Workaround Windows correctness issue in reduce_by_segment.pass #1771
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
Signed-off-by: Matthew Michel <[email protected]>
mmichel11
requested review from
danhoeflinger,
SergeyKopienko,
julianmi and
adamfidel
julianmi
reviewed
Aug 7, 2024
Signed-off-by: Matthew Michel <[email protected]>
|
I've added this to the milestone even though it's a test change, since it will be important for testing of other features in the milestone. |
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Outdated suggestions cannot be applied.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.
This PR is put up in anticipation of the reduce-then-scan implementation being committed. It can currently be produced on
dev/shared/reduce_than_scan_impl.When using the new reduce-then-scan implementation, a correctness issue has been identified on Windows with
reduce_by_segment.pass.cpp. The issue can only be produced with the current ordering of test cases (and I assume ordering in which kernels are compiled). After running the two flag predicate tests, the thirdMatrixPointtest return incorrect results. We run into this issue even if the flag predicate tests are not run and their kernels only compiled. Reordering the tests to run theMatrixPointtest prior to the flag predicate tests, or completely removing the flag predicate tests results in a pass.We believe this issue related to the order of kernel compilation is a compiler bug and are following up with the compiler team. For now, I have added a broken macro to reorder the tests on Windows.