Two-pass optimization for channelwise gated delta rule kernel (#21020)#21020
Open
JakeStevens wants to merge 1 commit into
Open
Two-pass optimization for channelwise gated delta rule kernel (#21020)#21020JakeStevens wants to merge 1 commit into
JakeStevens wants to merge 1 commit into
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21020
Note: Links to docs will display an error until the docs builds have been completed. ❌ 3 New FailuresAs of commit f3e0cfb with merge base f8b7f62 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
Contributor
|
@JakeStevens has exported this pull request. If you are a Meta employee, you can view the originating Diff in D112596724. |
This PR needs a
|
meta-codesync
Bot
changed the title
JakeStevens
added a commit
to JakeStevens/executorch
that referenced
this pull request
Jul 17, 2026
…h#21020) Summary: Reduce the channelwise gated delta rule kernel from four state traversals to two per token. The first pass computes the predicted value while folding in decay; the second applies decay and the rank-one update while accumulating the output. This preserves operation grouping and numerical behavior while reducing state-memory traffic. Add a standalone microbenchmark covering decode and representative prefill lengths. Differential Revision: D112596724
JakeStevens
force-pushed
the
export-D112596724
branch
from
July 17, 2026 19:56
dcb49cf to
1681b14
Compare
JakeStevens
added a commit
to JakeStevens/executorch
that referenced
this pull request
Jul 17, 2026
…h#21020) Summary: Reduce the channelwise gated delta rule kernel from four state traversals to two per token. The first pass computes the predicted value while folding in decay; the second applies decay and the rank-one update while accumulating the output. This preserves operation grouping and numerical behavior while reducing state-memory traffic. Add a standalone microbenchmark covering decode and representative prefill lengths. Differential Revision: D112596724
JakeStevens
force-pushed
the
export-D112596724
branch
from
July 17, 2026 20:38
1681b14 to
baee844
Compare
JakeStevens
added a commit
to JakeStevens/executorch
that referenced
this pull request
Jul 17, 2026
…h#21020) Summary: Reduce the channelwise gated delta rule kernel from four state traversals to two per token. The first pass computes the predicted value while folding in decay; the second applies decay and the rank-one update while accumulating the output. This preserves operation grouping and numerical behavior while reducing state-memory traffic. Add a standalone microbenchmark covering decode and representative prefill lengths. Differential Revision: D112596724
…h#21020) Summary: Reduce the channelwise gated delta rule kernel from four state traversals to two per token. The first pass computes the predicted value while folding in decay; the second applies decay and the rank-one update while accumulating the output. This preserves operation grouping and numerical behavior while reducing state-memory traffic. Add a standalone microbenchmark covering decode and representative prefill lengths. Reviewed By: billmguo Differential Revision: D112596724
JakeStevens
force-pushed
the
export-D112596724
branch
from
July 20, 2026 13:38
baee844 to
f3e0cfb
Compare
JakeStevens
added a commit
to JakeStevens/executorch
that referenced
this pull request
Jul 20, 2026
…h#21020) Summary: Reduce the channelwise gated delta rule kernel from four state traversals to two per token. The first pass computes the predicted value while folding in decay; the second applies decay and the rank-one update while accumulating the output. This preserves operation grouping and numerical behavior while reducing state-memory traffic. Add a standalone microbenchmark covering decode and representative prefill lengths. Reviewed By: billmguo Differential Revision: D112596724
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.
Summary:
Reduce the channelwise gated delta rule kernel from four state traversals to two per token. The first pass computes the predicted value while folding in decay; the second applies decay and the rank-one update while accumulating the output. This preserves operation grouping and numerical behavior while reducing state-memory traffic. Add a standalone microbenchmark covering decode and representative prefill lengths.
Reviewed By: billmguo
Differential Revision: D112596724