perf: Fix python API overhead when CUDAGraph is not enabled #969
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
|
After the changes, here is the norm python API performance (cpu and python overhead included, https://gist.github.com/yzh119/d9bf2abbb667abcbb806979f4bbea633): A100 PerformanceThis PRv0.1.6H100 PerformanceThis PR (no PDL)This PR (w/ PDL)v0.1.6 (no PDL) |
This was referenced Mar 24, 2025
Closed
5 tasks
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Outdated suggestions cannot be applied.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.
This PR fixes issue #960 , we identifies several performance bottlenecks for our python APIs when kernels are not captured by CUDAGraph:
with input.device as device:)These issues were introduced in JIT refactor after v0.1.6 (mainly for accelerating JIT compilation speed). In this PR, we changed back to get stream and device guard in C++).
@MichoChan @xiaoqi35