[ExecuTorch][llm] Fuse w1+w3 into single GEMM in quantized_moe_ffn - #21538
Merged
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21538
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 151 PendingAs of commit f7107f8 with merge base afffe4f ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
digantdesai
force-pushed
the
gh/digantdesai/74/orig
branch
from
August 1, 2026 13:57
265dc02 to
d5d8a4a
Compare
This PR needs a
|
digantdesai
force-pushed
the
gh/digantdesai/74/orig
branch
from
August 1, 2026 13:59
d5d8a4a to
674b468
Compare
Pull Request resolved: #21124 Fuse the up-projection (w1) and gate-projection (w3) into a single [2F, D] GEMM per expert. This halves the number of torchao activation quantizations per expert (from 2 to 1) and reduces total GEMM calls from 3 to 2 per active expert. At AOT time, w1 and w3 are concatenated before packing: pack_fn(cat([w1, w3], dim=0)). At runtime, a single expert_linear_dispatch produces [m_e, 2F], then a fused swiglu_and_compact pass reads the interleaved h1/h3 and writes [m_e, F] for the w2 down-projection. Schema changes from (packed_w1, packed_w3, packed_w2) to (packed_w13, packed_w2) — one fewer tensor arg (14 -> 13). ghstack-source-id: 409183086 @exported-using-ghexport Differential Revision: [D102799854](https://our.internmc.facebook.com/intern/diff/D102799854/)
digantdesai
force-pushed
the
gh/digantdesai/74/orig
branch
from
August 1, 2026 14:00
674b468 to
f7107f8
Compare
digantdesai
approved these changes
Aug 1, 2026
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
2 participants
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.
This PR was created by the merge bot to help merge the original PR into the main branch.
ghstack PR number: #21124 by @digantdesai
^ Please use this as the source of truth for the PR details, comments, and reviews
ghstack PR base: https://github.com/pytorch/executorch/tree/gh/digantdesai/74/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/digantdesai/74/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/gh/digantdesai/73/orig
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/digantdesai/74/orig
Differential Revision: D102799854
@diff-train-skip-merge