[ET-VK][qconv] Dynamically select between im2col path and general path #17387
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
This adds a dispatch layer to `q8ta_conv2d` that dynamically selects between the im2col-based and general convolution implementations at graph build time. The existing `q8ta_conv2d` function is renamed to `q8ta_conv2d_general`, and a new `q8ta_conv2d` dispatcher is introduced that chooses the im2col path when the convolution is non-grouped, has input channels divisible by 4, and kernel size ≤ 3x3. All other cases fall through to the general path. A separate `q8ta_conv2d_general` operator is also registered so tests can directly invoke the general path for comparison. The test suite is updated to exercise both the general and im2col implementations explicitly, and the default impl_selector is changed from "general" to empty (which triggers the new dispatcher). FP buffer storage types are removed from the test matrix since they are not needed. Differential Revision: [D93000162](https://our.internmc.facebook.com/intern/diff/D93000162/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/17387
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 1 Unrelated FailureAs of commit 1e5cfdd with merge base 964c565 ( NEW FAILURE - The following job has failed:
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 was referenced Feb 11, 2026
This PR needs a
|
…general path" This adds a dispatch layer to `q8ta_conv2d` that dynamically selects between the im2col-based and general convolution implementations at graph build time. The existing `q8ta_conv2d` function is renamed to `q8ta_conv2d_general`, and a new `q8ta_conv2d` dispatcher is introduced that chooses the im2col path when the convolution is non-grouped, has input channels divisible by 4, and kernel size ≤ 3x3. All other cases fall through to the general path. A separate `q8ta_conv2d_general` operator is also registered so tests can directly invoke the general path for comparison. The test suite is updated to exercise both the general and im2col implementations explicitly, and the default impl_selector is changed from "general" to empty (which triggers the new dispatcher). FP buffer storage types are removed from the test matrix since they are not needed. Differential Revision: [D93000162](https://our.internmc.facebook.com/intern/diff/D93000162/) [ghstack-poisoned]
…general path" This adds a dispatch layer to `q8ta_conv2d` that dynamically selects between the im2col-based and general convolution implementations at graph build time. The existing `q8ta_conv2d` function is renamed to `q8ta_conv2d_general`, and a new `q8ta_conv2d` dispatcher is introduced that chooses the im2col path when the convolution is non-grouped, has input channels divisible by 4, and kernel size ≤ 3x3. All other cases fall through to the general path. A separate `q8ta_conv2d_general` operator is also registered so tests can directly invoke the general path for comparison. The test suite is updated to exercise both the general and im2col implementations explicitly, and the default impl_selector is changed from "general" to empty (which triggers the new dispatcher). FP buffer storage types are removed from the test matrix since they are not needed. Differential Revision: [D93000162](https://our.internmc.facebook.com/intern/diff/D93000162/) [ghstack-poisoned]
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):
This adds a dispatch layer to
q8ta_conv2dthat dynamically selects between the im2col-based and general convolution implementations at graph build time. The existingq8ta_conv2dfunction is renamed toq8ta_conv2d_general, and a newq8ta_conv2ddispatcher is introduced that chooses the im2col path when the convolution is non-grouped, has input channels divisible by 4, and kernel size ≤ 3x3. All other cases fall through to the general path.A separate
q8ta_conv2d_generaloperator is also registered so tests can directly invoke the general path for comparison. The test suite is updated to exercise both the general and im2col implementations explicitly, and the default impl_selector is changed from "general" to empty (which triggers the new dispatcher). FP buffer storage types are removed from the test matrix since they are not needed.Differential Revision: D93000162