Add chicoma-gpu, GPU nodes on LANL Chicoma

E3SM-Ocean-Discussion · Feb 3, 2024 · b5db2b8 · b5db2b8
1 parent 6e7b8e6
commit b5db2b8
Show file tree

Hide file tree

Showing 5 changed files with 184 additions and 1 deletion.
diff --git a/cime_config/machines/Depends.chicoma-gpu.nvidia.cmake b/cime_config/machines/Depends.chicoma-gpu.nvidia.cmake
@@ -0,0 +1,10 @@
+list(APPEND REDUCE_OPT_LIST
+  homme/src/share/derivative_mod_base.F90
+)
+
+# Can use this flag to avoid internal compiler error for this file (with nvidia/21.11)
+if (NOT DEBUG)
+  foreach(ITEM IN LISTS REDUCE_OPT_LIST)
+    e3sm_add_flags("${ITEM}" " -Mnovect")
+  endforeach()
+endif()
diff --git a/cime_config/machines/Depends.chicoma-gpu.nvidiagpu.cmake b/cime_config/machines/Depends.chicoma-gpu.nvidiagpu.cmake
@@ -0,0 +1,10 @@
+list(APPEND REDUCE_OPT_LIST
+  homme/src/share/derivative_mod_base.F90
+)
+
+# Can use this flag to avoid internal compiler error for this file (with nvidia/21.11)
+if (NOT DEBUG)
+  foreach(ITEM IN LISTS REDUCE_OPT_LIST)
+    e3sm_add_flags("${ITEM}" " -Mnovect")
+  endforeach()
+endif()
diff --git a/cime_config/machines/config_batch.xml b/cime_config/machines/config_batch.xml
@@ -649,7 +649,35 @@
      <directive>--qos=standard </directive>
     </directives>
     <queues>
-      <queue walltimemax="24:00:00" nodemax="1792" default="true">standard</queue>
+      <queue walltimemax="16:00:00" nodemax="1792" default="true">standard</queue>
+    </queues>
+   </batch_system>
+
+   <batch_system MACH="chicoma-gpu" type="slurm">
+    <directives>
+      <directive> --partition=gpu</directive>
+    </directives>
+    <directives compiler="gnugpu">
+      <directive> --gpus-per-task=1</directive>
+    </directives>
+    <directives COMPSET="!.*MMF.*" compiler="gnugpu">
+      <directive> --gpu-bind=none</directive>
+    </directives>
+    <directives COMPSET=".*MMF.*" compiler="gnugpu">
+      <directive> --gpu-bind=map_gpu:0,1,2,3</directive>
+    </directives>
+    <directives compiler="nvidiagpu">
+      <directive> --gpus-per-task=1</directive>
+      <directive> --gpu-bind=none</directive>
+    </directives>
+    <directives compiler="gnu">
+      <directive> -G 0</directive>
+    </directives>
+    <directives compiler="nvidia">
+      <directive> -G 0</directive>
+    </directives>
+    <queues>
+      <queue walltimemax="16:00:00" nodemax="96" default="true">standard</queue>
     </queues>
    </batch_system>
 

diff --git a/cime_config/machines/config_machines.xml b/cime_config/machines/config_machines.xml
@@ -4229,6 +4229,140 @@
     </resource_limits>
   </machine>
 
+  <machine MACH="chicoma-gpu">
+    <DESC>Chicoma GPU nodes at LANL IC. Each GPU node has single 
+AMD EPYC 7713 64-Core (Milan) (256GB) and 4 nvidia A100'</DESC>
+    <NODENAME_REGEX>ch-fe*</NODENAME_REGEX>
+    <OS>Linux</OS>
+    <COMPILERS>gnugpu,gnu,nvidiagpu,nvidia</COMPILERS>
+    <MPILIBS>mpich</MPILIBS>
+    <CIME_OUTPUT_ROOT>/lustre/scratch4/turquoise/$ENV{USER}/E3SM/scratch/chicoma-cpu</CIME_OUTPUT_ROOT>
+    <DIN_LOC_ROOT>/usr/projects/e3sm/inputdata</DIN_LOC_ROOT>
+    <DIN_LOC_ROOT_CLMFORC>/usr/projects/e3sm/inputdata/atm/datm7</DIN_LOC_ROOT_CLMFORC>
+    <DOUT_S_ROOT>/lustre/scratch4/turquoise/$ENV{USER}/E3SM/archive/$CASE</DOUT_S_ROOT>
+    <BASELINE_ROOT>/lustre/scratch4/turquoise/$ENV{USER}/E3SM/input_data/ccsm_baselines/$COMPILER</BASELINE_ROOT>
+    <CCSM_CPRNC>/usr/projects/climate/SHARED_CLIMATE/software/badger/cprnc</CCSM_CPRNC>
+    <GMAKE_J>10</GMAKE_J>
+    <TESTS>e3sm_developer</TESTS>
+    <NTEST_PARALLEL_JOBS>4</NTEST_PARALLEL_JOBS>
+    <BATCH_SYSTEM>slurm</BATCH_SYSTEM>
+    <SUPPORTED_BY>e3sm</SUPPORTED_BY>
+    <MAX_TASKS_PER_NODE>128</MAX_TASKS_PER_NODE>
+    <MAX_TASKS_PER_NODE compiler="gnu">256</MAX_TASKS_PER_NODE>
+    <MAX_TASKS_PER_NODE compiler="nvidia">256</MAX_TASKS_PER_NODE>
+    <MAX_MPITASKS_PER_NODE>4</MAX_MPITASKS_PER_NODE>
+    <MAX_MPITASKS_PER_NODE compiler="gnu">64</MAX_MPITASKS_PER_NODE>
+    <MAX_MPITASKS_PER_NODE compiler="nvidia">64</MAX_MPITASKS_PER_NODE>
+    <PROJECT_REQUIRED>TRUE</PROJECT_REQUIRED>
+    <mpirun mpilib="default">
+      <executable>srun</executable>
+      <arguments>
+        <arg name="label"> --label</arg>
+        <arg name="num_tasks"> -n {{ total_tasks }} -N {{ num_nodes }}</arg>
+        <arg name="thread_count">-c $SHELL{echo 256/`./xmlquery --value MAX_MPITASKS_PER_NODE`|bc}</arg>
+        <arg name="binding"> $SHELL{if [ 128 -ge `./xmlquery --value MAX_MPITASKS_PER_NODE` ]; then echo "--cpu_bind=cores"; else echo "--cpu_bind=threads";fi;}</arg>
+        <arg name="placement"> -m plane=$SHELL{echo `./xmlquery --value MAX_MPITASKS_PER_NODE`}</arg>
+    </arguments>
+    </mpirun>
+    <module_system type="module" allow_error="true">
+      <init_path lang="perl">/usr/share/lmod/8.3.1/init/perl</init_path>
+      <!-- does not exist -->
+      <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="perl">/usr/share/lmod/lmod/libexec/lmod perl</cmd_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="unload">cray-hdf5-parallel</command>
+        <command name="unload">cray-netcdf-hdf5parallel</command>
+        <command name="unload">cray-parallel-netcdf</command>
+        <command name="unload">cray-netcdf</command>
+        <command name="unload">cray-hdf5</command>
+        <command name="unload">PrgEnv-gnu</command>
+        <command name="unload">PrgEnv-intel</command>
+        <command name="unload">PrgEnv-nvidia</command>
+        <command name="unload">PrgEnv-cray</command>
+        <command name="unload">PrgEnv-aocc</command>
+        <command name="unload">intel</command>
+        <command name="unload">intel-oneapi</command>
+        <command name="unload">nvidia</command>
+        <command name="unload">aocc</command>
+        <command name="unload">cudatoolkit</command>
+        <command name="unload">climate-utils</command>
+        <command name="unload">craype-accel-nvidia80</command>
+        <command name="unload">craype-accel-host</command>
+        <command name="unload">perftools-base</command>
+        <command name="unload">perftools</command>
+        <command name="unload">darshan</command>
+      </modules>
+
+      <modules compiler="gnu.*">
+        <command name="load">PrgEnv-gnu/8.4.0</command>
+        <command name="load">gcc/12.2.0</command>
+      </modules>
+
+      <modules compiler="nvidia.*">
+        <command name="load">PrgEnv-nvidia/8.4.0</command>
+        <command name="load">nvidia/22.7</command>
+      </modules>
+
+      <modules compiler="gnugpu">
+        <command name="load">cudatoolkit/22.7_11.7</command>
+        <command name="load">craype-accel-nvidia80</command>
+      </modules>
+
+      <modules compiler="nvidiagpu">
+        <command name="load">cudatoolkit/22.7_11.7</command>
+        <command name="load">craype-accel-nvidia80</command>
+        <command name="load">gcc-mixed/12.2.0</command>
+      </modules>
+
+      <modules compiler="gnu">
+        <command name="load">craype-accel-host</command>
+      </modules>
+
+      <modules compiler="nvidia">
+        <command name="load">craype-accel-host</command>
+      </modules>
+
+      <modules>
+        <command name="load">cray-libsci/23.05.1.4</command>
+        <command name="load">craype/2.7.21</command>
+        <command name="load">cray-mpich/8.1.26</command>
+        <command name="load">libfabric/1.15.2.0</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>
+        <command name="load">cmake/3.25.1</command>
+      </modules>
+    </module_system>
+
+    <RUNDIR>$CIME_OUTPUT_ROOT/$CASE/run</RUNDIR>
+    <EXEROOT>$CIME_OUTPUT_ROOT/$CASE/bld</EXEROOT>
+    <TEST_TPUT_TOLERANCE>0.1</TEST_TPUT_TOLERANCE>
+
+    <environment_variables>
+      <env name="MPICH_ENV_DISPLAY">1</env>
+      <env name="MPICH_VERSION_DISPLAY">1</env>
+      <env name="OMP_STACKSIZE">128M</env>
+      <env name="OMP_PROC_BIND">spread</env>
+      <env name="OMP_PLACES">threads</env>
+      <env name="HDF5_USE_FILE_LOCKING">FALSE</env>
+      <env name="PERL5LIB">/usr/projects/climate/SHARED_CLIMATE/software/chicoma-cpu/perl5-only-switch/lib/perl5</env>
+      <env name="PNETCDF_HINTS">romio_ds_write=disable;romio_ds_read=disable;romio_cb_write=enable;romio_cb_read=enable</env>
+      <env name="FI_CXI_RX_MATCH_MODE">software</env>
+      <env name="MPICH_COLL_SYNC">MPI_Bcast</env>
+      <env name="NETCDF_PATH">$ENV{CRAY_NETCDF_HDF5PARALLEL_PREFIX}</env>
+      <env name="PNETCDF_PATH">$ENV{CRAY_PARALLEL_NETCDF_PREFIX}</env>
+    </environment_variables>
+    <resource_limits>
+      <resource name="RLIMIT_STACK">-1</resource>
+    </resource_limits>
+  </machine>
+
   <machine MACH="mesabi">
     <DESC>Mesabi batch queue</DESC>
     <OS>LINUX</OS>

diff --git a/cime_config/machines/config_pio.xml b/cime_config/machines/config_pio.xml
@@ -66,6 +66,7 @@
       <value mach="grizzly">netcdf</value>
       <value mach="badger">netcdf</value>
       <value mach="chicoma-cpu">netcdf</value>
+      <value mach="chicoma-gpu">netcdf</value>
       <value mach="bebop" mpilib="impi" compset=".*CAM5.+MPAS.*">netcdf</value>
       <value mach="fugaku" compiler="gnu">netcdf</value>
     </values>