Improve rms_norm perf #727
Draft
Conversation
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
oulgen
force-pushed
the
oulgen/stack/108
branch
from
270db82 to
8bb79f4
Compare
meta-cla
bot
added
the
CLA Signed
xuanzhang816
reviewed
Oct 1, 2025
| normalized = x_tile * inv_rms_tile[:, None] | ||
| out[tile_m, :] = (normalized * weight[:].to(torch.float32)).to(out.dtype) | ||
| inv_rms[tile_m] = inv_rms_tile.to(out.dtype) | ||
| for tile_m in hl.tile(m): |
There was a problem hiding this comment.
That’s a smart way to address the performance issue!
Just to call out — longer term, it might be ideal for the autotuner to handle eviction policies and loop reduction setup automatically, rather than users specifying them directly.
There was a problem hiding this comment.
Yep, I agree, I was planning to implement that next but wanted to check if this is doable at all.
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Outdated suggestions cannot be applied.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.
Stacked PRs:
Improve rms_norm perf
Fixes #660
On my local 4080 laptop GPU perf improved from 3.37x to 6.12x. Will run
CI benchmarks on B200 to validate.