[ET-VK] Fused RMSNorm operator to fix fp16 overflow#18772
Merged
meta-codesync[bot] merged 3 commits intogh/SS-JIA/518/basefrom Apr 9, 2026
Merged
[ET-VK] Fused RMSNorm operator to fix fp16 overflow#18772meta-codesync[bot] merged 3 commits intogh/SS-JIA/518/basefrom
meta-codesync[bot] merged 3 commits intogh/SS-JIA/518/basefrom
Conversation
Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/) [ghstack-poisoned]
This was referenced Apr 8, 2026
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/18772
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (2 Unrelated Failures)As of commit 9864405 with merge base 4afd7f9 ( BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
SS-JIA
pushed a commit
that referenced
this pull request
Apr 8, 2026
Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/) ghstack-source-id: 364237333 Pull Request resolved: #18772
meta-cla
Bot
added
the
CLA Signed
This PR needs a
|
Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/) [ghstack-poisoned]
SS-JIA
pushed a commit
that referenced
this pull request
Apr 8, 2026
Pull Request resolved: #18772 Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. ghstack-source-id: 364280899 @exported-using-ghexport Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/)
manuelcandales
approved these changes
Apr 8, 2026
Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/) [ghstack-poisoned]
SS-JIA
pushed a commit
that referenced
this pull request
Apr 8, 2026
Pull Request resolved: #18772 Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. ghstack-source-id: 364514329 @exported-using-ghexport Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/)
meta-codesync
Bot
merged commit 76a5a52
into
gh/SS-JIA/518/base
159 of 164 checks passed
SS-JIA
pushed a commit
that referenced
this pull request
Apr 9, 2026
Pull Request resolved: #18772 Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. ghstack-source-id: 364514329 @exported-using-ghexport Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/)
SS-JIA
pushed a commit
that referenced
this pull request
Apr 9, 2026
Pull Request resolved: #18772 Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. ghstack-source-id: 364514329 @exported-using-ghexport Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/)
jpiat
pushed a commit
to jpiat/executorch
that referenced
this pull request
Apr 14, 2026
Pull Request resolved: pytorch#18772 Fused RMSNorm operator that performs squaring, mean, rsqrt, and weight scaling in a single shader dispatch. All accumulation is done in fp32 regardless of input dtype, preventing fp16 overflow when residual stream values exceed sqrt(65504) ≈ 256. The Python reference impl (`rms_norm_impl`) must preserve the input dtype — PyTorch type promotion would otherwise produce fp32 output from fp16 inputs, and the FusePatternsPass re-trace would propagate that incorrect dtype through the graph. Authored by Claude. ghstack-source-id: 364514329 @exported-using-ghexport Differential Revision: [D99841211](https://our.internmc.facebook.com/intern/diff/D99841211/)
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.
Stack from ghstack (oldest at bottom):
Fused RMSNorm operator that performs squaring, mean, rsqrt, and
weight scaling in a single shader dispatch. All accumulation is done
in fp32 regardless of input dtype, preventing fp16 overflow when
residual stream values exceed sqrt(65504) ≈ 256.
The Python reference impl (
rms_norm_impl) must preserve the inputdtype — PyTorch type promotion would otherwise produce fp32 output
from fp16 inputs, and the FusePatternsPass re-trace would propagate
that incorrect dtype through the graph.
Authored by Claude.
Differential Revision: D99841211