From cc190d2c7acb848d93344d4adcf412141f20a3bf Mon Sep 17 00:00:00 2001 From: Xylar Asay-Davis Date: Fri, 12 Jul 2024 02:29:03 -0700 Subject: [PATCH 1/2] Add gpus_per_node to pm-gpu Fix incorrect cores_per_node --- mache/machines/pm-gpu.cfg | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/mache/machines/pm-gpu.cfg b/mache/machines/pm-gpu.cfg index 1dbb2447..bb52dd1a 100644 --- a/mache/machines/pm-gpu.cfg +++ b/mache/machines/pm-gpu.cfg @@ -50,8 +50,11 @@ system = slurm # whether to use mpirun or srun to run a task parallel_executable = srun -# cores per node on the machine -cores_per_node = 256 +# cores per node on the machine (with hyperthreading) +cores_per_node = 128 + +# gpus per node on the machine +gpus_per_node = 4 # account for running diagnostics jobs account = e3sm From b3bce82586502abdbc50e402f37eee824a5cd77d Mon Sep 17 00:00:00 2001 From: Xylar Asay-Davis Date: Fri, 12 Jul 2024 02:34:13 -0700 Subject: [PATCH 2/2] Add gpus_per_node to frontier --- mache/machines/frontier.cfg | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/mache/machines/frontier.cfg b/mache/machines/frontier.cfg index f4bea107..97dc6e93 100644 --- a/mache/machines/frontier.cfg +++ b/mache/machines/frontier.cfg @@ -40,9 +40,12 @@ system = slurm # whether to use mpirun or srun to run a task parallel_executable = srun -# cores per node on the machine +# allocatable cores per node on the machine cores_per_node = 56 +# gpus per node on the machine +gpus_per_node = 4 + # account for running diagnostics jobs account = cli115