docs(torad): the aligned split's row, corrected: one slot prefills through the live-tile path and decodes through the mask-prefix one - #73
Merged
Conversation
…ive-tile path, and decodes through the mask-prefix one
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The train engine-10 past #72: 1 commit (3f8b2ee), TORAD.md only.
The row for 32e695e said the gate's one-slot legs "take the live-tile path too". Only their prefill does. The mask-prefix hint covers at most 16 rows (
fattn-common.cuh:Q->ne[1] <= 16), so a 512-row prefill ubatch splits live and its bits move with the aligned split. A one-slot decode takes the mask-prefix path, whose split was already a multiple of the output tiles (508 on a 5090), and its rate does not move: 101.5 against 100.7 tok/s at 245K. The 9 % is the multi-slot pool's.