ID=41, OPENMM.cuda

• Module/Version: openmm/8.1.1
• Toolchain/Version: gofbc/2023a
• CPU instruction set: avx2
• Example job submission script:

  #!/bin/bash
  #SBATCH -c1 --gres=gpu:h100:1  
  #SBATCH --men-per-cpu=4000 --time=1:0:0 
  
  module purge
  module load arch/avx2 StdEnv/2023 gcc/12.3 openmpi/4.1.5 cuda/12.2
  module load ambertools/23.5 openmm/8.1.1 
  
  virtualenv \${SLURM_TMPDIR}/env
  source \${SLURM_TMPDIR}/env/bin/activate
  pip install --no-index netCDF4
  
  python openmm_input.py

• Benchmark submission script:

    
  # Usage: bash $0 
  sbatch << EOF
  #!/bin/bash
  #SBATCH -c${1} --gres=gpu:h100:${1}  
  #SBATCH --mem-per-cpu=4000 --time=1:0:0 --partition=gpu_general
  
  module purge
  module load arch/avx2 StdEnv/2023 gcc/12.3 openmpi/4.1.5 cuda/12.2
  module load ambertools/23.5 openmm/8.1.1 
  
  virtualenv \${SLURM_TMPDIR}/env
  source \${SLURM_TMPDIR}/env/bin/activate
  pip install --no-index netCDF4
  
  cat /proc/cpuinfo | grep "model name" | uniq
  python openmm_input.py
  echo -n "Using ${1} GPUs, "
  hostname -s
  EOF