chundoong-lab-ta/SamsungDS23/cuda-mpi-example/run_performance.sh

#!/bin/bash

salloc -N 2 --exclusive                              \
  mpirun --bind-to none -mca btl ^openib -npernode 1 \
  -x OMP_PROC_BIND=true                              \
  numactl --physcpubind 0-31                         \
  ./main -n 10 8192 8192 8192