[NVPTX] Improve lowering of v2i16 logical ops. #67073
Merged
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
Bitwise logical ops can always be done as b32, regardless of availability of other v2i16 ops, that would need a new GPU.
Artem-B
referenced
this pull request
Sep 21, 2023
TargetLoweringBase makes almost all ops legal by default, so make ones that Expand explicit and remove redundant legal settings.
d0k
approved these changes
Sep 22, 2023
|
Heads up: this has caused xla/tests:float8_test_gpu to fail. |
MaskRay
added a commit
that referenced
this pull request
Sep 23, 2023
This reverts commit 6485790. Caused xla/tests:float8_test_gpu to fail ``` LLVM ERROR: Cannot select: t118: v2i16 = or t375, t401 t375: v2i16 = BUILD_VECTOR t374, t372 t374: i16 = select t247, Constant:i16<8960>, t360 t247: i1 = setcc t199, Constant:i16<7>, seteq:ch t199: i16 = extract_vector_elt t187, Constant:i64<0> t187: v2i16 = and t183, t410 t183: v2i16 = BUILD_VECTOR t383, t384 ... ``` Acked by author to revert
Filed an internal bug :) |
MaskRay
added a commit
that referenced
this pull request
Sep 23, 2023
The test is adapted from https://reviews.llvm.org/D46008
|
Re-landed w/ the fix in #67365 |
Guzhu-AMD
pushed a commit
to GPUOpen-Drivers/llvm-project
that referenced
this pull request
Sep 28, 2023
Local branch amd-gfx 5e06c87 Merged main:a6e81df289ec into amd-gfx:18e7078bbdc4 Remote branch main e01df87 [NVPTX] Test crash introduced by llvm#67073
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Outdated suggestions cannot be applied.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.
Bitwise logical ops can always be done as b32, regardless of availability of other v2i16 ops, that would need a new GPU.