Still setting up. If it stalls here for more than ~15 minutes the host is probably bad — use "Try a different host".

Idle rigs shut down automatically after 20 minutes with no jobs queued or running.

Open OnDemand

Full HPC portal — file manager, job composer, in-browser shell and terminal.

Write a job

Standard sbatch script — #SBATCH directives, then your commands. GPU tools (nvidia-smi, CUDA) are available system-wide.