Extend CompressAndTransposeQueriesMatrixAccumulationImpl and - #1007
Merged
Conversation
copybara-service
Bot
force-pushed
the
test_975094440
branch
3 times, most recently
from
September 3, 2026 09:23
01c50f9 to
0ac0678
Compare
CompressAndQuantizeQueriesMatrixAccumulationInt8Impl in gemma_cpp to support queries provided as BF16 in addition to float. Also update AbsMaxOfSpan to support BF16 inputs by promoting lower and upper vector halves to float in registers, computing absolute values, and reducing to the maximum float. PiperOrigin-RevId: 975615108
copybara-service
Bot
force-pushed
the
test_975094440
branch
from
September 3, 2026 09:44
0ac0678 to
f153c3b
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Extend CompressAndTransposeQueriesMatrixAccumulationImpl and
CompressAndQuantizeQueriesMatrixAccumulationInt8Impl in gemma_cpp
to support queries provided as BF16 in addition to float.
Also update AbsMaxOfSpan to support BF16 inputs by promoting lower
and upper vector halves to float in registers, computing absolute
values, and reducing to the maximum float.