[Mosaic GPU] Add a bank-conflict checker to tiled transfer + transfer planner #25149
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
![copybara-service[bot]](https://avatars.githubusercontent.com/in/44061?s=60&v=4){kind=small}
… planner Instead of only allowing a fixed set of layouts that we've hand verified as bank-conflict free, we now simulate the transactions performed within each warp and verify that no bank conflicts happen. If we detect that the simple schedule does not work out, we attempt to partition the threads in a warp into two groups and stagger the transfers in a way that lets us avoid conflicts. This allows us to match the hand-designed transfer schedule I wrote for 32-bit types, and even generalizes it to more cases automatically (e.g. swizzle=32). PiperOrigin-RevId: 700692053
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Outdated suggestions cannot be applied.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.
[Mosaic GPU] Add a bank-conflict checker to tiled transfer + transfer planner
Instead of only allowing a fixed set of layouts that we've hand verified as
bank-conflict free, we now simulate the transactions performed within each
warp and verify that no bank conflicts happen. If we detect that the simple
schedule does not work out, we attempt to partition the threads in a warp
into two groups and stagger the transfers in a way that lets us avoid conflicts.
This allows us to match the hand-designed transfer schedule I wrote for 32-bit
types, and even generalizes it to more cases automatically (e.g. swizzle=32).