Frequently Asked Questions (FAQ)¶

Q: My job is stuck in the queue.
A: Likely the cluster is busy or your resource request is large. Check squeue -u <username> or scontrol show job <jobID> for reasons. During conference deadlines, expect longer wait times. If unsure, ask in Slack #help-desk.

Q: How do I request multiple GPUs?
A: Request multiple GPUs in Slurm (--gpus-per-node=N).

Q: Can I run multiple jobs at once?
A: Yes, the scheduler allows multiple concurrent jobs. Fair-share might lower your priority if you exceed usage relative to others.

Q: I need software that requires sudo.
A: You do not have sudo access. If you are unable to install the software you need yourself through pip or conda please reach out to an admin.

For additional questions, ask a question on Slack in the help desk channel.