Avoids repeated heap allocations/deallocations of layout.total_bytes per attention subtask and kv_out_mem per token step by storing worker_workspaces and kv_out_mem on AttentionActivations / AttentionActivationsPtrs. This eliminates ~20M minor page faults and improves decode throughput by +18% to +26% on AMD Turin. - #1025
Merged
Conversation
copybara-service
Bot
force-pushed
the
test_979178228
branch
2 times, most recently
from
September 11, 2026 14:29
4535389 to
f951eff
Compare
…` per attention subtask and `kv_out_mem` per token step by storing `worker_workspaces` and `kv_out_mem` on `AttentionActivations` / `AttentionActivationsPtrs`. This eliminates ~20M minor page faults and improves decode throughput by +18% to +26% on AMD Turin. PiperOrigin-RevId: 979835873
copybara-service
Bot
force-pushed
the
test_979178228
branch
from
September 11, 2026 15:35
f951eff to
ffc1abc
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Avoids repeated heap allocations/deallocations of
layout.total_bytesper attention subtask andkv_out_memper token step by storingworker_workspacesandkv_out_memonAttentionActivations/AttentionActivationsPtrs. This eliminates ~20M minor page faults and improves decode throughput by +18% to +26% on AMD Turin.