[BACKEND] Support convert_layout with num_ctas > 1 Using Linear Layout
#4782
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
|
pin @lezcano because he expressed interests in reviewing LL related PRs. |
ThomasRaoux
approved these changes
Oct 2, 2024
| @@ -926,6 +937,7 @@ std::optional<LinearLayout> chooseStMatrixLayoutNoLeadingOffset( | |||
| StringAttr kWarp = S("warp"); | |||
| StringAttr kCol = S("dim1"); | |||
| StringAttr kRow = S("dim0"); | |||
| StringAttr kBlock = S("block"); | |||
There was a problem hiding this comment.
bikeshedding: I hope the name is explicit enough. Block term is way overloaded unfortunately. We could name it CTA but that's very nvidia specific.
There was a problem hiding this comment.
Sure, I had a try to change it to CTA but found it a bit annoying to change a lot of places. Let me sort it out in the future.
Luosuu
pushed a commit
to Luosuu/triton
that referenced
this pull request
Nov 13, 2024
…ayout (triton-lang#4782) Particularly, this PR implements layout conversion when a CGA contains more than one CTA. In such cases, a Triton tensor is split into multiple blocks, with each block being handled by a CTA. ``` block0 | block1 ---------------- block2 | block3 ``` If data transfer is required from block0 to block3, this PR cannot handle it, and we use `isCrossCTAConversion` to check this condition.
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Outdated suggestions cannot be applied.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.
Particularly, this PR implements layout conversion when a CGA contains more than one CTA. In such cases, a Triton tensor is split into multiple blocks, with each block being handled by a CTA.
If data transfer is required from block0 to block3, this PR cannot handle it, and we use
isCrossCTAConversionto check this condition.