Introduce native kernels for NestedTensor elementwise ops (Add/Mul) #2483
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
Contributor
Author
|
UT result: |
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Outdated suggestions cannot be applied.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.
UT Failure Analysis
#2412
The current RuntimeError of test_nested_tensor_dense_elementwise_embedding_dim_128_xpu_float16 stems from an upstream dispatcher check that explicitly rejects the required Nested Tensor + Dense Tensor mixed-type signature, preventing our kernel from being called.
Resolution
This PR implements the high-performance op_dense_esuhm XPU kernel, enabling core elementwise binary operations (such as add and mul} for the Nested Tensor broadcasting case on XPU devices. The kernel correctly handles the [B, *, D] op [B, 1, D] geometry.
Next Step
This PR provides the necessary compute infrastructure, and its merger is a critical prerequisite for the subsequent PR pytorch/pytorch#169928 that will fix the dispatcher logic to enable full functionality.