Batch UpdatePosition using cudaMemcpy #7964
Merged
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
trivialfis
reviewed
Jun 24, 2022
|
Overall looks good to me. Some small nitpicks. The optimization looks exciting! |
|
Sorry for the noise. I finally restored Jenkins back to health, and now builds jobs will launch automatically. |
RAMitchell
force-pushed
the
batch-position-memcpy
branch
from
0e9ebd7
to
776ef9f
Compare
trivialfis
reviewed
Jun 28, 2022
trivialfis
approved these changes
Jun 29, 2022
RAMitchell
force-pushed
the
batch-position-memcpy
branch
from
aad0d8e
to
9eddfce
Compare
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Outdated suggestions cannot be applied.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.
Alternative version of #7925 using cudaMemcpy and an unstable row partitioning scheme using warp aggregated atomics.
Considerably faster than master branch in row partitioning, but is unstable i.e. produces random orderings of rows in the partitioner. I think this is not a problem as the only consumer of the row partitions is the histogram kernel, which doesn't care about the ordering.
Will follow-up with some benchmarking and more extensive tests, in particular checking that we still have determinism.