Replace cuDF (de)serializer with cuDF spill-aware (de)serializer #1369
Conversation
This file contains 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
pentschev
added
2 - In Progress
pentschev
added
3 - Ready for Review
|
@ayushdg @trivialfis @dantegd @quasiben people who are using cuDF spill for whatever reason should test this, the change has potential to avoid double Dask+cuDF spilling of cuDF objects and therefore may improve performance. There's also a chance that this might introduce issues that are hard to predict, so please feel free to give it a try and report any issues and/or performance results, we plan to merge it in |
|
cc @rjzamora as well |
madsbk
requested changes
Jul 30, 2024
| if not cudf.get_option("spill"): | ||
| from cudf.comm import serialize | ||
|
|
||
|
|
There was a problem hiding this comment.
It would be good with a comment here saying why this it needed and include a link to #1363
madsbk
approved these changes
Jul 31, 2024
|
Thanks for review and approval @madsbk ! |
|
/merge |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Replace cuDF (de)serializer with cuDF spill-aware (de)serializer, using both together should be avoided as that will cause excessive spilling.
Additionally add:
LocalCUDACluster;dask cuda workerwarning to alert the user that cuDF spilling mechanism requires client/scheduler to enable it as well.Closes #1363 .