[ET-VK][qconv] Add layout-flexible impl of quantized depthwise conv2d #17263
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
pytorchbot
requested review from
SS-JIA,
kirklandsign and
larryliu0820
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/17263
Note: Links to docs will display an error until the docs builds have been completed. This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
SS-JIA
force-pushed
the
gh/SS-JIA/400/orig
branch
from
abfeb88 to
f310d2e
Compare
Pull Request resolved: #17108 Adds a new layout-agnostic quantized depthwise convolution operator `etvk.q8ta_conv2d_dw` that uses BufferMetadata and layout specialization constants to support arbitrary memory layouts (contiguous, channels-last, 4W4C block-packed, etc.). Key changes: 1. New shader `q8ta_conv2d_dw.glsl`: - Uses BufferMetadata for input/output tensor addressing - Layout-aware via `inp_layout`/`outp_layout` specialization constants - Processes 4 adjacent width positions × 4 channels per thread (4Wx4C tile) - Includes optimized paths for simple layouts (outer_block_size == 1) 2. New indexing utilities in `indexing.glslh`: - `texel_idx_to_tensor4d_idx()`: converts linear texel index to 4D tensor coords - `tensor4d_idx_to_texel_idx()`: converts 4D tensor index to texel index 3. Code refactoring: - Extract `Conv2DParams` struct and `create_conv2d_params()` to ConvolutionUtils.h - Create Q8taConv2dDW.cpp with new operator implementation - Add Q8taConv2d.h with public API declarations - Move `prepack_quantized_conv2d_dw_weight()` to new implementation file 4. New workgroup size helpers: - `pick_q8ta_conv2d_dw_global_wg_size()`: computes {W4, H, C4} dispatch size - `pick_q8ta_conv2d_dw_local_wg_size()`: adaptive local size based on tensor dims 5. Test updates: - Rename test to `test_q8_conv2d_dw.cpp` - Add `TestQ8taConv2d.cpp` with shared test utilities ghstack-source-id: 338638549 @exported-using-ghexport Differential Revision: [D92061368](https://our.internmc.facebook.com/intern/diff/D92061368/)
SS-JIA
force-pushed
the
gh/SS-JIA/401/orig
branch
from
cb35c3e to
b91589c
Compare
This PR needs a
|
SS-JIA
approved these changes
Feb 6, 2026
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
This PR was created by the merge bot to help merge the original PR into the main branch.
ghstack PR number: #17108 by @SS-JIA
^ Please use this as the source of truth for the PR details, comments, and reviews
ghstack PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/401/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/401/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/400/orig
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/401/orig
Differential Revision: D92061368
@diff-train-skip-merge