Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add ALCF Polaris to E3SM machines #6115

Merged
merged 7 commits into from
Feb 13, 2024
Merged

Add ALCF Polaris to E3SM machines #6115

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
6f8bd21
add polaris machine
xyuan Dec 6, 2023
bd072a2
Update Polaris config
amametjanov Dec 9, 2023
4edde11
Update paths and queues
amametjanov Dec 20, 2023
74a78e7
Update paths and modules, add gnu config
amametjanov Jan 25, 2024
359013d
Update gnugpu, nvidia and nvidiagpu cmake and modules
amametjanov Jan 31, 2024
66752a8
Set E3SMinput project paths and 1:1 MPI-to-GPU affinity
amametjanov Feb 1, 2024
1fa0cfa
Disable gpu env-vars in cpu runs; append cuda path to link flags
amametjanov Feb 6, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions cime_config/machines/cmake_macros/gnu_polaris.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,10 @@
string(APPEND CONFIG_ARGS " --host=cray")
if (COMP_NAME STREQUAL gptl)
string(APPEND CPPDEFS " -DHAVE_NANOTIME -DBIT64 -DHAVE_SLASHPROC -DHAVE_GETTIMEOFDAY")
endif()
set(MPICC "cc")
set(MPICXX "CC")
set(MPIFC "ftn")
set(SCC "gcc")
set(SCXX "g++")
set(SFC "gfortran")
18 changes: 18 additions & 0 deletions cime_config/machines/cmake_macros/gnugpu_polaris.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
string(APPEND CONFIG_ARGS " --host=cray")
if (COMP_NAME STREQUAL gptl)
string(APPEND CPPDEFS " -DHAVE_NANOTIME -DBIT64 -DHAVE_SLASHPROC -DHAVE_GETTIMEOFDAY")
endif()
set(USE_CUDA "TRUE")
string(APPEND CMAKE_C_FLAGS_RELEASE " -O2")
string(APPEND CMAKE_CXX_FLAGS_RELEASE " -O2")
string(APPEND CMAKE_Fortran_FLAGS_RELEASE " -O2")
string(APPEND CMAKE_CUDA_FLAGS " -ccbin CC -O2 -arch sm_80 --use_fast_math")
string(APPEND KOKKOS_OPTIONS " -DKokkos_ARCH_AMPERE80=On -DKokkos_ENABLE_CUDA=On -DKokkos_ENABLE_CUDA_LAMBDA=On -DKokkos_ENABLE_SERIAL=ON -DKokkos_ENABLE_OPENMP=Off")
set(E3SM_LINK_WITH_FORTRAN "TRUE")
set(MPICC "cc")
set(MPICXX "CC")
set(MPIFC "ftn")
set(SCC "cc")
set(SCXX "CC")
set(SFC "ftn")
string(APPEND CMAKE_EXE_LINKER_FLAGS " -L/soft/compilers/cudatoolkit/cuda-11.4.4/lib64/ ")
7 changes: 7 additions & 0 deletions cime_config/machines/cmake_macros/nvidia_polaris.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,7 @@
string(APPEND CONFIG_ARGS " --host=cray")
if (COMP_NAME STREQUAL gptl)
string(APPEND CPPDEFS " -DHAVE_NANOTIME -DBIT64 -DHAVE_SLASHPROC -DHAVE_GETTIMEOFDAY")
endif()
set(SCC "cc")
set(SCXX "CC")
set(SFC "ftn")
13 changes: 13 additions & 0 deletions cime_config/machines/cmake_macros/nvidiagpu_polaris.cmake
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
string(APPEND CONFIG_ARGS " --host=cray")
if (COMP_NAME STREQUAL gptl)
string(APPEND CPPDEFS " -DHAVE_NANOTIME -DBIT64 -DHAVE_SLASHPROC -DHAVE_GETTIMEOFDAY")
endif()
set(USE_CUDA "TRUE")
string(APPEND CPPDEFS " -DGPU -DMPAS_OPENACC")
string(APPEND CPPDEFS " -DTHRUST_IGNORE_CUB_VERSION_CHECK")
string(APPEND CUDA_FLAGS " -ccbin CC -O2 -arch sm_80 --use_fast_math")
string(APPEND CMAKE_EXE_LINKER_FLAGS " -acc=gpu -gpu=cc80,cuda11.0 -Minfo=accel")
set(SCC "cc")
set(SCXX "CC")
set(SFC "ftn")
string(APPEND CMAKE_Fortran_FLAGS " -acc=gpu -gpu=cc80,cuda11.0 -Minfo=accel")
12 changes: 12 additions & 0 deletions cime_config/machines/config_batch.xml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -560,6 +560,18 @@
</queues>
</batch_system>

<batch_system MACH="polaris" type="pbspro">
<batch_submit>/grand/E3SMinput/soft/qsub/throttle</batch_submit>
<directives>
<directive> -l filesystems=home:grand:eagle </directive>
</directives>
<queues>
<queue walltimemax="00:30:00" nodemin="1" nodemax="2" strict="true" default="true">debug</queue>
<queue walltimemax="00:30:00" nodemin="3" nodemax="9" strict="true">debug-scaling</queue>
<queue walltimemax="01:00:00" nodemin="10" nodemax="496" strict="true">prod</queue>
</queues>
</batch_system>

<batch_system MACH="cascade" type="slurm">
<directives>
<directive>--output=slurm.out</directive>
Expand Down
95 changes: 95 additions & 0 deletions cime_config/machines/config_machines.xml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3135,6 +3135,101 @@
</resource_limits>
</machine>

<machine MACH="polaris">
<DESC>ALCF Polaris 560 nodes, 2.8 GHz AMD EPYC Milan 7543P 32c CPU, 4 NVIDIA A100 GPUs</DESC>
<NODENAME_REGEX>polaris-*</NODENAME_REGEX>
<OS>Linux</OS>
<COMPILERS>gnu,gnugpu,nvidia,nvidiagpu</COMPILERS>
<MPILIBS>mpich</MPILIBS>
<PROJECT>E3SM_RRM</PROJECT>
<CHARGE_ACCOUNT>E3SM_RRM</CHARGE_ACCOUNT>
<SAVE_TIMING_DIR>/grand/E3SMinput/performance_archive</SAVE_TIMING_DIR>
<SAVE_TIMING_DIR_PROJECTS>SCREAM_Calib,E3SM_RRM</SAVE_TIMING_DIR_PROJECTS>
<CIME_OUTPUT_ROOT>/eagle/$PROJECT/$USER/scratch</CIME_OUTPUT_ROOT>
<DIN_LOC_ROOT>/grand/E3SMinput/data</DIN_LOC_ROOT>
<DIN_LOC_ROOT_CLMFORC>/grand/E3SMinput/data/atm/datm7</DIN_LOC_ROOT_CLMFORC>
<DOUT_S_ROOT>$CIME_OUTPUT_ROOT/archive/$CASE</DOUT_S_ROOT>
<BASELINE_ROOT>/grand/E3SMinput/baselines/$COMPILER</BASELINE_ROOT>
<CCSM_CPRNC>/grand/E3SMinput/soft/cprnc/cprnc</CCSM_CPRNC>
<GMAKE_J>4</GMAKE_J>
<TESTS>e3sm_developer</TESTS>
<NTEST_PARALLEL_JOBS>4</NTEST_PARALLEL_JOBS>
<BATCH_SYSTEM>pbspro</BATCH_SYSTEM>
<SUPPORTED_BY>e3sm</SUPPORTED_BY>
<MAX_TASKS_PER_NODE>64</MAX_TASKS_PER_NODE>
<MAX_MPITASKS_PER_NODE>4</MAX_MPITASKS_PER_NODE>
<MAX_MPITASKS_PER_NODE compiler="gnu">32</MAX_MPITASKS_PER_NODE>
<MAX_MPITASKS_PER_NODE compiler="nvidia">32</MAX_MPITASKS_PER_NODE>
<PROJECT_REQUIRED>TRUE</PROJECT_REQUIRED>
<mpirun mpilib="default">
<executable>mpiexec</executable>
<arguments>
<arg name="total_num_tasks">-np {{ total_tasks }} --label</arg>
<arg name="ranks_per_node">-ppn {{ tasks_per_node }}</arg>
<arg name="ranks_bind">--cpu-bind depth -envall</arg>
<arg name="threads_per_rank">-d $ENV{OMP_NUM_THREADS}</arg>
<arg name="gpu_maps">$ENV{GPU_TILE_COMPACT}</arg>
</arguments>
</mpirun>
<module_system type="module" allow_error="true">
<init_path lang="python">/usr/share/lmod/8.3.1/init/python</init_path>
<init_path lang="sh">/usr/share/lmod/8.3.1/init/sh</init_path>
<init_path lang="csh">/usr/share/lmod/8.3.1/init/csh</init_path>
<cmd_path lang="python">/usr/share/lmod/lmod/libexec/lmod python</cmd_path>
<cmd_path lang="sh">module</cmd_path>
<cmd_path lang="csh">module</cmd_path>
<modules>
<command name="purge"/>
<command name="load">cmake/3.23.2</command>
<command name="load">craype-x86-rome</command>
</modules>
<modules compiler="gnu.*">
<command name="load">PrgEnv-gnu/8.3.3</command>
</modules>
<modules compiler="gnugpu">
<command name="swap">gcc/12.2.0 gcc/11.2.0</command>
<command name="load">cudatoolkit-standalone/11.4.4</command>
</modules>
<modules compiler="nvidia.*">
<command name="load">PrgEnv-nvhpc/8.3.3</command>
</modules>
<modules compiler="nvidiagpu">
<command name="load">cudatoolkit-standalone/11.4.4</command>
<command name="load">craype-accel-nvidia80</command>
</modules>
<modules>
<command name="load">craype-network-ofi</command>
<command name="load">libfabric/1.15.2.0</command>
<command name="load">cray-libsci/23.02.1.1</command>
<command name="load">cray-hdf5-parallel/1.12.2.3</command>
<command name="load">cray-netcdf-hdf5parallel/4.9.0.3</command>
<command name="load">cray-parallel-netcdf/1.12.3.3</command>
</modules>
</module_system>
<RUNDIR>$CIME_OUTPUT_ROOT/$CASE/run</RUNDIR>
<EXEROOT>$CIME_OUTPUT_ROOT/$CASE/bld</EXEROOT>
<environment_variables>
<env name="NETCDF_PATH">$ENV{CRAY_NETCDF_HDF5PARALLEL_PREFIX}</env>
<env name="PNETCDF_PATH">$ENV{CRAY_PARALLEL_NETCDF_PREFIX}</env>
<env name="MPICH_GPU_SUPPORT_ENABLED">0</env>
<env name="CRAY_ACCEL_TARGET">host</env>
<env name="GPU_TILE_COMPACT"> </env>
</environment_variables>
<environment_variables compiler=".*gpu">
<env name="MPICH_GPU_SUPPORT_ENABLED">1</env>
<env name="CRAY_ACCEL_TARGET">nvidia80</env>
<env name="GPU_TILE_COMPACT">/grand/E3SMinput/soft/qsub/set_affinity_gpu_polaris.sh</env>
</environment_variables>
<environment_variables compiler="gnu.*">
<env name="LD_PRELOAD">/opt/cray/pe/gcc/11.2.0/snos/lib64/libstdc++.so</env>
</environment_variables>
<environment_variables BUILD_THREADED="TRUE">
<env name="OMP_STACKSIZE">128M</env>
<env name="OMP_PROC_BIND">spread</env>
<env name="OMP_PLACES">threads</env>
</environment_variables>
</machine>

<machine MACH="sunspot">
<DESC>ANL Sunspot Test and Development System (TDS), batch system is pbspro</DESC>
<NODENAME_REGEX>uan-.*</NODENAME_REGEX>
Expand Down