Excerpt | |||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Code Block | ||||||
---|---|---|---|---|---|---|
| ||||||
#!/bin/bash
#SBATCH -J orca_example
#SBATCH -e orca_example.err
#SBATCH -o orca_example.out
#SBATCH -p std
#SBATCH --ntasks=4
#SBATCH -t 02-00:00
module load apps/orca/4.1.0
##
# Modify the input and output files!
INPUT_FILE=orca_example.inp
OUTPUT_FILE=orca_example.out
##
# You don't need to modify nothing more
cp -r ${SLURM_SUBMIT_DIR}/${INPUT_FILE} ${SCRATCH}
cd ${SCRATCH}
srun $(which orca) ${INPUT_FILE} > ${OUTPUT_FILE}
cp ./${OUTPUT_FILE} ${SLURM_SUBMIT_DIR} |
Sbatch options:
The options shown in the example are detailed below. For more information and a more comprehensive list of available options, see the sbatch command page.
- -J: Name for the job's allocation.
- -e: Name of the stderr redirection filename.
- -o: Name of the stdout redirection filename.
- -p: Name of the partition (queue) where the job will be submitted.
For parallel calculations, the number of processes must be explicitly indicated in orca_example.inp file using:
Code Block %pal nprocs 4 end
-n: Number of tasks.
- -c: Number of cores per task.
- -t: Set the job's time limit. If the job don't finish before the time runs out, it will be killed.
Software execution information:
ORCA uses 3000MB per core by default. This value can be changed in orca_example.inp using the directive:
Code Block |
---|
%maxcore 3000 end |
It is recommended to use ~75% of the requested memory to SLURM, which is 3990 MB per core.