[LLVMGPU] Support CastTypeToFitMMA on TransformDialect script. #17884
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
raikonenfnu
requested review from
MaheshRavishankar,
qedawkins,
kuhar and
Groverkss
as code owners
qedawkins
approved these changes
Jul 12, 2024
|
|
||
| if (scheduleAttr) { | ||
| func.walk([&](vector::ContractionOp contract) { | ||
| contract->setAttr("iree.amdgpu.mma", scheduleAttr.getIntrinsic()); |
There was a problem hiding this comment.
nit: I would check if the contract already has this attribute set and only set it if not present. Also maybe add a comment
// Import mma type from dispatch schedule attribute if present.
There was a problem hiding this comment.
That makes sense :)
antiagainst
approved these changes
Jul 12, 2024
LLITCHEV
pushed a commit
to LLITCHEV/iree
that referenced
this pull request
Jul 30, 2024
…org#17884) Previously CastTypeToFitMMA relies on the `mma_schedule` attribute on the function's translationInfo to obtain information about `iree.amdgpu.mma`(intrnisic selected). While this is fine for C++ pipeline, the IR generated from TransformDialect script do not have such information. Instead IR generated in TD script typically annotate the `iree.amdgpu.mma`(intrnisic selected) directly on the vector.contractOps. This is a crucial part of enabling performant the latest attention compilation pipeline (with online attn + transpose fusion) which is based on TD scripts. --------- Co-authored-by: Kunwar Grover <[email protected]> Signed-off-by: Lubo Litchev <[email protected]>
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Outdated suggestions cannot be applied.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.
Previously CastTypeToFitMMA relies on the
mma_scheduleattribute on the function's translationInfo to obtain information aboutiree.amdgpu.mma(intrnisic selected).While this is fine for C++ pipeline, the IR generated from TransformDialect script do not have such information. Instead IR generated in TD script typically annotate the
iree.amdgpu.mma(intrnisic selected) directly on the vector.contractOps.This is a crucial part of enabling performant the latest attention compilation pipeline (with online attn + transpose fusion) which is based on TD scripts.