- --gpus-per-task is not working

Updated: This is fixed. 

Original Post:

After the recent Slurm upgrade, the option --gpus-per-task is currently not functioning as expected. Users attempting to submit jobs with this option may encounter the following error message:

salloc: error: Job submit/allocate failed: Requested node configuration is not available

As a workaround, users should use the --gpus-per-node option instead of --gpus-per-task. We have opened a ticket with the vendor about this issue.