Jobs can be submitted to Koko through two methods, SRUN and SBATCH scripts. We have listed a few sample SBATCH scripts to assist users in building their own scripts to submit jobs. Please see our website here for partition information.
To submit your SBATCH script to SLURM once you are finished, please save the file and start it with the command: ‘$ sbatch <your script name>.sh’. This will submit your job to the SLURM partition/queue you specify in your script.
NOTE: These scripts are not directly usable to submit jobs, please edit them to meet the requirements of your job submission.
CPU single node with 20 CPU cores:
#!/bin/bash
#SBATCH --partition=<choose a partition/queue>
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=20
#SBATCH --exclusive
#SBATCH --job-name="<your job name here>"
#SBATCH --output=test.out
#SBATCH --mail-user=<your email here>
#SBATCH --mail-type=ALL
<add your modules and code here>
GPU job:
#!/bin/bash
#SBATCH --partition=<choose a partition/queue>
#SBATCH --nodes=1
#SBATCH --gres=gpu:<gpu type>:<number of gpu's>
#SBATCH --job-name="<your job name here>"
#SBATCH --output=test.out
#SBATCH --mail-user=<your email here>
#SBATCH --mail-type=ALL
<add your modules and code here>
For more info and additional SBATCH commands, please visit the SLURM SBATCH manual page here.