forked from jdherman/hpc-submission-scripts
-
Notifications
You must be signed in to change notification settings - Fork 0
/
example-slurm-script.sh
executable file
·21 lines (17 loc) · 1001 Bytes
/
example-slurm-script.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
#!/bin/bash
#SBATCH -D /full/directory/path
#SBATCH -o /full/path/to/output/file/myjob.%j.%N.out # Name of the output file (eg. myMPI.oJobID)
#SBATCH -n 1024 # Total number of processors to request (16 cores per node)
#SBATCH -p normal # Queue name "normal"
#SBATCH -t 24:00:00 # Run time (hh:mm:ss) - 24 hours
#SBATCH --mail-user=email@address # address for email notification
#SBATCH --mail-type=ALL # email at Begin and End of job
# Job script for SLURM system
# Everything in this script will be one job submission
# Note that this loops through SEEDS, but NUM is set as an environment variable in submit-slurm-loop.sh
# In other words you can divide up the work between the two scripts however you want -- decide how many total jobs there should be (in submit-slurm-loop) and how much work should be done for each job (in the loop below).
SEEDS=$(seq 1 10)
for SEED in ${SEEDS}
do
ibrun ./my_program -s ${SEED} -c ${VARIABLE}
done