smyalygames/FiniteVolumeGPU_HIP
1
0
Fork 0
mirror of https://github.com/smyalygames/FiniteVolumeGPU_HIP.git synced 2025-05-18 14:34:12 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

fix: use MPI in slurm job

Browse Source
This commit is contained in:
Anthony Berg 2025-03-30 20:44:40 +02:00
parent 28a96382ff
commit cf102131df
1 changed files with 10 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
Jobs/job_lumi.slrum
View File

@ -1,39 +1,27 @@
#!/bin/bash -e #!/bin/bash -l
#SBATCH --job-name=lumi #SBATCH --job-name=lumi
#SBATCH --account=project_4650000xx #SBATCH --account=project_4650000xx
#SBATCH --time=00:10:00 #SBATCH --time=00:10:00
#SBATCH --partition=dev-g #SBATCH --partition=dev-g
#SBATCH --nodes=1 #SBATCH --nodes=1
#SBATCH --ntasks-per-node=8 #SBATCH --ntasks-per-node=8
#SBATCH --gpus=8
#SBATCH --gpus-per-node=8 #SBATCH --gpus-per-node=8
#SBATCH -o %x-%j.out #SBATCH --output=%x-%j.out
#SBATCH --exclusive #SBATCH --exclusive
#
N=$SLURM_JOB_NUM_NODES N=$SLURM_JOB_NUM_NODES
echo "--nbr of nodes:", $N echo "--nbr of nodes:", $N
echo "--total nbr of gpus:", $SLURM_NTASKS echo "--total nbr of gpus:", $SLURM_NTASKS
Mydir=/project/project_4650000xx Mydir=/project/${project}
Myapplication=${Mydir}/FiniteVolumeGPU_hip/mpiTesting.py Myapplication=${Mydir}/FiniteVolumeGPU_HIP/mpiTesting.py
CondaEnv=${Mydir}/FiniteVolumeGPU_HIP/MyCondaEnv/bin
#modules export PATH="${CondaEnv}:$PATH"
ml LUMI/24.03 partition/G
ml lumi-container-wrapper
ml cray-python/3.11.7
ml rocm/6.2.2
ml craype-accel-amd-gfx90a CPU_BIND="map_cpu:49,57,17,25,1,9,33,41"
ml cray-mpich/8.1.29
export PATH="/project/project_4650000xx/FiniteVolumeGPU_hip/MyCondaEnv/bin:$PATH" export MPICH_GPU_SUPPORT_ENABLED=1
#missing library srun --cpu-bind=${CPU_BIND} --mpi=pmi2 \
export LD_LIBRARY_PATH=/opt/cray/pe/mpich/8.1.29/ofi/cray/17.0/lib-abi-mpich:$LD_LIBRARY_PATH
#Binding mask
bind_mask="0x${fe}000000000000,0x${fe}00000000000000,0x${fe}0000,0x${fe}000000,0x${fe},0x${fe}00,0x${fe}00000000,0x${fe}0000000000"
srun --cpu-bind=mask_cpu:$bind_mask \
python ${Myapplication} -nx 1024 -ny 1024 --profile python ${Myapplication} -nx 1024 -ny 1024 --profile