[TMA] Correctly get TMA Block Shape for SwizzledShared Blocks #7275
+20
−2
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
peterbell10
requested changes
Jun 23, 2025
Sorry, you’re right — there were a few commits I needed to cherry-pick after our cutoff to see the correct behavior. With those included, the descriptor is showing up as nvmma as expected. Appreciate the clarification. |
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Outdated suggestions cannot be applied.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.
When calculating the TMA Block Shape, we previously assumed the encoding was always an
NVMMASharedEncodingAttr. However, when the inner dimension is less than 8, the actual encoding is aSwizzledSharedEncodingAttr. Under our old assumption, casting toNVMMASharedEncodingAttrwould result in reading garbage values for fields likeelementBitWidth,swizzleBytes,fp4Padded,transposed, andpackedSize. In practice, this often led tofp4Paddedbeing incorrectly interpreted asTrue, causing the calculated Block Shape to be multiplied by 2. In certain cases, this results in unpredictable behavior during TMA load/store operations.When we have a
SwizzledSharedEncodingAttr, the only modification necessary to calculate the Block Shape is to limit each dimension to 256, the max size for a TMA block pointer.We also need to fix something similar when calculating the unpacked offset layout.
New contributor declaration
I am not making a trivial change, such as fixing a typo in a comment.
I have written a PR description following these
rules.
I have run
pre-commit run --from-ref origin/main --to-ref HEAD.Select one of the following.
/testforlittests/unittestfor C++ tests/python/testfor end-to-end teststest_tensor_descriptor_storeandtest_make_tensor_descriptor_matmulin test/unit/language/test_tensor_descriptor.py. And bytest_tensor_descriptor_reduce,test_tma_gather,test_tma_scatter,test_host_tensor_descriptor_load, andtest_host_tensor_descriptor_matmulin test/unit/cuda/test_tensor_descriptor.pySelect one of the following.
littests.littests I have added follow these best practices,including the "tests should be minimal" section. (Usually running Python code
and using the instructions it generates is not minimal.)