fix(geometry): retry point-in-polygon with finer batching on GPU OOM#62
Open
EliHei2 wants to merge 1 commit into
Open
fix(geometry): retry point-in-polygon with finer batching on GPU OOM#62EliHei2 wants to merge 1 commit into
EliHei2 wants to merge 1 commit into
Conversation
The batched quadtree point-in-polygon join could OOM on very large inputs (notably MERSCOPE, with millions of transcripts) and crash the run. Wrap the batch loop so a CUDA out-of-memory error retries with progressively finer batching (doubling the batch count up to 256) before giving up, and return an empty match frame when there are no results instead of failing in cudf.concat. What to review: the OOM-detection (errors only; non-OOM re-raised immediately) and the bounded retry in _points_in_polygons_contains. Keeps main's quadtree API; no change to the happy path.
Collaborator
|
Good catch, down to merge if it makes things more robust. Two things first:
Could you share a log showing it crash in points-in-polygon on your end?
Happy to merge meanwhile, but would like to confirm it's really points-in-polygon before. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The batched quadtree point-in-polygon join could OOM on very large inputs (notably MERSCOPE, with millions of transcripts) and crash the run. Wrap the batch loop so a CUDA out-of-memory error retries with progressively finer batching (doubling the batch count up to 256) before giving up, and return an empty match frame when there are no results instead of failing in cudf.concat.
What to review: the OOM-detection (errors only; non-OOM re-raised immediately) and the bounded retry in _points_in_polygons_contains. Keeps main's quadtree API; no change to the happy path.