Fix vectorized MLIR derivative constants#2962
Open
jacobleft wants to merge 3 commits into
Open
Conversation
jacobleft
force-pushed
the
jacobleft/fix-vector-constantfp
branch
from
July 20, 2026 14:38
647f4d1 to
092febe
Compare
jacobleft
marked this pull request as draft
July 20, 2026 14:54
jacobleft
force-pushed
the
jacobleft/fix-vector-constantfp
branch
from
July 20, 2026 15:14
092febe to
a98cc16
Compare
jacobleft
marked this pull request as ready for review
July 20, 2026 16:02
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #2961.
Change
Zero-operand floating constants emitted into derivative expressions now use the derivative shadow type. The generator also propagates that the produced value is already vectorized, preventing the generic vectorization path from broadcasting it a second time.
An equivalent native derivative rule now uses the operand-less representation, so the core MLIR test suite directly exercises this code path.
Regression
test/MLIR/ForwardMode/batched_derivative_constant.mlirfails against this PR's parent while constructingarith.addf(tensor<2x4xf32>, tensor<4xf32>). With this change, it verifies that the generated1has typetensor<2x4xf32>.Invariant
A generated zero-operand derivative constant must have the same shadow type and vectorization state as the tangent expression it participates in.
This replaces the closed #2955.