ID=43, GROMACS.cuda.mpi

• Module/Version: gromacs/2024.4
• Toolchain/Version: gofbc/2023a
• CPU instruction set: avx512
• Example job submission script:

  #!/bin/bash
  #SBATCH --mem-per-cpu=2000 -c16 --ntasks=2 --gpus-per-node=2  --nodes=1 --time=1:0:0 
  
  export OMP_NUM_THREADS="${SLURM_CPUS_PER_TASK:-1}"
  module load StdEnv/2023  gcc/12.3  openmpi/4.1.5 cuda/12.2 gromacs/2024.4
  srun gmx_mpi mdrun -s topol.tpr

• Benchmark submission script:

    
  #!/bin/bash
  #SBATCH --mem-per-cpu=2000 -c16 --ntasks=2 --gpus-per-node=2  --nodes=1 --time=1:0:0 
  
  export OMP_NUM_THREADS="${SLURM_CPUS_PER_TASK:-1}"
  module load StdEnv/2023  gcc/12.3  openmpi/4.1.5 cuda/12.2 gromacs/2024.4
  srun gmx_mpi mdrun -s topol.tpr -cpt 0
  
  grep Brand md.log
  grep "The number of OpenMP threads" md.log
  grep "MPI pro" md.log
  echo ntasks=$SLURM_TASKS_PER_NODE
  grep Performance: md.log