Skip to content

Add warning for unused GPU when n_clusters < n_ranks in batch all_neighbors#1072

Merged
rapids-bot[bot] merged 11 commits intorapidsai:branch-25.08from
jinsolp:fix-an-mg-rank
Jul 23, 2025
Merged

Add warning for unused GPU when n_clusters < n_ranks in batch all_neighbors#1072
rapids-bot[bot] merged 11 commits intorapidsai:branch-25.08from
jinsolp:fix-an-mg-rank

Conversation

@jinsolp
Copy link
Copy Markdown
Contributor

@jinsolp jinsolp commented Jun 30, 2025

In MG all_neighbors, some GPUs stay idle when n_clusters < n_ranks (this is the expected behavior). Adding warning to let users know that they are not fully using all GPUs and providing directions for better utilization.

@jinsolp jinsolp self-assigned this Jun 30, 2025
@jinsolp jinsolp requested a review from a team as a code owner June 30, 2025 18:28
@jinsolp jinsolp added improvement Improves an existing functionality non-breaking Introduces a non-breaking change labels Jun 30, 2025
@github-actions github-actions Bot added the cpp label Jun 30, 2025
Comment on lines +343 to +345
"Rank %d is not used for computation. This happens because the total number of ranks (%d) "
"> n_clusters (%lu). Consider increasing n_clusters or reduce the number of GPUs for "
"better utilization.",
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

@jinsolp
Copy link
Copy Markdown
Contributor Author

jinsolp commented Jul 23, 2025

/merge

@rapids-bot rapids-bot Bot merged commit b230ebe into rapidsai:branch-25.08 Jul 23, 2025
53 checks passed
@jinsolp jinsolp deleted the fix-an-mg-rank branch July 24, 2025 18:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

cpp improvement Improves an existing functionality non-breaking Introduces a non-breaking change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants