SCIAMA
High Performance Compute Cluster
Using SLURM on SCIAMA (for PBS users)
For a full list of SLURM commands, please consult the SLURM documentation.
Submitting jobs
For details on how to submit (batch or interactive)jobs on SCIAMA using slurm, please read this article on how to do this.
Migration guide for Torque/PBS users
For people who are used to the "old" Torque/PBS system, here a quick translation table for the most important commands:
| Function | Torque | SLURM |
|---|---|---|
| Interactive shell on compute node | qsub -I | sinteractive |
| Batch job submission | qsub | sbatch |
| Queue status | qstat | squeue |
| Delete job | qdel | scancel |
| Hold job | qhold | scontrol hold |
| Release job | qrls | scontrol release |
Below is a simple job submission script. Basically #PBS is replaced by #SBATCH.
The Torque line of the form "#PBS -l nodes=2:ppn=2" is replaced by the lines "#SBATCH –nodes=2" and "#SBATCH –ntasks-per-node=2"
A "queue (-q)" in Torque is replaced by a "partition(-p)" in SLURM.
#!/bin/bash
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=2
#SBATCH --time=0-2:00
#SBATCH --job-name=starccm
#SBATCH -p sciama4.q
#SBATCH -D /users/burtong/test-starccm
#SBATCh --error=starccm.err.%j
#SBATCH --output=starccm.out.%j
##SLURM: ====> Job Node List (DO NOT MODIFY)
echo "Slurm nodes assigned :$SLURM_JOB_NODELIST"
echo "SLURM_JOBID="$SLURM_JOBID
echo "SLURM_JOB_NODELIST"=$SLURM_JOB_NODELIST
echo "SLURM_NNODES"=$SLURM_NNODES
echo "SLURMTMPDIR="$SLURMTMPDIR
echo "working directory = "$SLURM_SUBMIT_DIR
echo "SLURM_NTASKS="$SLURM_NTASKS
echo ------------------------------------------------------
echo 'This job is allocated on '${SLURM_NTASKS}' cpu(s)'
echo 'Job is running on node(s): '
echo $SLURM_JOB_NODELIST
echo ------------------------------------------------------
module purge
module add starccm/12.06.011
starccm+ -rsh ssh -batch test1.sim -fabric tcp -power -podkey -np ${SLURM_NTASKS}
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=2
#SBATCH --time=0-2:00
#SBATCH --job-name=starccm
#SBATCH -p sciama4.q
#SBATCH -D /users/burtong/test-starccm
#SBATCh --error=starccm.err.%j
#SBATCH --output=starccm.out.%j
##SLURM: ====> Job Node List (DO NOT MODIFY)
echo "Slurm nodes assigned :$SLURM_JOB_NODELIST"
echo "SLURM_JOBID="$SLURM_JOBID
echo "SLURM_JOB_NODELIST"=$SLURM_JOB_NODELIST
echo "SLURM_NNODES"=$SLURM_NNODES
echo "SLURMTMPDIR="$SLURMTMPDIR
echo "working directory = "$SLURM_SUBMIT_DIR
echo "SLURM_NTASKS="$SLURM_NTASKS
echo ------------------------------------------------------
echo 'This job is allocated on '${SLURM_NTASKS}' cpu(s)'
echo 'Job is running on node(s): '
echo $SLURM_JOB_NODELIST
echo ------------------------------------------------------
module purge
module add starccm/12.06.011
starccm+ -rsh ssh -batch test1.sim -fabric tcp -power -podkey -np ${SLURM_NTASKS}