[Executorch][quant] Optimize per channel dequantize#5622
Closed
kimishpatel wants to merge 1 commit into
Closed
Conversation
When using quantized kv cache, dequantization routine takes significantly long. This diff just vectorizes dequant per channel for common case. Differential Revision: [D63338858](https://our.internmc.facebook.com/intern/diff/D63338858/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/5622
Note: Links to docs will display an error until the docs builds have been completed. ❌ 28 New FailuresAs of commit bed6c96 with merge base b2517d6 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
This was referenced Sep 25, 2024
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D63338858 |
This was referenced Sep 25, 2024
kimishpatel
added a commit
that referenced
this pull request
Sep 25, 2024
When using quantized kv cache, dequantization routine takes significantly long. This diff just vectorizes dequant per channel for common case. Differential Revision: [D63338858](https://our.internmc.facebook.com/intern/diff/D63338858/) ghstack-source-id: 244549440 Pull Request resolved: #5622
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Stack from ghstack (oldest at bottom):
When using quantized kv cache, dequantization routine takes significantly long.
This diff just vectorizes dequant per channel for common case.
Differential Revision: D63338858