ROAR User Guide   »   Using GPUs on Roar Collab
Feedback [ + ]

Using GPUs on Roar Collab

GPUs are available on RC to users that are added to paid GPU compute accounts. To use GPUs, add the --gpus resource directive:

 

#!/bin/bash

#SBATCH --job-name=apythonjob   # give the job a name
#SBATCH --account=<gpu_acct>    # specify the account
#SBATCH --partition=sla-prio    # specify the partition
#SBATCH --nodes=1               # request a node
#SBATCH --ntasks=1              # request a task / cpu
#SBATCH --mem=1G                # request the memory required per node
#SBATCH --gpus=1                # request a gpu
#SBATCH --time=00:01:00         # set a limit on the total run time

python pyscript.py

 

Only software that has been explicitly written to run on GPUs can take advantage of GPUs. Adding the --gpus option to a Slurm script for a CPU-only program will not speed up the execution time and will just waste resources and increase the queue time. Furthermore, some codes are only written to use a single GPU, so avoid requesting multiple GPUs unless the program can use them.