Add "scalar" quantize kernels, and use them to implement convert assuming they exist.#9591
Merged
copybara-service[bot] merged 1 commit intomasterfrom Feb 25, 2026
Merged
Conversation
copybara-service
bot
force-pushed
the
test_874865981
branch
from
1282e60 to
2f1c6fc
Compare
…ming they exist. This is a stepping stone towards removing the "params" design, which comes from XNNPACK. I think the params design is a headache and needs to go. This will unblock a number of improvements, such as computing the reciprocal of the scales as a separate op (or computing integer scaling parameters), which will allow lifting it out of the inner loops, even for ops like channelwise requantization. PiperOrigin-RevId: 874898592
copybara-service
bot
force-pushed
the
test_874865981
branch
from
2f1c6fc to
6425d79
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add "scalar" quantize kernels, and use them to implement convert assuming they exist.
This is a stepping stone towards removing the "params" design, which comes from XNNPACK. I think the params design is a headache and needs to go. This will unblock a number of improvements, such as computing the reciprocal of the scales as a separate op (or computing integer scaling parameters), which will allow lifting it out of the inner loops, even for ops like channelwise requantization.