Issue with compiling Tensorflow

[rahasayantan]

System information

Have I written custom code (as opposed to using a stock example script provided in TensorFlow): No

OS Platform and Distribution (e.g., Linux Ubuntu):14.04.3-->Trusty

TensorFlow installed from (source or binary):Source

TensorFlow version (use command below):1.0 (Steps-> Downloaded tensorflow from https://github.com/benoitsteiner/tensorflow-opencl, ./configure - to configure project)

Bazel version (if compiling from source):0.5.4

CUDA/cuDNN version:NA

OPENCL Version:
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.0 AMD-APP (1800.11)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices

GPU model and memory:
Platform Name: AMD Accelerated Parallel Processing
Number of devices: 2
Device Type: CL_DEVICE_TYPE_GPU
Board name: AMD Radeon (TM) R5 M335
Memory: 4096M

Exact command to reproduce:
run the python script -- ipython keras_code.py

** G++/GCC version**:
g++-4.8

I have compiled CPP programs, they work fine.

ComputeCPP: 0.3.4
-- ** Python**: I am using Anaconda distribution Python for 2.7.2. (Anaconda - 2.4.3)

Describe the problem:

I have tried installing as per the steps given in :
https://www.codeplay.com/portal/03-30-17-setting-up-tensorflow-with-opencl-using-sycl

I have cloned the following git Repo:
git clone https://github.com/benoitsteiner/tensorflow-opencl.git

Everything works fine till the compile step: I use the following command:
bazel build --local_resources 2048,.5,1.0 -c opt --config=sycl //tensorflow/tools/pip_package:build_pip_package

================

Error

ERROR: /home/sayantan/Downloads/tensorflow-opencl/tensorflow/core/kernels/BUILD:2512:1: C++ compilation of rule '//tensorflow/core/kernels:softmax_op' failed (Exit 1).
In file included from tensorflow/core/kernels/softmax_op.cc:20:
In file included from ./tensorflow/core/kernels/softmax_op.h:23:
In file included from ./third_party/eigen3/unsupported/Eigen/CXX11/Tensor:4:
In file included from external/eigen_archive/unsupported/Eigen/CXX11/Tensor:14:
In file included from external/eigen_archive/Eigen/Core:299:
In file included from external/local_config_sycl/crosstool/../sycl/include/SYCL/sycl.hpp:20:
In file included from external/local_config_sycl/crosstool/../sycl/include/SYCL/sycl_interface.h:54:
external/local_config_sycl/crosstool/../sycl/include/SYCL/multi_pointer.h:342:3: error: multiple overloads of 'global_ptr' instantiate to the same signature 'void (pointer_t)' (aka 'void (attribute((address_space(1))) float *)')
global_ptr(pointer_t ptr) : Base(ptr) {}
^
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorSyclExprConstructor.h:42:74: note: in instantiation of template class 'cl::sycl::global_ptr<attribute((address_space(1))) float>' requested here
EvalToLHSConstructor(const utility::tuple::Tuple<Params...> &t) : expr(ConvertToActualTypeSycl(typename Eigen::internal::remove_all::type, utility::tuple::get(t))) {}
^
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorDeviceSycl.h:31:88: note: expanded from macro 'ConvertToActualTypeSycl'
#define ConvertToActualTypeSycl(Scalar, buf_acc) reinterpret_cast<typename cl::sycl::global_ptr::pointer_t>((&(*buf_acc.get_pointer())))
^
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorSyclExprConstructor.h:226:1: note: in instantiation of member function 'Eigen::TensorSycl::internal::EvalToLHSConstructor<attribute((address_space(1))) float *, 0, cl::sycl::accessor<unsigned char, 1, cl::sycl::access::mode::write, cl::sycl::access::target::global_buffer, cl::sycl::codeplay::access::placeholder::false_t>, cl::sycl::accessor<unsigned char, 1, cl::sycl::access::mode::read, cl::sycl::access::target::global_buffer, cl::sycl::codeplay::access::placeholder::false_t> >::EvalToLHSConstructor' requested here
EVALTO(const)
^
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorSyclExprConstructor.h:223:41: note: expanded from macro 'EVALTO'
: nestedExpression(funcD.xprExpr, t), buffer(t), expr(buffer.expr, nestedExpression.expr) {}
^
external/eigen_archive/unsupported/Eigen/CXX11/src/Tensor/TensorSyclExprConstructor.h:506:10: note: in instantiation of function template specialization 'Eigen::TensorSycl::internal::ExprConstructor<const Eigen::TensorEvalToOp<const Eigen::TensorReductionOp<Eigen::internal::MaxReducer, const Eigen::IndexList<Eigen::type2index<1>>, const Eigen::TensorMap<Eigen::Tensor<const float, 2, 1, long>, 16, MakeGlobalPointer>, MakeGlobalPointer>, MakeGlobalPointer>, const Eigen::TensorEvalToOp<const Eigen::TensorSycl::internal::PlaceHolder<const Eigen::TensorReductionOp<Eigen::internal::MaxReducer, const Eigen::IndexList<Eigen::type2index<1>>, const Eigen::TensorMap<Eigen::Tensor<const float, 2, 1, long>, 16, MakePointer>, MakePointer>, 1>, MakePointer>, cl::sycl::accessor<unsigned char, 1, cl::sycl::access::mode::write, cl::sycl::access::target::global_buffer, cl::sycl::codeplay::access::placeholder::false_t>, cl::sycl::accessor<unsigned char, 1, cl::sycl::access::mode::read, cl::sycl::access::target::global_buffer, cl::sycl::codeplay::access::placeholder::false_t> >::ExprConstructor<Eigen::TensorSycl::internal::FunctorExtractor<Eigen::TensorEvaluator<const Eigen::TensorEvalToOp<const Eigen::TensorReductionOp<Eigen::internal::MaxReducer, const Eigen::IndexList<Eigen::type2index<1>>, const Eigen::TensorMap<Eigen::Tensor<const float, 2, 1, long>, 16, MakePointer>, MakePointer>, MakePointer>, const Eigen::SyclDevice> > >' requested here
return ExprConstructor<OrigExpr, IndexExpr, Params...>(funcD, t);
.....

2 errors generated.
Target //tensorflow/tools/pip_package:build_pip_package failed to build
Use --verbose_failures to see the command lines of failed build steps.
INFO: Elapsed time: 3715.231s, Critical Path: 167.06s

Can you please help?

Thanks and regards
Sayantan Raha

[mirh]

This has been specifically mentioned as fixed on official upstream.
Can you try it or this?