archer2.job

#!/bin/bash

# Slurm job options (name, compute nodes, job time)
#SBATCH --job-name=benchio
#SBATCH --time=00:05:00
#SBATCH --output=%x-%j.out
#SBATCH --nodes=1
#SBATCH --tasks-per-node=1
#SBATCH --cpus-per-task=1
#SBATCH --qos=short
#SBATCH --partition=standard

module use /work/z19/shared/sfarr/modulefiles
module load adios/2.8.3
module load cray-hdf5-parallel
module load cray-netcdf-hdf5parallel

# Set the number of threads to 1
#   This prevents any threaded system libraries from automatically
#   using threading.

export OMP_NUM_THREADS=1
export SRUN_CPUS_PER_TASK=${SLURM_CPUS_PER_TASK}

ulimit -s unlimited

# Launch the parallel job

# Magic setting to improve IO performance with default MPI
export FI_OFI_RXM_SAR_LIMIT=64K

srun --unbuffered --distribution=block:block \
     ./benchio 512 512 512 local serial unstriped