Fix handling of 3D DotOp with M=1 #8561
Merged
+27
−14
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
ThomasRaoux
reviewed
Oct 27, 2025
neildhar
force-pushed
the
fix-dotop-lowering
branch
3 times, most recently
from
c0d7dd3 to
ecb7e9b
Compare
neildhar
changed the title
neildhar
force-pushed
the
fix-dotop-lowering
branch
from
ecb7e9b to
972975e
Compare
We currently always look at index 1 in the shape to determine if it is an outer product. However, for 3D operands, we actually have to look at index 2. This causes us to incorrectly crash when supplied a 3D operand where the first dimension is 1. Eliminate the isOuter check entirely, since my understanding is that: 1. it is purely defensive, since we will just crash when isOuter is true and mmaLayout is non-null. 2. it does not disqualify all the invalid K values, so it might be confusing.
neildhar
force-pushed
the
fix-dotop-lowering
branch
from
972975e to
2501d30
Compare
|
@ThomasRaoux is this good to merge? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
We currently always look at index 1 in the shape to determine if it is
an outer product. However, for 3D operands, we actually have to look at
index 2. This causes us to incorrectly crash when supplied a 3D operand
where the first dimension is 1.
Eliminate the isOuter check entirely, since my understanding is that:
true and mmaLayout is non-null.
confusing.