Skip to content

Conversation

@malago86
Copy link
Member

I added a few lines to retrieve the GPUs that are Idle (by using the nvidia-smi output).

  • Idle GPUs will be utilized first when running MCGPU
  • If there is not enough idle GPUs, the process will be distributed on the available idle GPUs
  • If there is no idle GPU available, it will wait until at least one is available

Take a look at the changes and let me know what you think! There is room for improvement, but the idea is there and seems to be working so far!

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.

1 participant