From a9766e0ef2b922be1f45aaa19b1f6b8a70523118 Mon Sep 17 00:00:00 2001 From: Christoph Knote Date: Fri, 21 Oct 2022 15:40:55 +0200 Subject: [PATCH] More fixes for openMPI --- machine_profiles/alcc_spack/batch_preambles/main | 4 +--- machine_profiles/alcc_spack/batch_preambles/post | 3 +-- machine_profiles/alcc_spack/batch_preambles/pre | 3 +-- machine_profiles/alcc_spack/batch_preambles/spinup | 3 +-- machine_profiles/alcc_spack/batch_preambles/staging | 3 +-- machine_profiles/alcc_spack/profile.bash | 9 ++++----- 6 files changed, 9 insertions(+), 16 deletions(-) diff --git a/machine_profiles/alcc_spack/batch_preambles/main b/machine_profiles/alcc_spack/batch_preambles/main index bcfbcf3..51d7947 100644 --- a/machine_profiles/alcc_spack/batch_preambles/main +++ b/machine_profiles/alcc_spack/batch_preambles/main @@ -1,13 +1,11 @@ #!/bin/bash -l #SBATCH --partition=alcc1,epyc -#SBATCH --account=mbees -#SBATCH --exclude=alcc132 #SBATCH -o __runDir__/__mainJobName__.%j.%N.out #SBATCH -D __runDir__ #SBATCH -J __mainJobName__ #SBATCH --nodes=__nNodesMain__ #SBATCH --ntasks-per-node=__nTasksPerNodeMain__ -#SBATCH --mem-per-cpu=2000 +#SBATCH --mem-per-cpu=2200 #SBATCH --mail-type=FAIL #SBATCH --mail-user=christoph.knote@med.uni-augsburg.de #SBATCH --time=18:00:00 diff --git a/machine_profiles/alcc_spack/batch_preambles/post b/machine_profiles/alcc_spack/batch_preambles/post index 318f931..a6001d8 100644 --- a/machine_profiles/alcc_spack/batch_preambles/post +++ b/machine_profiles/alcc_spack/batch_preambles/post @@ -1,6 +1,5 @@ #!/bin/bash -l -#SBATCH --partition=alcc1 -#SBATCH --account=mbees +#SBATCH --partition=alcc1,epyc #SBATCH -o __runDir__/__postJobName__.%j.%N.out #SBATCH -D __runDir__ #SBATCH -J __postJobName__ diff --git a/machine_profiles/alcc_spack/batch_preambles/pre b/machine_profiles/alcc_spack/batch_preambles/pre index 7f93b14..05d934d 100644 --- a/machine_profiles/alcc_spack/batch_preambles/pre +++ b/machine_profiles/alcc_spack/batch_preambles/pre @@ -1,6 +1,5 @@ #!/bin/bash -l -#SBATCH --partition=alcc1 -#SBATCH --account=mbees +#SBATCH --partition=alcc1,epyc #SBATCH -o __runDir__/__preJobName__.%j.%N.out #SBATCH -D __runDir__ #SBATCH -J __preJobName__ diff --git a/machine_profiles/alcc_spack/batch_preambles/spinup b/machine_profiles/alcc_spack/batch_preambles/spinup index 3c3a716..5693ae6 100644 --- a/machine_profiles/alcc_spack/batch_preambles/spinup +++ b/machine_profiles/alcc_spack/batch_preambles/spinup @@ -1,12 +1,11 @@ #!/bin/bash -l #SBATCH --partition=alcc1,epyc -#SBATCH --account=mbees #SBATCH -o __runDir__/__spinupJobName__.%j.%N.out #SBATCH -D __runDir__ #SBATCH -J __spinupJobName__ #SBATCH --nodes=__nNodesSpinup__ #SBATCH --ntasks-per-node=__nTasksPerNodeSpinup__ -#SBATCH --mem-per-cpu=2000 +#SBATCH --mem-per-cpu=2200 #SBATCH --mail-type=FAIL #SBATCH --mail-user=christoph.knote@med.uni-augsburg.de #SBATCH --time=01:00:00 diff --git a/machine_profiles/alcc_spack/batch_preambles/staging b/machine_profiles/alcc_spack/batch_preambles/staging index d1507a9..e288f77 100644 --- a/machine_profiles/alcc_spack/batch_preambles/staging +++ b/machine_profiles/alcc_spack/batch_preambles/staging @@ -1,6 +1,5 @@ #!/bin/bash -l -#SBATCH --partition=alcc1 -#SBATCH --account=mbees +#SBATCH --partition=alcc1,epyc #SBATCH -o __runDir__/__stagingJobName__.%j.%N.out #SBATCH -D __runDir__ #SBATCH -J __stagingJobName__ diff --git a/machine_profiles/alcc_spack/profile.bash b/machine_profiles/alcc_spack/profile.bash index 37a43f7..5621441 100644 --- a/machine_profiles/alcc_spack/profile.bash +++ b/machine_profiles/alcc_spack/profile.bash @@ -11,15 +11,14 @@ export UCX_NET_DEVICES=eth5 # --- MPI settings --- mpiCommandPre="mpirun" -#mpiCommandMain="mpirun -mca pml ucx -mca coll ^hcoll -x UCX_TLS=tcp,shm -x UCX_NET_DEVICES=eth5 /usr/bin/time -v" -mpiCommandMain="mpirun" +mpiCommandMain="mpirun -mca pml ucx -mca coll ^hcoll -x UCX_TLS=tcp,shm -x UCX_NET_DEVICES=eth5" mpiCommandReal=${mpiCommandPre} -nNodesMain=3 -nTasksPerNodeMain=28 +nNodesMain=1 +nTasksPerNodeMain=64 nNodesSpinup=1 -nTasksPerNodeSpinup=28 +nTasksPerNodeSpinup=64 # --- Batch system --- -- GitLab