Skip to content

Conversation

sjpb
Copy link
Collaborator

@sjpb sjpb commented Jul 31, 2025

If not using MIG or NVLINK GPUs, it is possible to use the nvidia GRES autodetect mechanism rather than the nvml one. The former does not require rebuilding slurm against the NVIDIA management libraries, so this can be skipped when building cuda images.

This does not change the current default.

@sjpb sjpb marked this pull request as ready for review July 31, 2025 15:28
@sjpb sjpb requested a review from a team as a code owner July 31, 2025 15:28
@sjpb sjpb merged commit 64ddd42 into main Aug 1, 2025
2 checks passed
@sjpb sjpb deleted the feat/nvidia-autodetect branch August 1, 2025 09:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants