Seitenvergleich

...

Codeblock

language	bash
title	Lise (using mpirun): on Nvidia A100 GPU nodes

#!/bin/bash 
#SBATCH --partition=gpu-a100  
#SBATCH --time 12:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=4
#SBATCH --cpus-per-task=18
#SBATCH --time=06:00:00
#SBATCH --job-name=cp2k

export SLURM_CPU_BIND=none
export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}    
export OMP_PLACES=cores
export OMP_PROC_BIND=close

export OMPI_MCA_coll="^hcoll"
export OMPI_MCA_btl="smcuda,tcp,vader,self"  

module load gcc/11.3.0 openmpi/gcc.11/4.1.4 cp2k/2023.1

# gpu_bind.sh (see the following script) should be placed inside the same directory where cp2k will be executed
mpirun --bind-to core --map-by numa:PE=${SLURM_CPUS_PER_TASK} ./gpu_bind.sh cp2k.psmp input > output

Codeblock

language	bash
title	gpu_bind.sh

#!/bin/bash
export CUDA_VISIBLE_DEVICES=$OMPI_COMM_WORLD_LOCAL_RANK
$@

Codeblock

language	bash
title	Emmy (using srun)

#!/bin/bash
#SBATCH --time 12:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=24
#SBATCH --cpus-per-task=4
#SBATCH --time=06:00:00
#SBATCH --job-name=cp2k

export OMP_NUM_THREADS=${SLURM_CPUS_PER_TASK}

module load intel/2022.2 impi/2021.6 cp2k/2023.1
srun cp2k.psmp input > output

...

Versionen im Vergleich

Alte Version 9

Neue Version 10

Schlüssel