isaac-sim / IsaacLab

Unified framework for robot learning built on NVIDIA Isaac Sim
https://isaac-sim.github.io/IsaacLab
Other
1.9k stars 728 forks source link

[Proposal] Submitting training jobs for HPCs using a PBS job scheduler #599

Closed shafeef901 closed 2 months ago

shafeef901 commented 2 months ago

Proposal

Add support for training using IsaacLab on clusters that use PBS job scheduler.

Motivation

The current example for training on a cluster assumes a SLURM (Simple Linux Utility for Resource Management) job scheduler/workload manager. However, PBS (Portable Batch System) is another widely adopted job scheduler. I have managed to submit jobs on a PBS job scheduler-based HPC by adapting the existing job submission scripts.

If this feature seems valuable for the community, I would be happy to contribute by giving the option to choose between the two job schedulers and running the appropriate scripts.

Checklist

Mayankm96 commented 2 months ago

Definitely. Would love to extend the support to as many cluster systems as possible :)