[ET-VK][ez] Constrain out repsets individually #15748
Merged
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
added 3 commits
November 10, 2025 13:32
Pull Request resolved: #15702 Title says it all! With the way the pass is currently written only the first arg will be inspected for q/dq node replacement. As a consequence, the second arg for i.e. binary ops may not have the quantized op be replaced. ghstack-source-id: 322214453 @exported-using-ghexport Differential Revision: [D86674169](https://our.internmc.facebook.com/intern/diff/D86674169/)
Pull Request resolved: #15703 Title says it all! This diff allows quantize/dequantize ops to consume/produce tensors in the `CONTIGUOUS_BUFFER` layout. This can help reduce the number of memory layout transitions needed to execute a model. ghstack-source-id: 322214457 @exported-using-ghexport Differential Revision: [D86674166](https://our.internmc.facebook.com/intern/diff/D86674166/)
Pull Request resolved: #15704 Address the TODO comment in the `tag_memory_meta_pass.py` graph pass. ``` # TODO(ssjia): For most ops, inputs and outputs must be synchronized, so there # is no need to constrain output repsets explicitly. Currently, the exceptions # (i.e. choose qparams) already define constrined repsets for the output, so # there is again no need to explicitly constrain the outputs. If an operator # appears later on that does not sync input and output representations, and # defines ambiguous repsets for the output tensor(s), then we will need to add # additional logic to this function to constrain the output repsets separately # from the input repsets. ``` This condition is now fulfilled with the below diff. ghstack-source-id: 322214459 @exported-using-ghexport Differential Revision: [D86674164](https://our.internmc.facebook.com/intern/diff/D86674164/)
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15748
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 111 PendingAs of commit 27bed93 with merge base aba44fd ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
SS-JIA
approved these changes
Nov 11, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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: #15704 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/362/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/362/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/gh/SS-JIA/361/orig
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/SS-JIA/362/orig
Differential Revision: D86674164
@diff-train-skip-merge