(), + &incy_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ GERU. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during GERU:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::geru(main_queue, m, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda); +#else + TEST_RUN_CT(main_queue, onemkl::blas::geru, + (main_queue, m, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during GERU:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, m, n, lda, std::max(m, n), std::cout); + } + + return good; +} + +class GeruTests : public ::testing::TestWithParam {}; + +TEST_P(GeruTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 1, 1, 42)); +} +TEST_P(GeruTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), 25, 30, alpha, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(GeruTestSuite, GeruTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/hbmv.cpp b/tests/unit_tests/blas/level2/hbmv.cpp new file mode 100644 index 000000000..020c599f0 --- /dev/null +++ b/tests/unit_tests/blas/level2/hbmv.cpp @@ -0,0 +1,152 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, int k, fp alpha, fp beta, int incx, + int incy, int lda) { + // Prepare data. + + vector x, y, y_ref, A; + + rand_vector(x, n, incx); + rand_vector(y, n, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + + // Call Reference HBMV. + const int n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + const int k_ref = k; + using fp_ref = typename ref_type_info::type; + + ::hbmv(convert_to_cblas_uplo(upper_lower), &n_ref, &k_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + &lda_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), + &incy_ref); + + // Call DPC++ HBMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::hbmv(main_queue, upper_lower, n, k, alpha, A_buffer, lda, x_buffer, incx, + beta, y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::hbmv, + (main_queue, upper_lower, n, k, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, n, incy, n, std::cout); + } + + return good; +} + +class HbmvTests : public ::testing::TestWithParam {}; + +TEST_P(HbmvTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 1, 1, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 1, 1, 42)); +} +TEST_P(HbmvTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, -2, + -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, -2, + -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 1, 1, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(HbmvTestSuite, HbmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/hemv.cpp b/tests/unit_tests/blas/level2/hemv.cpp new file mode 100644 index 000000000..d190121de --- /dev/null +++ b/tests/unit_tests/blas/level2/hemv.cpp @@ -0,0 +1,150 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, fp beta, int incx, int incy, + int lda) { + // Prepare data. + vector x, y, y_ref, A; + + rand_vector(x, n, incx); + rand_vector(y, n, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + + // Call Reference HEMV. + const int n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::hemv(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + &lda_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), + &incy_ref); + + // Call DPC++ HEMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HEMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::hemv(main_queue, upper_lower, n, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::hemv, + (main_queue, upper_lower, n, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HEMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, n, incy, n, std::cout); + } + + return good; +} + +class HemvTests : public ::testing::TestWithParam {}; + +TEST_P(HemvTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1, 42)); +} +TEST_P(HemvTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1, 42)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(HemvTestSuite, HemvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/her.cpp b/tests/unit_tests/blas/level2/her.cpp new file mode 100644 index 000000000..291222d98 --- /dev/null +++ b/tests/unit_tests/blas/level2/her.cpp @@ -0,0 +1,141 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp_scalar alpha, int incx, int lda) { + // Prepare data. + vector x, A_ref, A; + rand_vector(x, n, incx); + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + A_ref = A; + + // Call Reference HER. + const int n_ref = n, incx_ref = incx, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + using fp_scalar_mkl = typename ref_type_info::type; + + ::her(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_scalar_mkl *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ HER. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HER:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::her(main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer, lda); +#else + TEST_RUN_CT(main_queue, onemkl::blas::her, + (main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HER:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, lda, n, std::cout); + } + + return good; +} + +class HerTests : public ::testing::TestWithParam {}; + +TEST_P(HerTests, ComplexSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE( + (test, float>(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 42))); + EXPECT_TRUE( + (test, float>(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 42))); + EXPECT_TRUE( + (test, float>(GetParam(), onemkl::uplo::lower, 30, alpha, -2, 42))); + EXPECT_TRUE( + (test, float>(GetParam(), onemkl::uplo::upper, 30, alpha, -2, 42))); + EXPECT_TRUE( + (test, float>(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 42))); + EXPECT_TRUE( + (test, float>(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 42))); +} +TEST_P(HerTests, ComplexDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 42))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 42))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::lower, 30, alpha, -2, 42))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::upper, 30, alpha, -2, 42))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 42))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 42))); +} + +INSTANTIATE_TEST_SUITE_P(HerTestSuite, HerTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/her2.cpp b/tests/unit_tests/blas/level2/her2.cpp new file mode 100644 index 000000000..f3706f8c4 --- /dev/null +++ b/tests/unit_tests/blas/level2/her2.cpp @@ -0,0 +1,135 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, int incx, int incy, + int lda) { + // Prepare data. + vector x, y, A_ref, A; + + rand_vector(x, n, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + A_ref = A; + + // Call Reference HER2. + const int n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::her2(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)y.data(), &incy_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ HER2. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HER2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::her2(main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, + A_buffer, lda); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::her2, + (main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HER2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, lda, n, std::cout); + } + + return good; +} + +class Her2Tests : public ::testing::TestWithParam {}; + +TEST_P(Her2Tests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1, 42)); +} +TEST_P(Her2Tests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(Her2TestSuite, Her2Tests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/hpmv.cpp b/tests/unit_tests/blas/level2/hpmv.cpp new file mode 100644 index 000000000..05073af29 --- /dev/null +++ b/tests/unit_tests/blas/level2/hpmv.cpp @@ -0,0 +1,140 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, fp beta, int incx, + int incy) { + // Prepare data. + vector x, y, y_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, n, n, n); + + // Call Reference HPMV. + const int n_ref = n, incx_ref = incx, incy_ref = incy; + using fp_ref = typename ref_type_info::type; + + ::hpmv(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), &incy_ref); + + // Call DPC++ HPMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HPMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::hpmv(main_queue, upper_lower, n, alpha, A_buffer, x_buffer, incx, beta, + y_buffer, incy); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::hpmv, + (main_queue, upper_lower, n, alpha, A_buffer, x_buffer, incx, beta, y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HPMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, n, incy, n, std::cout); + } + + return good; +} + +class HpmvTests : public ::testing::TestWithParam {}; + +TEST_P(HpmvTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1)); +} +TEST_P(HpmvTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3)); + EXPECT_TRUE( + test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1)); +} + +INSTANTIATE_TEST_SUITE_P(HpmvTestSuite, HpmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/hpr.cpp b/tests/unit_tests/blas/level2/hpr.cpp new file mode 100644 index 000000000..440ee1f65 --- /dev/null +++ b/tests/unit_tests/blas/level2/hpr.cpp @@ -0,0 +1,136 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp_scalar alpha, int incx) { + // Prepare data. + vector x, A_ref, A; + rand_vector(x, n, incx); + rand_matrix(A, onemkl::transpose::nontrans, n, n, n); + A_ref = A; + + // Call Reference HPR. + const int n_ref = n, incx_ref = incx; + using fp_ref = typename ref_type_info::type; + using fp_scalar_mkl = typename ref_type_info::type; + + ::hpr(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_scalar_mkl *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)A_ref.data()); + + // Call DPC++ HPR. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HPR:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::hpr(main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::hpr, + (main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HPR:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, n, n, std::cout); + } + + return good; +} + +class HprTests : public ::testing::TestWithParam {}; + +TEST_P(HprTests, ComplexSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, 30, alpha, 2))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, 30, alpha, 2))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, 30, alpha, -2))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, 30, alpha, -2))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, 30, alpha, 1))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, 30, alpha, 1))); +} + +TEST_P(HprTests, ComplexDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::lower, 30, alpha, 2))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::upper, 30, alpha, 2))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::lower, 30, alpha, -2))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::upper, 30, alpha, -2))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::lower, 30, alpha, 1))); + EXPECT_TRUE( + (test, double>(GetParam(), onemkl::uplo::upper, 30, alpha, 1))); +} + +INSTANTIATE_TEST_SUITE_P(HprTestSuite, HprTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/hpr2.cpp b/tests/unit_tests/blas/level2/hpr2.cpp new file mode 100644 index 000000000..d4c0b5c48 --- /dev/null +++ b/tests/unit_tests/blas/level2/hpr2.cpp @@ -0,0 +1,132 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, int incx, int incy) { + // Prepare data. + vector x, y, A_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, n, n, n); + A_ref = A; + + // Call Reference HPR2. + const int n_ref = n, incx_ref = incx, incy_ref = incy; + using fp_ref = typename ref_type_info::type; + + ::hpr2(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)y.data(), &incy_ref, (fp_ref *)A_ref.data()); + + // Call DPC++ HPR2. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during HPR2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::hpr2(main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, + A_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::hpr2, + (main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during HPR2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, n, n, std::cout); + } + + return good; +} + +class Hpr2Tests : public ::testing::TestWithParam {}; + +TEST_P(Hpr2Tests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1)); +} +TEST_P(Hpr2Tests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1)); +} + +INSTANTIATE_TEST_SUITE_P(Hpr2TestSuite, Hpr2Tests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/sbmv.cpp b/tests/unit_tests/blas/level2/sbmv.cpp new file mode 100644 index 000000000..d78824657 --- /dev/null +++ b/tests/unit_tests/blas/level2/sbmv.cpp @@ -0,0 +1,137 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, int k, fp alpha, fp beta, int incx, + int incy, int lda) { + // Prepare data. + vector x, y, y_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + + // Call Reference SBMV. + const int n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + const int k_ref = k; + using fp_ref = typename ref_type_info::type; + + ::sbmv(convert_to_cblas_uplo(upper_lower), &n_ref, &k_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + &lda_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), + &incy_ref); + + // Call DPC++ SBMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + try { +#ifdef CALL_RT_API + onemkl::blas::sbmv(main_queue, upper_lower, n, k, alpha, A_buffer, lda, x_buffer, incx, + beta, y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::sbmv, + (main_queue, upper_lower, n, k, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, n, incy, n, std::cout); + } + + return good; +} + +class SbmvTests : public ::testing::TestWithParam {}; + +TEST_P(SbmvTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 1, 1, 42)); +} +TEST_P(SbmvTests, RealDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, 5, alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, 5, alpha, beta, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(SbmvTestSuite, SbmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/spmv.cpp b/tests/unit_tests/blas/level2/spmv.cpp new file mode 100644 index 000000000..ec101f719 --- /dev/null +++ b/tests/unit_tests/blas/level2/spmv.cpp @@ -0,0 +1,136 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, fp beta, int incx, + int incy) { + // Prepare data. + vector x, y, y_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, n, n, n); + + // Call Reference SPMV. + const int n_ref = n, incx_ref = incx, incy_ref = incy; + using fp_ref = typename ref_type_info::type; + + ::spmv(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), &incy_ref); + + // Call DPC++ SPMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SPMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::spmv(main_queue, upper_lower, n, alpha, A_buffer, x_buffer, incx, beta, + y_buffer, incy); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::spmv, + (main_queue, upper_lower, n, alpha, A_buffer, x_buffer, incx, beta, y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SPMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, n, incy, n, std::cout); + } + + return good; +} + +class SpmvTests : public ::testing::TestWithParam {}; + +TEST_P(SpmvTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1)); +} +TEST_P(SpmvTests, RealDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1)); +} + +INSTANTIATE_TEST_SUITE_P(SpmvTestSuite, SpmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/spr.cpp b/tests/unit_tests/blas/level2/spr.cpp new file mode 100644 index 000000000..5fce74a62 --- /dev/null +++ b/tests/unit_tests/blas/level2/spr.cpp @@ -0,0 +1,128 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, int incx) { + // Prepare data. + vector x, A_ref, A; + rand_vector(x, n, incx); + rand_matrix(A, onemkl::transpose::nontrans, n, n, n); + A_ref = A; + + // Call Reference SPR. + const int n_ref = n, incx_ref = incx; + using fp_ref = typename ref_type_info::type; + + ::spr(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)A_ref.data()); + + // Call DPC++ SPR. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SPR:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::spr(main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::spr, + (main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SPR:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, n, n, std::cout); + } + + return good; +} + +class SprTests : public ::testing::TestWithParam {}; + +TEST_P(SprTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1)); +} +TEST_P(SprTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1)); +} + +INSTANTIATE_TEST_SUITE_P(SprTestSuite, SprTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/spr2.cpp b/tests/unit_tests/blas/level2/spr2.cpp new file mode 100644 index 000000000..b04f61556 --- /dev/null +++ b/tests/unit_tests/blas/level2/spr2.cpp @@ -0,0 +1,132 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, int incx, int incy) { + // Prepare data. + vector x, y, A_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, n, n, n); + A_ref = A; + + // Call Reference SPR2. + const int n_ref = n, incx_ref = incx, incy_ref = incy; + using fp_ref = typename ref_type_info::type; + + ::spr2(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)y.data(), &incy_ref, (fp_ref *)A_ref.data()); + + // Call DPC++ SPR2. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SPR2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::spr2(main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, + A_buffer); +#else + TEST_RUN_CT(main_queue, onemkl::blas::spr2, + (main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SPR2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, n, n, std::cout); + } + + return good; +} + +class Spr2Tests : public ::testing::TestWithParam {}; + +TEST_P(Spr2Tests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1)); +} +TEST_P(Spr2Tests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1)); +} + +INSTANTIATE_TEST_SUITE_P(Spr2TestSuite, Spr2Tests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/symv.cpp b/tests/unit_tests/blas/level2/symv.cpp new file mode 100644 index 000000000..05d8a7987 --- /dev/null +++ b/tests/unit_tests/blas/level2/symv.cpp @@ -0,0 +1,137 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, fp beta, int incx, int incy, + int lda) { + // Prepare data. + vector x, y, y_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + y_ref = y; + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + + // Call Reference SYMV. + const int n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::symv(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)A.data(), + &lda_ref, (fp_ref *)x.data(), &incx_ref, (fp_ref *)&beta, (fp_ref *)y_ref.data(), + &incy_ref); + + // Call DPC++ SYMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SYMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::symv(main_queue, upper_lower, n, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy); +#else + TEST_RUN_CT(main_queue, onemkl::blas::symv, + (main_queue, upper_lower, n, alpha, A_buffer, lda, x_buffer, incx, beta, + y_buffer, incy)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SYMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto y_accessor = y_buffer.template get_access(); + good = check_equal_vector(y_accessor, y_ref, n, incy, n, std::cout); + } + + return good; +} + +class SymvTests : public ::testing::TestWithParam {}; + +TEST_P(SymvTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1, 42)); +} +TEST_P(SymvTests, RealDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, beta, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, beta, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(SymvTestSuite, SymvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/syr.cpp b/tests/unit_tests/blas/level2/syr.cpp new file mode 100644 index 000000000..f22a116ed --- /dev/null +++ b/tests/unit_tests/blas/level2/syr.cpp @@ -0,0 +1,128 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, int incx, int lda) { + // Prepare data. + vector x, A_ref, A; + rand_vector(x, n, incx); + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + A_ref = A; + + // Call Reference SYR. + const int n_ref = n, incx_ref = incx, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::syr(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ SYR. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SYR:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::syr(main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer, lda); +#else + TEST_RUN_CT(main_queue, onemkl::blas::syr, + (main_queue, upper_lower, n, alpha, x_buffer, incx, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SYR:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, lda, n, std::cout); + } + + return good; +} + +class SyrTests : public ::testing::TestWithParam {}; + +TEST_P(SyrTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 42)); +} +TEST_P(SyrTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(SyrTestSuite, SyrTests, ::testing::ValuesIn(devices), ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/syr2.cpp b/tests/unit_tests/blas/level2/syr2.cpp new file mode 100644 index 000000000..4248e38da --- /dev/null +++ b/tests/unit_tests/blas/level2/syr2.cpp @@ -0,0 +1,134 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device &dev, onemkl::uplo upper_lower, int n, fp alpha, int incx, int incy, + int lda) { + // Prepare data. + vector x, y, A_ref, A; + rand_vector(x, n, incx); + rand_vector(y, n, incy); + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + A_ref = A; + + // Call Reference SYR2. + const int n_ref = n, incx_ref = incx, incy_ref = incy, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::syr2(convert_to_cblas_uplo(upper_lower), &n_ref, (fp_ref *)&alpha, (fp_ref *)x.data(), + &incx_ref, (fp_ref *)y.data(), &incy_ref, (fp_ref *)A_ref.data(), &lda_ref); + + // Call DPC++ SYR2. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const &e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const &e) { + std::cout << "Caught asynchronous SYCL exception during SYR2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer y_buffer = make_buffer(y); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::syr2(main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, + A_buffer, lda); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::syr2, + (main_queue, upper_lower, n, alpha, x_buffer, incx, y_buffer, incy, A_buffer, lda)); +#endif + } + catch (exception const &e) { + std::cout << "Caught synchronous SYCL exception during SYR2:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto A_accessor = A_buffer.template get_access(); + good = check_equal_matrix(A_accessor, A_ref, n, n, lda, n, std::cout); + } + + return good; +} + +class Syr2Tests : public ::testing::TestWithParam {}; + +TEST_P(Syr2Tests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1, 42)); +} +TEST_P(Syr2Tests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 2, 3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, -2, -3, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, 30, alpha, 1, 1, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, 30, alpha, 1, 1, 42)); +} + +INSTANTIATE_TEST_SUITE_P(Syr2TestSuite, Syr2Tests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/tbmv.cpp b/tests/unit_tests/blas/level2/tbmv.cpp new file mode 100644 index 000000000..40525f30f --- /dev/null +++ b/tests/unit_tests/blas/level2/tbmv.cpp @@ -0,0 +1,224 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::diag unit_nonunit, int n, int k, int incx, int lda) { + // Prepare data. + vector x, x_ref, A; + rand_vector(x, n, incx); + x_ref = x; + rand_matrix(A, transa, n, n, lda); + + // Call Reference TBMV. + const int n_ref = n, incx_ref = incx, lda_ref = lda; + const int k_ref = k; + using fp_ref = typename ref_type_info::type; + + ::tbmv(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(transa), + convert_to_cblas_diag(unit_nonunit), &n_ref, &k_ref, (fp_ref*)A.data(), &lda_ref, + (fp_ref*)x_ref.data(), &incx_ref); + + // Call DPC++ TBMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::tbmv(main_queue, upper_lower, transa, unit_nonunit, n, k, A_buffer, lda, + x_buffer, incx); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::tbmv, + (main_queue, upper_lower, transa, unit_nonunit, n, k, A_buffer, lda, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_vector(x_accessor, x_ref, n, incx, n, std::cout); + } + + return good; +} + +class TbmvTests : public ::testing::TestWithParam {}; + +TEST_P(TbmvTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); +} +TEST_P(TbmvTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); +} +TEST_P(TbmvTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); +} +TEST_P(TbmvTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); +} + +INSTANTIATE_TEST_SUITE_P(TbmvTestSuite, TbmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/tbsv.cpp b/tests/unit_tests/blas/level2/tbsv.cpp new file mode 100644 index 000000000..c0792ad68 --- /dev/null +++ b/tests/unit_tests/blas/level2/tbsv.cpp @@ -0,0 +1,224 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::diag unit_nonunit, int n, int k, int incx, int lda) { + // Prepare data. + vector x, x_ref, A; + rand_vector(x, n, incx); + x_ref = x; + rand_trsm_matrix(A, transa, n, n, lda); + + // Call Reference TBSV. + const int n_ref = n, incx_ref = incx, lda_ref = lda; + const int k_ref = k; + using fp_ref = typename ref_type_info::type; + + ::tbsv(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(transa), + convert_to_cblas_diag(unit_nonunit), &n_ref, &k_ref, (fp_ref*)A.data(), &lda_ref, + (fp_ref*)x_ref.data(), &incx_ref); + + // Call DPC++ TBSV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TBSV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::tbsv(main_queue, upper_lower, transa, unit_nonunit, n, k, A_buffer, lda, + x_buffer, incx); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::tbsv, + (main_queue, upper_lower, transa, unit_nonunit, n, k, A_buffer, lda, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TBSV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_trsv_vector(x_accessor, x_ref, n, incx, n, std::cout); + } + + return good; +} + +class TbsvTests : public ::testing::TestWithParam {}; + +TEST_P(TbsvTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); +} +TEST_P(TbsvTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); +} +TEST_P(TbsvTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); +} +TEST_P(TbsvTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 30, 5, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 5, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 5, 2, 42)); +} + +INSTANTIATE_TEST_SUITE_P(TbsvTestSuite, TbsvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/tpmv.cpp b/tests/unit_tests/blas/level2/tpmv.cpp new file mode 100644 index 000000000..5c54e3093 --- /dev/null +++ b/tests/unit_tests/blas/level2/tpmv.cpp @@ -0,0 +1,210 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::diag unit_nonunit, int n, int incx) { + // Prepare data. + vector x, x_ref, A; + rand_vector(x, n, incx); + x_ref = x; + rand_matrix(A, transa, n, n, n); + + // Call Reference TPMV. + const int n_ref = n, incx_ref = incx; + using fp_ref = typename ref_type_info::type; + + ::tpmv(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(transa), + convert_to_cblas_diag(unit_nonunit), &n_ref, (fp_ref*)A.data(), (fp_ref*)x_ref.data(), + &incx_ref); + + // Call DPC++ TPMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::tpmv(main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, x_buffer, + incx); +#else + TEST_RUN_CT(main_queue, onemkl::blas::tpmv, + (main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TBMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_vector(x_accessor, x_ref, n, incx, n, std::cout); + } + + return good; +} + +class TpmvTests : public ::testing::TestWithParam {}; + +TEST_P(TpmvTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); +} +TEST_P(TpmvTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); +} +TEST_P(TpmvTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); +} +TEST_P(TpmvTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::lower, onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::upper, onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); +} + +INSTANTIATE_TEST_SUITE_P(TpmvTestSuite, TpmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/tpsv.cpp b/tests/unit_tests/blas/level2/tpsv.cpp new file mode 100644 index 000000000..7bb7e9e3c --- /dev/null +++ b/tests/unit_tests/blas/level2/tpsv.cpp @@ -0,0 +1,210 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::diag unit_nonunit, int n, int incx) { + // Prepare data. + vector x, x_ref, A; + rand_vector(x, n, incx); + x_ref = x; + rand_trsm_matrix(A, transa, n, n, n); + + // Call Reference TPSV. + const int n_ref = n, incx_ref = incx; + using fp_ref = typename ref_type_info::type; + + ::tpsv(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(transa), + convert_to_cblas_diag(unit_nonunit), &n_ref, (fp_ref*)A.data(), (fp_ref*)x_ref.data(), + &incx_ref); + + // Call DPC++ TPSV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TPSV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::tpsv(main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, x_buffer, + incx); +#else + TEST_RUN_CT(main_queue, onemkl::blas::tpsv, + (main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TPSV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_trsv_vector(x_accessor, x_ref, n, incx, n, std::cout); + } + + return good; +} + +class TpsvTests : public ::testing::TestWithParam {}; + +TEST_P(TpsvTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); +} +TEST_P(TpsvTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); +} +TEST_P(TpsvTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); +} +TEST_P(TpsvTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::lower, onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::upper, onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2)); +} + +INSTANTIATE_TEST_SUITE_P(TpsvTestSuite, TpsvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/trmv.cpp b/tests/unit_tests/blas/level2/trmv.cpp new file mode 100644 index 000000000..6980e804b --- /dev/null +++ b/tests/unit_tests/blas/level2/trmv.cpp @@ -0,0 +1,221 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::diag unit_nonunit, int n, int incx, int lda) { + // Prepare data. + vector x, x_ref, A; + rand_vector(x, n, incx); + x_ref = x; + rand_matrix(A, transa, n, n, lda); + + // Call Reference TRMV. + const int n_ref = n, incx_ref = incx, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::trmv(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(transa), + convert_to_cblas_diag(unit_nonunit), &n_ref, (fp_ref*)A.data(), &lda_ref, + (fp_ref*)x_ref.data(), &incx_ref); + + // Call DPC++ TRMV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TRMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::trmv(main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, lda, + x_buffer, incx); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::trmv, + (main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, lda, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TRMV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_vector(x_accessor, x_ref, n, incx, n, std::cout); + } + + return good; +} + +class TrmvTests : public ::testing::TestWithParam {}; + +TEST_P(TrmvTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); +} +TEST_P(TrmvTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); +} +TEST_P(TrmvTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); +} +TEST_P(TrmvTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); +} + +INSTANTIATE_TEST_SUITE_P(TrmvTestSuite, TrmvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level2/trsv.cpp b/tests/unit_tests/blas/level2/trsv.cpp new file mode 100644 index 000000000..0f414d41d --- /dev/null +++ b/tests/unit_tests/blas/level2/trsv.cpp @@ -0,0 +1,221 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose transa, + onemkl::diag unit_nonunit, int n, int incx, int lda) { + // Prepare data. + vector x, x_ref, A; + rand_vector(x, n, incx); + x_ref = x; + rand_trsm_matrix(A, transa, n, n, lda); + + // Call Reference TRSV. + const int n_ref = n, incx_ref = incx, lda_ref = lda; + using fp_ref = typename ref_type_info::type; + + ::trsv(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(transa), + convert_to_cblas_diag(unit_nonunit), &n_ref, (fp_ref*)A.data(), &lda_ref, + (fp_ref*)x_ref.data(), &incx_ref); + + // Call DPC++ TRSV. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TRSV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer x_buffer = make_buffer(x); + buffer A_buffer = make_buffer(A); + + try { +#ifdef CALL_RT_API + onemkl::blas::trsv(main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, lda, + x_buffer, incx); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::trsv, + (main_queue, upper_lower, transa, unit_nonunit, n, A_buffer, lda, x_buffer, incx)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TRSV:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto x_accessor = x_buffer.template get_access(); + good = check_equal_trsv_vector(x_accessor, x_ref, n, incx, n, std::cout); + } + + return good; +} + +class TrsvTests : public ::testing::TestWithParam {}; + +TEST_P(TrsvTests, RealSinglePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); +} +TEST_P(TrsvTests, RealDoublePrecision) { + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); +} +TEST_P(TrsvTests, ComplexSinglePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + onemkl::diag::nonunit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); +} +TEST_P(TrsvTests, ComplexDoublePrecision) { + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, onemkl::diag::unit, 30, 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 30, 2, + 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 30, + 2, 42)); +} + +INSTANTIATE_TEST_SUITE_P(TrsvTestSuite, TrsvTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/CMakeLists.txt b/tests/unit_tests/blas/level3/CMakeLists.txt new file mode 100644 index 000000000..12fffff37 --- /dev/null +++ b/tests/unit_tests/blas/level3/CMakeLists.txt @@ -0,0 +1,49 @@ +#=============================================================================== +# Copyright 2020 Intel Corporation +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions +# and limitations under the License. +# +# +# SPDX-License-Identifier: Apache-2.0 +#=============================================================================== + +# Build object from all test sources +set(L3_SOURCES "gemm.cpp" "symm.cpp" "syrk.cpp" "hemm.cpp" "herk.cpp" "syr2k.cpp" "her2k.cpp" "trmm.cpp" "trsm.cpp") + +if(BUILD_SHARED_LIBS) + add_library(blas_level3_rt OBJECT ${L3_SOURCES}) + target_compile_options(blas_level3_rt PRIVATE -DCALL_RT_API) + target_include_directories(blas_level3_rt + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../include + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../include + PUBLIC ${PROJECT_SOURCE_DIR}/include + PUBLIC ${PROJECT_SOURCE_DIR}/deps/googletest/include + PUBLIC ${CMAKE_BINARY_DIR}/bin/onemkl + PUBLIC ${CBLAS_INCLUDE} + ) + target_link_libraries(blas_level3_rt PUBLIC ONEMKL::SYCL::SYCL) +endif() + +add_library(blas_level3_ct OBJECT ${L3_SOURCES}) +target_compile_options(blas_level3_ct PRIVATE) +target_include_directories(blas_level3_ct + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../include + PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../include + PUBLIC ${PROJECT_SOURCE_DIR}/include + PUBLIC ${PROJECT_SOURCE_DIR}/deps/googletest/include + PUBLIC ${CMAKE_BINARY_DIR}/bin/onemkl + PUBLIC ${CBLAS_INCLUDE} +) +target_link_libraries(blas_level3_ct PUBLIC ONEMKL::SYCL::SYCL) + + diff --git a/tests/unit_tests/blas/level3/gemm.cpp b/tests/unit_tests/blas/level3/gemm.cpp new file mode 100644 index 000000000..58f3a30c8 --- /dev/null +++ b/tests/unit_tests/blas/level3/gemm.cpp @@ -0,0 +1,206 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::transpose transa, onemkl::transpose transb, int m, int n, + int k, int lda, int ldb, int ldc, fp alpha, fp beta) { + // Prepare data. + vector> A, B, C, C_ref; + rand_matrix(A, transa, m, k, lda); + rand_matrix(B, transb, k, n, ldb); + rand_matrix(C, onemkl::transpose::nontrans, m, n, ldc); + C_ref = C; + + // Call Reference GEMM. + const int m_ref = m, n_ref = n, k_ref = k; + const int lda_ref = lda, ldb_ref = ldb, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + + ::gemm(convert_to_cblas_trans(transa), convert_to_cblas_trans(transb), &m_ref, &n_ref, &k_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B.data(), &ldb_ref, + (fp_ref*)&beta, (fp_ref*)C_ref.data(), &ldc_ref); + + // Call DPC++ GEMM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during GEMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::gemm(main_queue, transa, transb, m, n, k, alpha, A_buffer, lda, B_buffer, ldb, + beta, C_buffer, ldc); +#else + TEST_RUN_CT(main_queue, onemkl::blas::gemm, + (main_queue, transa, transb, m, n, k, alpha, A_buffer, lda, B_buffer, ldb, beta, + C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during GEMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + auto C_accessor = C_buffer.template get_access(); + bool good = check_equal_matrix(C_accessor, C_ref, m, n, ldc, 10 * k, std::cout); + + return good; +} + +class GemmTests : public ::testing::TestWithParam {}; + +TEST_P(GemmTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::nontrans, onemkl::transpose::nontrans, + 79, 83, 91, 103, 105, 106, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::nontrans, onemkl::transpose::trans, 79, + 83, 91, 103, 105, 106, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, onemkl::transpose::nontrans, 79, + 83, 91, 103, 105, 106, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, onemkl::transpose::trans, 79, 83, + 91, 103, 105, 106, alpha, beta)); +} + +TEST_P(GemmTests, RealDoublePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::nontrans, onemkl::transpose::nontrans, + 79, 83, 91, 103, 105, 106, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::nontrans, onemkl::transpose::trans, 79, + 83, 91, 103, 105, 106, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, onemkl::transpose::nontrans, 79, + 83, 91, 103, 105, 106, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::transpose::trans, onemkl::transpose::trans, 79, 83, + 91, 103, 105, 106, alpha, beta)); +} + +TEST_P(GemmTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, + onemkl::transpose::nontrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, + onemkl::transpose::trans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, + onemkl::transpose::nontrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, + onemkl::transpose::trans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, + onemkl::transpose::conjtrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, + onemkl::transpose::conjtrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, + onemkl::transpose::nontrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, + onemkl::transpose::trans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, + onemkl::transpose::conjtrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); +} + +TEST_P(GemmTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, + onemkl::transpose::nontrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, + onemkl::transpose::trans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, + onemkl::transpose::nontrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, + onemkl::transpose::trans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::nontrans, + onemkl::transpose::conjtrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::trans, + onemkl::transpose::conjtrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, + onemkl::transpose::nontrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, + onemkl::transpose::trans, 79, 83, 91, 103, 105, 106, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::transpose::conjtrans, + onemkl::transpose::conjtrans, 79, 83, 91, 103, 105, 106, + alpha, beta)); +} + +INSTANTIATE_TEST_SUITE_P(GemmTestSuite, GemmTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/hemm.cpp b/tests/unit_tests/blas/level3/hemm.cpp new file mode 100644 index 000000000..b75aa1864 --- /dev/null +++ b/tests/unit_tests/blas/level3/hemm.cpp @@ -0,0 +1,146 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::side left_right, onemkl::uplo upper_lower, int m, int n, + int lda, int ldb, int ldc, fp alpha, fp beta) { + // Prepare data. + vector> A, B, C, C_ref; + if (left_right == onemkl::side::left) + rand_matrix(A, onemkl::transpose::nontrans, m, m, lda); + else + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + rand_matrix(B, onemkl::transpose::nontrans, m, n, ldb); + rand_matrix(C, onemkl::transpose::nontrans, m, n, ldc); + C_ref = C; + + // Call Reference HEMM. + const int m_ref = m, n_ref = n; + const int lda_ref = lda, ldb_ref = ldb, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + + ::hemm(convert_to_cblas_side(left_right), convert_to_cblas_uplo(upper_lower), &m_ref, &n_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B.data(), &ldb_ref, + (fp_ref*)&beta, (fp_ref*)C_ref.data(), &ldc_ref); + + // Call DPC++ HEMM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during HEMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::hemm(main_queue, left_right, upper_lower, m, n, alpha, A_buffer, lda, + B_buffer, ldb, beta, C_buffer, ldc); +#else + TEST_RUN_CT(main_queue, onemkl::blas::hemm, + (main_queue, left_right, upper_lower, m, n, alpha, A_buffer, lda, B_buffer, ldb, + beta, C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during HEMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto C_accessor = C_buffer.template get_access(); + good = check_equal_matrix(C_accessor, C_ref, m, n, ldc, 10 * std::max(m, n), std::cout); + } + return good; +} + +class HemmTests : public ::testing::TestWithParam {}; + +TEST_P(HemmTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); +} +TEST_P(HemmTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); +} + +INSTANTIATE_TEST_SUITE_P(HemmTestSuite, HemmTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/her2k.cpp b/tests/unit_tests/blas/level3/her2k.cpp new file mode 100644 index 000000000..d4127986d --- /dev/null +++ b/tests/unit_tests/blas/level3/her2k.cpp @@ -0,0 +1,153 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose trans, int n, int k, + int lda, int ldb, int ldc, fp alpha, fp_scalar beta) { + // Prepare data. + vector> A, B, C, C_ref; + rand_matrix(A, trans, n, k, lda); + rand_matrix(B, trans, n, k, ldb); + rand_matrix(C, onemkl::transpose::nontrans, n, n, ldc); + C_ref = C; + + // Call Reference HER2K. + const int n_ref = n, k_ref = k; + const int lda_ref = lda, ldb_ref = ldb, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + using fp_scalar_mkl = typename ref_type_info::type; + + ::her2k(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(trans), &n_ref, &k_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B.data(), &ldb_ref, + (fp_scalar_mkl*)&beta, (fp_ref*)C_ref.data(), &ldc_ref); + + // Call DPC++ HER2K. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during HER2K:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::her2k(main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, B_buffer, + ldb, beta, C_buffer, ldc); +#else + TEST_RUN_CT(main_queue, onemkl::blas::her2k, + (main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, B_buffer, ldb, + beta, C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during HER2K:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto C_accessor = C_buffer.template get_access(); + good = check_equal_matrix(C_accessor, C_ref, n, n, ldc, 10 * std::max(n, k), std::cout); + } + + return good; +} + +class Her2kTests : public ::testing::TestWithParam {}; + +TEST_P(Her2kTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + float beta(1.0); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 72, 27, 101, 102, + 103, alpha, beta))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 72, 27, 101, 102, + 103, alpha, beta))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, 72, 27, 101, 102, + 103, alpha, beta))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, 72, 27, 101, 102, + 103, alpha, beta))); +} +TEST_P(Her2kTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + double beta(1.0); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 72, 27, 101, 102, + 103, alpha, beta))); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 72, 27, 101, 102, + 103, alpha, beta))); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, 72, 27, 101, 102, + 103, alpha, beta))); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, 72, 27, 101, 102, + 103, alpha, beta))); +} + +INSTANTIATE_TEST_SUITE_P(Her2kTestSuite, Her2kTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/herk.cpp b/tests/unit_tests/blas/level3/herk.cpp new file mode 100644 index 000000000..9db3b28c5 --- /dev/null +++ b/tests/unit_tests/blas/level3/herk.cpp @@ -0,0 +1,150 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose trans, int n, int k, + int lda, int ldc, fp_scalar alpha, fp_scalar beta) { + // Prepare data. + vector> A, C, C_ref; + rand_matrix(A, trans, n, k, lda); + rand_matrix(C, onemkl::transpose::nontrans, n, n, ldc); + C_ref = C; + + // Call Reference HERK. + const int n_ref = n, k_ref = k; + const int lda_ref = lda, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + + ::herk(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(trans), &n_ref, &k_ref, + (fp_scalar*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_scalar*)&beta, + (fp_ref*)C_ref.data(), &ldc_ref); + + // Call DPC++ HERK. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during HERK:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::herk(main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, beta, + C_buffer, ldc); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::herk, + (main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, beta, C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during HERK:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto C_accessor = C_buffer.template get_access(); + good = check_equal_matrix(C_accessor, C_ref, n, n, ldc, 10 * std::max(n, k), std::cout); + } + + return good; +} + +class HerkTests : public ::testing::TestWithParam {}; + +TEST_P(HerkTests, ComplexSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 72, 27, 101, 103, + alpha, beta))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 72, 27, 101, 103, + alpha, beta))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, 72, 27, 101, 103, + alpha, beta))); + EXPECT_TRUE((test, float>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, 72, 27, 101, 103, + alpha, beta))); +} +TEST_P(HerkTests, ComplexDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 72, 27, 101, 103, + alpha, beta))); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 72, 27, 101, 103, + alpha, beta))); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::conjtrans, 72, 27, 101, 103, + alpha, beta))); + EXPECT_TRUE((test, double>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::conjtrans, 72, 27, 101, 103, + alpha, beta))); +} + +INSTANTIATE_TEST_SUITE_P(HerkTestSuite, HerkTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/symm.cpp b/tests/unit_tests/blas/level3/symm.cpp new file mode 100644 index 000000000..4f0270eaf --- /dev/null +++ b/tests/unit_tests/blas/level3/symm.cpp @@ -0,0 +1,171 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::side left_right, onemkl::uplo upper_lower, int m, int n, + int lda, int ldb, int ldc, fp alpha, fp beta) { + // Prepare data. + vector> A, B, C, C_ref; + if (left_right == onemkl::side::left) + rand_matrix(A, onemkl::transpose::nontrans, m, m, lda); + else + rand_matrix(A, onemkl::transpose::nontrans, n, n, lda); + rand_matrix(B, onemkl::transpose::nontrans, m, n, ldb); + rand_matrix(C, onemkl::transpose::nontrans, m, n, ldc); + C_ref = C; + + // Call Reference SYMM. + const int m_ref = m, n_ref = n; + const int lda_ref = lda, ldb_ref = ldb, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + + ::symm(convert_to_cblas_side(left_right), convert_to_cblas_uplo(upper_lower), &m_ref, &n_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B.data(), &ldb_ref, + (fp_ref*)&beta, (fp_ref*)C_ref.data(), &ldc_ref); + + // Call DPC++ SYMM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during SYMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::symm(main_queue, left_right, upper_lower, m, n, alpha, A_buffer, lda, + B_buffer, ldb, beta, C_buffer, ldc); +#else + TEST_RUN_CT(main_queue, onemkl::blas::symm, + (main_queue, left_right, upper_lower, m, n, alpha, A_buffer, lda, B_buffer, ldb, + beta, C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during SYMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto C_accessor = C_buffer.template get_access(); + good = check_equal_matrix(C_accessor, C_ref, m, n, ldc, 10 * std::max(m, n), std::cout); + } + + return good; +} + +class SymmTests : public ::testing::TestWithParam {}; + +TEST_P(SymmTests, RealSinglePrecision) { + float alpha(2.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, 72, 27, 101, 102, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, 72, 27, 101, 102, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, 72, 27, 101, 102, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, 72, 27, 101, 102, + 103, alpha, beta)); +} +TEST_P(SymmTests, RealDoublePrecision) { + double alpha(2.0); + double beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, 72, 27, 101, 102, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, 72, 27, 101, 102, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, 72, 27, 101, 102, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, 72, 27, 101, 102, + 103, alpha, beta)); +} +TEST_P(SymmTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); +} +TEST_P(SymmTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, 72, + 27, 101, 102, 103, alpha, beta)); +} + +INSTANTIATE_TEST_SUITE_P(SymmTestSuite, SymmTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/syr2k.cpp b/tests/unit_tests/blas/level3/syr2k.cpp new file mode 100644 index 000000000..c1d45c1ee --- /dev/null +++ b/tests/unit_tests/blas/level3/syr2k.cpp @@ -0,0 +1,174 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose trans, int n, int k, + int lda, int ldb, int ldc, fp alpha, fp beta) { + // Prepare data. + vector> A, B, C, C_ref; + rand_matrix(A, trans, n, k, lda); + rand_matrix(B, trans, n, k, ldb); + rand_matrix(C, onemkl::transpose::nontrans, n, n, ldc); + C_ref = C; + + // Call Reference SYR2K. + const int n_ref = n, k_ref = k; + const int lda_ref = lda, ldb_ref = ldb, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + + ::syr2k(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(trans), &n_ref, &k_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B.data(), &ldb_ref, + (fp_ref*)&beta, (fp_ref*)C_ref.data(), &ldc_ref); + + // Call DPC++ SYR2K. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during SYR2K:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::syr2k(main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, B_buffer, + ldb, beta, C_buffer, ldc); +#else + TEST_RUN_CT(main_queue, onemkl::blas::syr2k, + (main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, B_buffer, ldb, + beta, C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during SYR2K:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto C_accessor = C_buffer.template get_access(); + good = check_equal_matrix(C_accessor, C_ref, n, n, ldc, 10 * std::max(n, k), std::cout); + } + + return good; +} + +class Syr2kTests : public ::testing::TestWithParam {}; + +TEST_P(Syr2kTests, RealSinglePrecision) { + float alpha(3.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, 73, 27, + 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, 73, 27, + 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, 73, 27, 101, + 102, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, 73, 27, 101, + 102, 103, alpha, beta)); +} +TEST_P(Syr2kTests, RealDoublePrecision) { + double alpha(3.0); + double beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, 73, 27, + 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, 73, 27, + 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, 73, 27, 101, + 102, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, 73, 27, 101, + 102, 103, alpha, beta)); +} +TEST_P(Syr2kTests, ComplexSinglePrecision) { + std::complex alpha(3.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 73, 27, 101, 102, 103, alpha, + beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 73, 27, 101, 102, 103, alpha, + beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + 73, 27, 101, 102, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + 73, 27, 101, 102, 103, alpha, beta)); +} +TEST_P(Syr2kTests, ComplexDoublePrecision) { + std::complex alpha(3.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 73, 27, 101, 102, 103, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 73, 27, 101, 102, 103, + alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::trans, 73, 27, 101, 102, 103, alpha, + beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::trans, 73, 27, 101, 102, 103, alpha, + beta)); +} + +INSTANTIATE_TEST_SUITE_P(Syr2kTestSuite, Syr2kTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/syrk.cpp b/tests/unit_tests/blas/level3/syrk.cpp new file mode 100644 index 000000000..8ba9df6aa --- /dev/null +++ b/tests/unit_tests/blas/level3/syrk.cpp @@ -0,0 +1,170 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::uplo upper_lower, onemkl::transpose trans, int n, int k, + int lda, int ldc, fp alpha, fp beta) { + // Prepare data. + vector> A, C, C_ref; + rand_matrix(A, trans, n, k, lda); + rand_matrix(C, onemkl::transpose::nontrans, n, n, ldc); + C_ref = C; + + // Call Reference SYRK. + const int n_ref = n, k_ref = k; + const int lda_ref = lda, ldc_ref = ldc; + + using fp_ref = typename ref_type_info::type; + + ::syrk(convert_to_cblas_uplo(upper_lower), convert_to_cblas_trans(trans), &n_ref, &k_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)&beta, (fp_ref*)C_ref.data(), + &ldc_ref); + + // Call DPC++ SYRK. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during SYRK:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer C_buffer(C.data(), range<1>(C.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::syrk(main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, beta, + C_buffer, ldc); +#else + TEST_RUN_CT( + main_queue, onemkl::blas::syrk, + (main_queue, upper_lower, trans, n, k, alpha, A_buffer, lda, beta, C_buffer, ldc)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during SYRK:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto C_accessor = C_buffer.template get_access(); + good = check_equal_matrix(C_accessor, C_ref, n, n, ldc, 10 * std::max(n, k), std::cout); + } + + return good; +} + +class SyrkTests : public ::testing::TestWithParam {}; + +TEST_P(SyrkTests, RealSinglePrecision) { + float alpha(3.0); + float beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, 73, 27, + 101, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, 73, 27, + 101, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, 73, 27, 101, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, 73, 27, 101, + 103, alpha, beta)); +} +TEST_P(SyrkTests, RealDoublePrecision) { + double alpha(3.0); + double beta(3.0); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::nontrans, 73, 27, + 101, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::nontrans, 73, 27, + 101, 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, 73, 27, 101, + 103, alpha, beta)); + EXPECT_TRUE(test(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, 73, 27, 101, + 103, alpha, beta)); +} +TEST_P(SyrkTests, ComplexSinglePrecision) { + std::complex alpha(3.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 73, 27, 101, 103, alpha, + beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 73, 27, 101, 103, alpha, + beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, + 73, 27, 101, 103, alpha, beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, + 73, 27, 101, 103, alpha, beta)); +} +TEST_P(SyrkTests, ComplexDoublePrecision) { + std::complex alpha(3.0, -0.5); + std::complex beta(3.0, -1.5); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::lower, + onemkl::transpose::nontrans, 73, 27, 101, 103, alpha, + beta)); + EXPECT_TRUE(test>(GetParam(), onemkl::uplo::upper, + onemkl::transpose::nontrans, 73, 27, 101, 103, alpha, + beta)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::lower, onemkl::transpose::trans, 73, 27, 101, 103, alpha, beta)); + EXPECT_TRUE(test>( + GetParam(), onemkl::uplo::upper, onemkl::transpose::trans, 73, 27, 101, 103, alpha, beta)); +} + +INSTANTIATE_TEST_SUITE_P(SyrkTestSuite, SyrkTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/trmm.cpp b/tests/unit_tests/blas/level3/trmm.cpp new file mode 100644 index 000000000..207d0d4fd --- /dev/null +++ b/tests/unit_tests/blas/level3/trmm.cpp @@ -0,0 +1,279 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_nonunit, int m, int n, int lda, int ldb, + fp alpha) { + // Prepare data. + vector> A, B, B_ref; + if (left_right == onemkl::side::right) + rand_matrix(A, transa, n, n, lda); + else + rand_matrix(A, transa, m, m, lda); + + rand_matrix(B, onemkl::transpose::nontrans, m, n, ldb); + B_ref = B; + + // Call Reference TRMM. + const int m_ref = m, n_ref = n; + const int lda_ref = lda, ldb_ref = ldb; + + using fp_ref = typename ref_type_info::type; + + ::trmm(convert_to_cblas_side(left_right), convert_to_cblas_uplo(upper_lower), + convert_to_cblas_trans(transa), convert_to_cblas_diag(unit_nonunit), &m_ref, &n_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B_ref.data(), &ldb_ref); + + // Call DPC++ TRMM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TRMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::trmm(main_queue, left_right, upper_lower, transa, unit_nonunit, m, n, alpha, + A_buffer, lda, B_buffer, ldb); +#else + TEST_RUN_CT(main_queue, onemkl::blas::trmm, + (main_queue, left_right, upper_lower, transa, unit_nonunit, m, n, alpha, + A_buffer, lda, B_buffer, ldb)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TRMM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto B_accessor = B_buffer.template get_access(); + good = check_equal_matrix(B_accessor, B_ref, m, n, ldb, 10 * std::max(m, n), std::cout); + } + + return good; +} + +class TrmmTests : public ::testing::TestWithParam {}; + +TEST_P(TrmmTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); +} +TEST_P(TrmmTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); +} +TEST_P(TrmmTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); +} +TEST_P(TrmmTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); +} + +INSTANTIATE_TEST_SUITE_P(TrmmTestSuite, TrmmTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/blas/level3/trsm.cpp b/tests/unit_tests/blas/level3/trsm.cpp new file mode 100644 index 000000000..acec92b71 --- /dev/null +++ b/tests/unit_tests/blas/level3/trsm.cpp @@ -0,0 +1,372 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include +#include +#include + +#include +#include "allocator_helper.hpp" +#include "cblas.h" +#include "config.hpp" +#include "onemkl/onemkl.hpp" +#include "onemkl_blas_helper.hpp" +#include "reference_blas_templates.hpp" +#include "test_common.hpp" +#include "test_helper.hpp" + +#include + +using namespace cl::sycl; +using std::vector; + +extern std::vector devices; + +namespace { + +template +bool test(const device& dev, onemkl::side left_right, onemkl::uplo upper_lower, + onemkl::transpose transa, onemkl::diag unit_nonunit, int m, int n, int lda, int ldb, + fp alpha) { + // Prepare data. + vector> A, B, B_ref; + if (left_right == onemkl::side::right) + rand_trsm_matrix(A, transa, n, n, lda); + else + rand_trsm_matrix(A, transa, m, m, lda); + + rand_matrix(B, onemkl::transpose::nontrans, m, n, ldb); + B_ref = B; + + // Call Reference TRSM. + const int m_ref = m, n_ref = n; + const int lda_ref = lda, ldb_ref = ldb; + + using fp_ref = typename ref_type_info::type; + + ::trsm(convert_to_cblas_side(left_right), convert_to_cblas_uplo(upper_lower), + convert_to_cblas_trans(transa), convert_to_cblas_diag(unit_nonunit), &m_ref, &n_ref, + (fp_ref*)&alpha, (fp_ref*)A.data(), &lda_ref, (fp_ref*)B_ref.data(), &ldb_ref); + + // Call DPC++ TRSM. + + // Catch asynchronous exceptions. + auto exception_handler = [](exception_list exceptions) { + for (std::exception_ptr const& e : exceptions) { + try { + std::rethrow_exception(e); + } + catch (exception const& e) { + std::cout << "Caught asynchronous SYCL exception during TRSM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + } + }; + + queue main_queue(dev, exception_handler); + + buffer A_buffer(A.data(), range<1>(A.size())); + buffer B_buffer(B.data(), range<1>(B.size())); + + try { +#ifdef CALL_RT_API + onemkl::blas::trsm(main_queue, left_right, upper_lower, transa, unit_nonunit, m, n, alpha, + A_buffer, lda, B_buffer, ldb); +#else + TEST_RUN_CT(main_queue, onemkl::blas::trsm, + (main_queue, left_right, upper_lower, transa, unit_nonunit, m, n, alpha, + A_buffer, lda, B_buffer, ldb)); +#endif + } + catch (exception const& e) { + std::cout << "Caught synchronous SYCL exception during TRSM:\n" + << e.what() << std::endl + << "OpenCL status: " << e.get_cl_code() << std::endl; + } + + // Compare the results of reference implementation and DPC++ implementation. + bool good; + { + auto B_accessor = B_buffer.template get_access(); + good = + check_equal_trsm_matrix(B_accessor, B_ref, m, n, ldb, 10 * std::max(m, n), std::cout); + } + + return good; +} + +class TrsmTests : public ::testing::TestWithParam {}; + +TEST_P(TrsmTests, RealSinglePrecision) { + float alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); +} +TEST_P(TrsmTests, RealDoublePrecision) { + double alpha(2.0); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); + EXPECT_TRUE(test(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, 101, 102, + alpha)); +} +TEST_P(TrsmTests, ComplexSinglePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, + 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, + 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, + 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, 101, + 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); +} +TEST_P(TrsmTests, ComplexDoublePrecision) { + std::complex alpha(2.0, -0.5); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::unit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::nontrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::trans, onemkl::diag::nonunit, 72, 27, + 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::lower, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::left, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); + EXPECT_TRUE(test>(GetParam(), onemkl::side::right, onemkl::uplo::upper, + onemkl::transpose::conjtrans, onemkl::diag::nonunit, 72, + 27, 101, 102, alpha)); +} + +INSTANTIATE_TEST_SUITE_P(TrsmTestSuite, TrsmTests, ::testing::ValuesIn(devices), + ::DeviceNamePrint()); + +} // anonymous namespace diff --git a/tests/unit_tests/include/test_helper.hpp b/tests/unit_tests/include/test_helper.hpp new file mode 100644 index 000000000..4fd5df293 --- /dev/null +++ b/tests/unit_tests/include/test_helper.hpp @@ -0,0 +1,72 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#ifndef _TEST_HELPER_HPP_ +#define _TEST_HELPER_HPP_ + +#include + +#ifdef ENABLE_MKLCPU_BACKEND + #define TEST_RUN_INTELCPU(q, func, args) \ + func args +#else + #define TEST_RUN_INTELCPU(q, func, args) +#endif + +#ifdef ENABLE_MKLGPU_BACKEND + #define TEST_RUN_INTELGPU(q, func, args) \ + func args +#else + #define TEST_RUN_INTELGPU(q, func, args) +#endif + +#define TEST_RUN_CT(q, func, args) \ + do { \ + if (q.is_host() || q.get_device().is_cpu()) \ + TEST_RUN_INTELCPU(q, func, args); \ + else if (q.get_device().is_gpu()) { \ + unsigned int vendor_id = static_cast( \ + q.get_device().get_info()); \ + if (vendor_id == INTEL_ID) \ + TEST_RUN_INTELGPU(q, func, args); \ + } \ + } while (0); + +class DeviceNamePrint { +public: + std::string operator()(testing::TestParamInfo dev) const { + if (dev.param.is_cpu()) + return std::string("CPU"); + if (dev.param.is_host()) + return std::string("HOST"); + if (dev.param.is_gpu()) { + unsigned int vendor_id = + static_cast(dev.param.get_info()); + switch (vendor_id) { + case INTEL_ID: + return std::string("INTELGPU"); + } + } + if (dev.param.is_accelerator()) + return std::string("ACCELERATOR"); + return std::string("UNKNOWN"); + } +}; + +#endif // _TEST_HELPER_HPP_ diff --git a/tests/unit_tests/main_test.cpp b/tests/unit_tests/main_test.cpp new file mode 100644 index 000000000..9c95299b7 --- /dev/null +++ b/tests/unit_tests/main_test.cpp @@ -0,0 +1,176 @@ +/******************************************************************************* +* Copyright 2020 Intel Corporation +* +* Licensed under the Apache License, Version 2.0 (the "License"); +* you may not use this file except in compliance with the License. +* You may obtain a copy of the License at +* +* http://www.apache.org/licenses/LICENSE-2.0 +* +* Unless required by applicable law or agreed to in writing, +* software distributed under the License is distributed on an "AS IS" BASIS, +* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +* See the License for the specific language governing permissions +* and limitations under the License. +* +* +* SPDX-License-Identifier: Apache-2.0 +*******************************************************************************/ + +#include +#include +#include +#include "config.hpp" +#include "onemkl/onemkl.hpp" + +#define MAX_STR 128 + +using ::testing::EmptyTestEventListener; +using ::testing::InitGoogleTest; +using ::testing::Test; +using ::testing::TestCase; +using ::testing::TestEventListeners; +using ::testing::TestInfo; +using ::testing::TestPartResult; +using ::testing::UnitTest; + +cl::sycl::device host, cpu, gpu; +std::vector devices; + +std::string gtestInFile; + +namespace { +// Provides alternative output mode which produces minimal amount of +// information about tests. +class TersePrinter : public EmptyTestEventListener { +private: + // Called before any test activity starts. + void OnTestProgramStart(const UnitTest& /* unit_test */) override {} + + // Called after all test activities have ended. + void OnTestProgramEnd(const UnitTest& unit_test) override { + fprintf(stdout, "TEST %s\n", unit_test.Passed() ? "PASSED" : "FAILED"); + fflush(stdout); + } + + // Called before a test starts. + void OnTestStart(const TestInfo& test_info) override { + fprintf(stdout, "*** Test %s.%s starting.\n", test_info.test_case_name(), test_info.name()); + fflush(stdout); + } + + // Called after a failed assertion or a SUCCEED() invocation. + void OnTestPartResult(const TestPartResult& test_part_result) override { + const char* file_name = test_part_result.file_name(); + fprintf(stdout, "%s in %s:%d\n%s\n", test_part_result.failed() ? "*** Failure" : "Success", + file_name ? file_name : "unknown file", test_part_result.line_number(), + test_part_result.summary()); + fflush(stdout); + } + + // Called after a test ends. + void OnTestEnd(const TestInfo& test_info) override { + fprintf(stdout, "*** Test %s.%s ending.\n", test_info.test_case_name(), test_info.name()); + fflush(stdout); + } +}; // class TersePrinter + +} // anonymous namespace + +int main(int argc, char** argv) { + std::set unique_devices; + + auto platforms = cl::sycl::platform::get_platforms(); + for (auto plat : platforms) { + if (!plat.is_host()) { + auto plat_devs = plat.get_devices(); + for (auto dev : plat_devs) { + try { + if (unique_devices.find(dev.get_info()) == + unique_devices.end()) { + unique_devices.insert(dev.get_info()); + unsigned int vendor_id = static_cast( + dev.get_info()); +#ifndef ENABLE_MKLCPU_BACKEND + if (dev.is_cpu()) + continue; +#endif +#ifndef ENABLE_MKLGPU_BACKEND + if (dev.is_gpu() && vendor_id == INTEL_ID) + continue; +#endif + if (!dev.is_accelerator()) + devices.push_back(dev); + } + } + catch (std::exception const& e) { + std::cout << "Exception while accessing device: " << e.what() << "\n"; + } + } + } + } + +#ifdef ENABLE_MKLCPU_BACKEND + devices.push_back(cl::sycl::device(cl::sycl::host_selector())); +#endif + + // start Google Test pickup and output + testing::InitGoogleTest(&argc, argv); + + bool terse_output = false; + if (argc > 1 && strcmp(argv[1], "--terse_output") == 0) + terse_output = true; + else + printf("%s\n", + "Run this program with --terse_output to change the way it prints its output."); + + for (int i = 0; i < argc; i++) { + if (strncmp(argv[i], "--input_file=", 13) == 0) { + std::string tmp(argv[i]); + gtestInFile = tmp.substr(13); + break; + } + } + + UnitTest& unit_test = *UnitTest::GetInstance(); + + // If we are given the --terse_output command line flag, suppresses the + // standard output and attaches own result printer. + if (terse_output) { + TestEventListeners& listeners = unit_test.listeners(); + + // Removes the default console output listener from the list so it will + // not receive events from Google Test and won't print any output. Since + // this operation transfers ownership of the listener to the caller we + // have to delete it as well. + delete listeners.Release(listeners.default_result_printer()); + + // Adds the custom output listener to the list. It will now receive + // events from Google Test and print the alternative output. We don't + // have to worry about deleting it since Google Test assumes ownership + // over it after adding it to the list. + listeners.Append(new TersePrinter); + } + int ret_val = RUN_ALL_TESTS(); + + // This is an example of using the UnitTest reflection API to inspect test + // results. Here we discount failures from the tests we expected to fail. + int unexpectedly_failed_tests = 0; + for (int i = 0; i < unit_test.total_test_case_count(); ++i) { + const TestCase& test_case = *unit_test.GetTestCase(i); + for (int j = 0; j < test_case.total_test_count(); ++j) { + const TestInfo& test_info = *test_case.GetTestInfo(j); + // Counts failed tests that were not meant to fail (those without + // 'Fails' in the name). + if (test_info.result()->Failed() && strcmp(test_info.name(), "Fails") != 0) { + unexpectedly_failed_tests++; + } + } + } + + // Test that were meant to fail should not affect the test program outcome. + if (unexpectedly_failed_tests == 0) + ret_val = 0; + + return ret_val; +}