MOLECULAR DYNAMICS PERFORMANCE GUIDE - Digital Research Alliance of CANADA

BENCHMARK DETAILS

ID=181

  • Dataset: 6n4o
  • Software: PMEMD.cuda.MPI (amber/20.9-20.15-gomklc-2020a-avx512)
  • Resource: 2 tasks, 1 cores, nodes, 2 GPUs, no NVLink
  • CPU: Xeon Gold 5120 (Skylake), 2.2 GHz
  • GPU: Tesla-V100-PCIE-16GB, 4 cores/GPU
  • Simulation speed: 40.948 ns/day
  • Efficiency: 50.7 %
  • Site: Graham
  • Date: Oct. 17, 2022, 8:07 p.m.
  • Submission script:

    #!/bin/bash
    #SBATCH -c1 --gpus=v100l:1
    #SBATCH --mem-per-cpu=2000 --time=1:0:0
    INPFILE=pmemd_prod.in
    LOGFILE=md_${SLURM_JOBID}.log
    module --force purge
    ml StdEnv/2020 gcc/9.3.0 cuda/11.0 openmpi/4.0.3 amber/20.9-20.15
    pmemd.cuda -O -i $INPFILE -o $LOGFILE -p prmtop.parm7 -c restart.rst7

  • Notes:

    Benchmark submission:

    #!/bin/bash
    #SBATCH -c1 --gpus=v100l:1
    #SBATCH --mem-per-cpu=2000 --time=1:0:0
    # Usage: sbatch submit.cuda.sh
    INPFILE=pmemd_prod.in
    STEPS=10000
    # End of user input
    TMPFILE=tf_${SLURM_JOBID}
    LOGFILE=md_${SLURM_JOBID}.log
    module --force purge
    ml StdEnv/2020 gcc/9.3.0 cuda/11.0 openmpi/4.0.3 amber/20.9-20.15
    # Print resource info
    echo ${SLURM_NODELIST} running on ${SLURM_CPUS_PER_TASK} cores
    cat /proc/cpuinfo | grep "model name" | uniq
    nvidia-smi --query-gpu=gpu_name --format=csv
    pmemd.cuda -O -i $INPFILE -o $LOGFILE -p prmtop.parm7 -c restart.rst7
    grep "NonSetup CPU time" $LOGFILE > $TMPFILE
    # Print average of three runs.
    echo -n "ns/day:"
    awk -v steps=$STEPS '{total += $5; count++ } END { print count*3.6*2.4*steps*0.01/total}' $TMPFILE
    rm $TMPFILE $LOGFILE mdinfo mdcrd restrt

  • Simulation input file:

    Benchmark
    &cntrl
    imin=0,irest=1,ntx=5,
    nstlim=20000,dt=0.001,
    ntc=2,ntf=2,
    cut=8.0,
    ntpr=100, ntwx=1000,
    ntb=2, ntp=1, taup=2.0,
    ntt=3, gamma_ln=2.0, temp0=300.0,
    /
    &ewald
    nfft1=128,
    nfft2=128,
    nfft3=128,
    /