Re: Slurm Script for Multi Node Run

From: Bennion, Brian (bennion1_at_llnl.gov)
Date: Mon Apr 17 2017 - 11:52:58 CDT

if you are really using slurm and you are using an mpi version of namd2, then you should prabably use srun to start the multinode job

srun -N #nodes -n #mpitasks per node namd2 input.namd > output.namd

make sure the namd2 executable is mpi aware

________________________________
From: owner-namd-l_at_ks.uiuc.edu <owner-namd-l_at_ks.uiuc.edu> on behalf of Zeki Zeybek <zeki.zeybek_at_bilgiedu.net>
Sent: Sunday, April 16, 2017 9:42:41 AM
To: namd-l_at_ks.uiuc.edu
Subject: namd-l: Slurm Script for Multi Node Run

Hi !

In the server I am using each node have 15 cores in it and each user is allowed to use at most 96 cores. My slurm script for single node works without any error but I just cannot construct the proper job script to use 80 cores for instance.

My current script is;

#!/bin/sh
#SBATCH --partition=NAME
#SBATCH --account=NAME
#SBATCH --job-name=parallel_job
#SBATCH --mail-type=ALL
#SBATCH --mail-user=some_at_someedu.net
#SBATCH --nodes=10
#SBATCH --ntasks=10
#SBATCH --cpus-per-task=8
#SBATCH --time=sometime
#SBATCH --output=parallel_%j.out

Required Libraries and Modules

echo "SLURM_NODELIST $SLURM_NODELIST"
echo "NUMBER OF CORES $SLURM_NTASKS"
echo "Running program on $SLURM_CPUS_ON_NODE CPU cores"

mpirun $NAMD_DIR/namd2 namd_input.conf > namd_multinode_output.log

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:13 CST