SLURM is a job scheduler that helps take full advantage of the ECPS cluster by allocating workloads evenly across users and nodes.

Getting started

1. Log in to ECPS via ssh (instructions for Mac, for Windows)
2. Type the following command to get an overview of the cluster's state:

   sinfo -N -l

   

Serial script

here is an example of serial script and parallel script

Serial

#!/bin/bash
#SBATCH --time=1:00:00
#SBATCH --nodes=1
#SBATCH --ntasks=8
#SBATCH --partition=Compute

srun ~/build_2/programs/simulateflow -T 10000 -s 1000 uinit.nc

Parallel

#!/bin/bash
#SBATCH --time=1:00:00
#SBATCH --nodes=2
#SBATCH --ntasks=8
#SBATCH --partition=FastCompute

mpirun ~/build_2/programs/simulateflow -T 10000 -s 1000 uinit.nc

See also

• SLURM Wiki page for administrators
• Running SLURM from Matlab
• SLURM tutorial at CÉCI-HPC
• SCITAS offers SLURM introductory courses on-site (Écublens campus)