Asynchronously copy table data to the host during shuffle #11280
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
Signed-off-by: Jason Lowe <jlowe@nvidia.com>
revans2
approved these changes
Jul 31, 2024
|
build |
abellina
approved these changes
Jul 31, 2024
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Outdated suggestions cannot be applied.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.
Leverages rapidsai/cudf#16429 to asynchronously copy the partitioned table data to the host. This avoids unnecessary stream synchronization between each device buffer being copied back to the host and better overlaps CPU and GPU work during sliceInternalOnCpu. This has no measurable performance difference on NDS runs because the schema of shuffled data is relatively narrow (i.e.: not many separate buffers to copy back to the host and thus not many unnecessary CUDA stream synchronizations to save), but for wide shuffled schemas, this can make a significant difference. For example, sliceInternalOnCpu for a repartition of a table with 512 integer columns takes half the time with this asynchronous copy.