Merge branch 'develop' into experiment/kripke

.github/workflows/bin/license

@@ -24,7 +24,7 @@ licensed_files = [
     r"^configs\/[^\/]*\/[^\/]*",
     r"^docs\/[^\/]*\.rst$",
     r"^experiments\/[^\/]*\/[^\/]*",
-    r"^repo\/[^\/]*\/[^\/]*",
+    r"^repo\/[^\/]*\/[^\/]*(?<!\.patch)$",
 ]
 
 

.github/workflows/requirements/style.txt

@@ -1,4 +1,4 @@
-black==23.12.1
+black==24.1.1
 flake8==7.0.0
 isort==5.13.2
 codespell==2.2.6

configs/LLNL-Sierra-IBM-power9-V100-Infiniband/spack.yaml

@@ -11,11 +11,7 @@ spack:
       spack_spec: [email protected]
     compiler-gcc:
       spack_spec: [email protected]
-    compiler-cuda10:
-      spack_spec: [email protected]{cuda10_version}
-    cublas-cuda-10:
-      spack_spec: cublas@{cuda10_version}
-    cublas-cuda-default:
+    cublas-cuda:
       spack_spec: cublas@{default_cuda_version} 
     lapack:
       spack_spec: [email protected]

configs/LLNL-Sierra-IBM-power9-V100-Infiniband/variables.yaml

@@ -11,7 +11,5 @@ variables:
   batch_nodes: '#BSUB -nnodes {n_nodes}'
   batch_ranks: '#BSUB -nnodes {n_nodes}'
   batch_timeout: '#BSUB -W {batch_time}'
-  cuda10_version: '10.1.243'
   default_cuda_version: '11.8.0'
   cuda_arch: '70'
-  enable_mps: '/usr/tcetmp/bin/enable_mps'

docs/add-a-site-specific-system-config.rst

@@ -3,9 +3,9 @@
 
    SPDX-License-Identifier: Apache-2.0
 
-=============================
+======================================
 Adding a Specific System Configuration
-=============================
+======================================
 
 For a specific system, one can (optionally) add more information about the software installed on the system
 by adding Spack config files in ``benchpark/configs/$SITE/SYSTEMNAME-GENERICSYSTEM/auxiliary_software_files/``.

docs/add-a-system-config.rst

@@ -111,4 +111,4 @@ spack section in the `Ramble configuration file
       batch_timeout: ''
 
 If defining a specific system, one can be more specific with available software versions
-and packages, as demonstrated in :doc:`add-a-specific-system-config.rst`.
+and packages, as demonstrated in :doc:`add-a-site-specific-system-config`.

docs/tables/current-system-definitions.csv

@@ -1,20 +1,20 @@
-,Magma,Sierra,Tioga,AWS_PCluster_Hpc7a-zen4-EFA,HPECray-zen3-MI250X-Slingshot,x86_64
-**site**,LLNL,LLNL,LLNL,,,
-**system**,Penguin-icelake-OmniPath,IBM-power9-V100-Infiniband,HPECray-zen3-MI250X-Slingshot,AWS_PCluster_Hpc7a-zen4-EFA,HPECray-zen3-MI250X-Slingshot,x86_64
-**integrator.vendor**,PenguinComputing,IBM,HPECray,AWS,HPECray,
-**integrator.name**,RelionCluster,AC922,EX235a,ParallelCluster3.7.2-Hpc7a,EX235a,
-**processor.vendor**,Intel,IBM,AMD,AMD,AMD,
-**processor.name**,XeonPlatinum924248C,POWER9,EPYC-Zen3,EPYC-Zen4,EPYC-Zen3,
-**processor.ISA**,x86_64,ppc64le,x86_64,x86_64,x86_64,x86_64
-**processor.uArch**,icelake,power9,zen3,zen4,zen3,
-**accelerator.vendor**,,NVIDIA,AMD,,AMD,
-**accelerator.name**,,V100,MI250X,,MI250X,
-**accelerator.ISA**,,PTX,GCN,,GCN,
-**accelerator.uArch**,,sm_70,gfx90a,,gfx90a,
-**interconnect.vendor**,Intel,Mellanox,HPECray,AWS,HPECray,
-**interconnect.name**,OmniPath,EDR-Infiniband,Slingshot11,EFA,Slingshot11,
-**system-tested.site**,LLNL,LLNL,LLNL,AWS,LLNL,
-**system-tested.name**,magma,lassen,tioga,,tioga,
-**system-tested.installation-year**,2019,2018,2022,,2022,
-**system-tested.description**,`top500 <https://www.top500.org/system/179774>`_,`top500 <https://www.top500.org/system/179398>`_,`top500 <https://www.top500.org/system/180052>`_,`aws/hpc7a <https://aws.amazon.com/ec2/instance-types/hpc7a>`_,`top500 <https://www.top500.org/system/180052>`_,
-**top500-system-instances**,Magma (LLNL),Sierra (LLNL),"Frontier (ORNL), Lumi     (CSC), Tioga    (LLNL)",,"Frontier (ORNL), Lumi     (CSC), Tioga    (LLNL)",
+,Sierra,AWS_PCluster_Hpc7a-zen4-EFA,HPECray-zen3-MI250X-Slingshot,Magma,Tioga,x86_64
+**site**,LLNL,,,LLNL,LLNL,
+**system**,IBM-power9-V100-Infiniband,AWS_PCluster_Hpc7a-zen4-EFA,HPECray-zen3-MI250X-Slingshot,Penguin-icelake-OmniPath,HPECray-zen3-MI250X-Slingshot,x86_64
+**integrator.vendor**,IBM,AWS,HPECray,PenguinComputing,HPECray,
+**integrator.name**,AC922,ParallelCluster3.7.2-Hpc7a,EX235a,RelionCluster,EX235a,
+**processor.vendor**,IBM,AMD,AMD,Intel,AMD,
+**processor.name**,POWER9,EPYC-Zen4,EPYC-Zen3,XeonPlatinum924248C,EPYC-Zen3,
+**processor.ISA**,ppc64le,x86_64,x86_64,x86_64,x86_64,x86_64
+**processor.uArch**,power9,zen4,zen3,icelake,zen3,
+**accelerator.vendor**,NVIDIA,,AMD,,AMD,
+**accelerator.name**,V100,,MI250X,,MI250X,
+**accelerator.ISA**,PTX,,GCN,,GCN,
+**accelerator.uArch**,sm_70,,gfx90a,,gfx90a,
+**interconnect.vendor**,Mellanox,AWS,HPECray,Intel,HPECray,
+**interconnect.name**,EDR-Infiniband,EFA,Slingshot11,OmniPath,Slingshot11,
+**system-tested.site**,LLNL,AWS,LLNL,LLNL,LLNL,
+**system-tested.name**,lassen,,tioga,magma,tioga,
+**system-tested.installation-year**,2018,,2022,2019,2022,
+**system-tested.description**,`top500 <https://www.top500.org/system/179398>`_,`aws/hpc7a <https://aws.amazon.com/ec2/instance-types/hpc7a>`_,`top500 <https://www.top500.org/system/180052>`_,`top500 <https://www.top500.org/system/179774>`_,`top500 <https://www.top500.org/system/180052>`_,
+**top500-system-instances**,Sierra (LLNL),,"Frontier (ORNL), Lumi     (CSC), Tioga    (LLNL)",Magma (LLNL),"Frontier (ORNL), Lumi     (CSC), Tioga    (LLNL)",

experiments/amg2023/cuda/execute_experiment.tpl

@@ -11,6 +11,4 @@ cd {experiment_run_dir}
 
 {spack_setup}
 
-{experiment_setup} #TODO: Add experiment specific code here
-
 {command}

experiments/amg2023/cuda/ramble.yaml

@@ -27,21 +27,11 @@ ramble:
             gtl: ['gtl', 'nogtl']
             gtlflag: ['-M"-gpu"', '']
           experiments:
-            # because MPS only works for cuda<=v10
-            amg2023_cuda_10_problem1_{gtl}_{n_nodes}_{px}_{py}_{pz}_{nx}_{ny}_{nz}:
+            amg2023_cuda_problem1_{gtl}_{n_nodes}_{px}_{py}_{pz}_{nx}_{ny}_{nz}:
               variables:
-                experiment_setup: '{enable_mps}'
-                env_name: amg2023-cuda-10
-                processes_per_node: '8'
-                n_nodes: '1'
-              matrices:
-                - size:
-                  - n
-            amg2023_cuda_default_problem1_{gtl}_{n_nodes}_{px}_{py}_{pz}_{nx}_{ny}_{nz}:
-              variables:
-                env_name: amg2023-cuda-default
-                processes_per_node: ['4', '2']
-                n_nodes: ['2', '4']
+                env_name: amg2023
+                processes_per_node: '4'
+                n_nodes: '2'
               zips:
                 gtl_info:
                 - gtl
@@ -52,22 +42,18 @@ ramble:
                   - gtl_info
   spack:
     concretized: true
-    variables:
-      cuda_version: ['{cuda10_version}', '{default_cuda_version}']
-      cuda_compiler: ['compiler-cuda10', 'default-compiler']
-      versions: ['10', 'default']
     packages:
-      hypre-cuda-{versions}:
-        spack_spec: '[email protected] +mpi+cuda+mixedint cuda_arch=={cuda_arch} ^cuda@{cuda_version}'
-        compiler: '{cuda_compiler}'
-      amg2023-cuda-{versions}:
-        spack_spec: 'amg2023@develop +mpi+cuda ^cuda@{cuda_version}'
-        compiler: '{cuda_compiler}'
+      hypre:
+        spack_spec: [email protected] +mpi+cuda+mixedint cuda_arch=={cuda_arch} ^cuda@{default_cuda_version}
+        compiler: default-compiler
+      amg2023:
+        spack_spec: amg2023@develop +mpi+cuda ^cuda@{default_cuda_version}
+        compiler: default-compiler
     environments:
-      amg2023-cuda-{versions}:
+      amg2023:
         packages:
-        - cublas-cuda-{versions}
+        - cublas-cuda
         - lapack
         - default-mpi
-        - hypre-cuda-{versions}
-        - amg2023-cuda-{versions}
+        - hypre
+        - amg2023

experiments/saxpy/cuda/execute_experiment.tpl

@@ -5,7 +5,6 @@
 # SPDX-License-Identifier: Apache-2.0
 
 {batch_nodes}
-{batch_ranks}
 {batch_timeout}
 
 cd {experiment_run_dir}

repo/gromacs/gmxDetectCpu-cmake-3.14.patch

@@ -0,0 +1,12 @@
+--- a/cmake/gmxDetectCpu.cmake
++++ b/cmake/gmxDetectCpu.cmake
+@@ -83,7 +83,7 @@ function(gmx_run_cpu_detection TYPE)
+                 set(GCC_INLINE_ASM_DEFINE "-DGMX_X86_GCC_INLINE_ASM=0")
+             endif()
+
+-            set(_compile_definitions "${GCC_INLINE_ASM_DEFINE} -I${PROJECT_SOURCE_DIR}/src -DGMX_CPUINFO_STANDALONE ${GMX_STDLIB_CXX_FLAGS} -DGMX_TARGET_X86=${GMX_TARGET_X86_VALUE}")
++            set(_compile_definitions ${GCC_INLINE_ASM_DEFINE} -I${PROJECT_SOURCE_DIR}/src -DGMX_CPUINFO_STANDALONE ${GMX_STDLIB_CXX_FLAGS} -DGMX_TARGET_X86=${GMX_TARGET_X86_VALUE})
+             set(LINK_LIBRARIES "${GMX_STDLIB_LIBRARIES}")
+             try_compile(CPU_DETECTION_COMPILED
+                 "${PROJECT_BINARY_DIR}"
+

repo/gromacs/gmxDetectSimd-cmake-3.14.patch

@@ -0,0 +1,11 @@
+--- a/cmake/gmxDetectSimd.cmake
++++ b/cmake/gmxDetectSimd.cmake
+@@ -77,7 +77,7 @@ function(gmx_suggest_simd _suggested_simd)
+     else()
+         set(GMX_TARGET_X86_VALUE 0)
+     endif()
+-    set(_compile_definitions "${GCC_INLINE_ASM_DEFINE} -I${CMAKE_SOURCE_DIR}/src -DGMX_CPUINFO_STANDALONE ${GMX_STDLIB_CXX_FLAGS} -DGMX_TARGET_X86=${GMX_TARGET_X86_VALUE}")
++    set(_compile_definitions ${GCC_INLINE_ASM_DEFINE} -I${CMAKE_SOURCE_DIR}/src -DGMX_CPUINFO_STANDALONE ${GMX_STDLIB_CXX_FLAGS} -DGMX_TARGET_X86=${GMX_TARGET_X86_VALUE})
+
+     # Prepare a default suggestion
+     set(OUTPUT_SIMD "None")