[Matrix] Add comparison operator tests - #1447
Merged
Merged
Conversation
farzonl
added
the
test-all
Member
Author
|
So I don't have to come back and remove a bunch of xfails lets wait on this PR until llvm/llvm-project#216791 merges |
Member
Author
|
the tests that have 64/16 bit requires are not as exhaustive with operator testing as the 32bit tests. |
kmpeng
self-requested a review
August 19, 2026 17:22
Member
Author
|
Failures to investigate: DXC MetalClang DXC MetalThe issue in Metal seems to be offloader specific. We don't have a uint32 Format. |
farzonl
removed
the
test-all
farzonl
force-pushed
the
task/issue-1446
branch
2 times, most recently
from
August 20, 2026 17:40
61d2f34 to
b054c3a
Compare
kmpeng
reviewed
Aug 20, 2026
kmpeng
approved these changes
Aug 24, 2026
Icohedron
reviewed
Aug 25, 2026
Icohedron
left a comment
Icohedron
left a comment
Contributor
There was a problem hiding this comment.
I don't see why the 32-bit tests have to be the only comprehensive ones.
The 32-bit tests aren't even that much bigger than the other tests in line count either.
e.g. the fp16 test is 72 lines while the fp32 test is 105 lines
| Buffers: | ||
| - Name: UintIn | ||
| Format: UInt32 | ||
| Data: [ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 ] |
Contributor
There was a problem hiding this comment.
[SUGGESTION] The int and uint tests should exercise values where the sign bit is actually set. As-is, there is no observable difference if the compiler used signed/unsigned operations where it should have been unsigned/signed.
alsepkow
reviewed
Aug 27, 2026
alsepkow
left a comment
alsepkow
left a comment
Collaborator
There was a problem hiding this comment.
LGTM assuming @Icohedron comments are addressed.
And looks like there's a legalization issue causing a crash in the 32 bit tests?
Member
Author
Yes I am waiting on reviews for this pr that will fix that: llvm/llvm-project#219017 care to take a look? It is really short. |
resolves llvm#1446 cover cmp operators across matrix sizes and base types
…are doing elementwise compares
…-element LHS/RHS matrices
farzonl
force-pushed
the
task/issue-1446
branch
from
September 16, 2026 15:41
50d3613 to
317f890
Compare
Icohedron
approved these changes
Sep 16, 2026
Icohedron
left a comment
Icohedron
left a comment
Contributor
There was a problem hiding this comment.
Would be good to have uint tests exercise extra large integers which have the most significant bit set to be sure it's not interpreted as a negative signed integer instead.
But otherwise LGTM
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
5 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.
resolves #1446
cover cmp operators across matrix sizes and base types
assisted by co-pilot MAI model