feat(cuda): Optimize MoE scatter_output with parallel CUDA schedule #3398
Conversation
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
1 participant
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Outdated suggestions cannot be applied.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.
Optimizes the scatter_output operator used in Mixture-of-Experts (MoE) inference by replacing the default serial execution with a custom TVM schedule.
The new schedule (_schedule_scatter_output) fuses the token and hidden dimension loops and explicitly binds them to CUDA blocks and threads (1024 threads/block). This ensures the scatter operation runs in parallel on the GPU , significantly improving memory bandwidth utilization compared to the previous T.serial implementation.