[ExecuTorch][WebGPU] Add aten.sub.Tensor with TensorMeta broadcast#20985
Open
JCNTH wants to merge 2 commits into
Open
[ExecuTorch][WebGPU] Add aten.sub.Tensor with TensorMeta broadcast#20985JCNTH wants to merge 2 commits into
JCNTH wants to merge 2 commits into
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20985
Note: Links to docs will display an error until the docs builds have been completed. ❌ 128 New Failures, 3 Unrelated Failures, 2 Unclassified FailuresAs of commit 92ff4d3 with merge base c2b273e ( NEW FAILURES - The following jobs have failed:
UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:
FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
This was referenced Jul 16, 2026
This was referenced Jul 16, 2026
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Stack from ghstack (oldest at bottom):
Add
aten.sub.Tensorwith TensorMeta broadcastProblem: The WebGPU backend had no
subop. A flat modulo-tiled broadcast (the old rigaddform) is wrong for a middle/spatial broadcast[N,C,1,1]— e.g. InstanceNorm'sx - mean, wheremeanis[N,C,1,1]broadcast acrossH,W. A flat tiling cannot express that middle/spatial broadcast, so the result is numerically wrong.Solution: Add
aten.sub.Tensormirroringmul's 3xTensorMetaUBO broadcast — the same-shape elementwise fast path plus an out-index relinearize on shape mismatch — with analpha_pipelineoverrideconstant so the kernel computesin1 - alpha_ * in2.Implementation:
runtime/ops/sub/binary_sub.wgsl+BinaryOp.cppregisteraten.sub.Tensor. The handler fills three broadcastTensorMetauniforms (out/in1/in2) and rebuilds them plus the dispatch on dynamic resize;alphais read from the scalar arg and baked into the pipeline as thealpha_override (fixed for the op's lifetime, so resize never rewrites it). Mirrors Vulkanaten.sub.Tensor/ themulop.Constraints: fp32-only (every operand
nbytes == numel * 4), rank <= 4 (NCHW).Co-authored-with: Claude Code.
Differential Revision: D112378933