Align the m dimension of pack_a outputs - #11136
Open
copybara-service[bot] wants to merge 1 commit into
Open
Conversation
`transpose_a` dot kernels read multiples of `tile_m` rows at a time. This change adds a new dimension that correctly requires the padding, instead of relying on incidental allocation padding. Additionally fixes two issues with `transpose_a` that broke multi-reduction and convolution targets: 1. Corrects multi-reduction dimension stride ordering in `define_transpose_a` so that intermediate reduction dimensions (k2, k3) precede `k1_outer`. 2. Disallows rewriting `transpose_a(stencil_copy)` to `stencil_copy(transpose_a)` when `tile_m > 1`, since `stencil_copy` cannot slide across tiled dimensions. PiperOrigin-RevId: 976076613
copybara-service
Bot
force-pushed
the
test_976076613
branch
from
September 4, 2026 04:31
dd102cb to
5f3633a
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Align the
mdimension ofpack_aoutputstranspose_adot kernels read multiples oftile_mrows at a time. This change adds a new dimension that correctly requires the padding, instead of relying on incidental allocation padding.Additionally fixes two issues with
transpose_athat broke multi-reduction and convolution targets:define_transpose_aso that intermediate reduction dimensions (k2, k3) precedek1_outer.transpose_a(stencil_copy)tostencil_copy(transpose_a)whentile_m > 1, sincestencil_copycannot slide across tiled dimensions.