[release/2.3] [ROCm] Correct numerical issues in layer norm backwards kernel (#140259)#1766
Merged
pruthvistony merged 1 commit intorelease/2.3from Dec 6, 2024
Merged
Conversation
…ch#140259) It was raised that the backwards layer norm on AMD was slightly off the accuracy of the equivalent NVIDIA implementation. On AMD we call into a helper kernel `cuLoadWriteStridedInputs` which processes strided input and accumulates the partial gradients into shared memory. In this kernel (pytorch#87635) we truncated `mean` and `rstd` from T_ACC type to T which causes numerical issues in the warp buffers created in this kernel. This PR will use the correct accumulator type for mean and rstd. Note: Only AMD call into this call stack for backwards layer norm, so this was not an issue for NV. Pull Request resolved: pytorch#140259 Approved by: https://github.com/jianyuh (cherry picked from commit 001f736)
|
Jenkins build for 376941ea81baea6726add3a82a7e78144ed2bebe commit finished as FAILURE |
pruthvistony
approved these changes
Dec 6, 2024
This was referenced Dec 13, 2024
rocm-mici
pushed a commit
that referenced
this pull request
Dec 13, 2024
… kernel (pytorch#140259) (#1766) It was raised that the backwards layer norm on AMD was slightly off the accuracy of the equivalent NVIDIA implementation. On AMD we call into a helper kernel `cuLoadWriteStridedInputs` which processes strided input and accumulates the partial gradients into shared memory. In this kernel (pytorch#87635) we truncated `mean` and `rstd` from T_ACC type to T which causes numerical issues in the warp buffers created in this kernel. This PR will use the correct accumulator type for mean and rstd. Note: Only AMD call into this call stack for backwards layer norm, so this was not an issue for NV. Pull Request resolved: pytorch#140259 Approved by: https://github.com/jianyuh (cherry picked from commit 001f736)
rocm-mici
pushed a commit
that referenced
this pull request
Dec 13, 2024
… kernel (pytorch#140259) (#1766) It was raised that the backwards layer norm on AMD was slightly off the accuracy of the equivalent NVIDIA implementation. On AMD we call into a helper kernel `cuLoadWriteStridedInputs` which processes strided input and accumulates the partial gradients into shared memory. In this kernel (pytorch#87635) we truncated `mean` and `rstd` from T_ACC type to T which causes numerical issues in the warp buffers created in this kernel. This PR will use the correct accumulator type for mean and rstd. Note: Only AMD call into this call stack for backwards layer norm, so this was not an issue for NV. Pull Request resolved: pytorch#140259 Approved by: https://github.com/jianyuh (cherry picked from commit 001f736)
rocm-mici
pushed a commit
that referenced
this pull request
Dec 13, 2024
… kernel (pytorch#140259) (#1766) It was raised that the backwards layer norm on AMD was slightly off the accuracy of the equivalent NVIDIA implementation. On AMD we call into a helper kernel `cuLoadWriteStridedInputs` which processes strided input and accumulates the partial gradients into shared memory. In this kernel (pytorch#87635) we truncated `mean` and `rstd` from T_ACC type to T which causes numerical issues in the warp buffers created in this kernel. This PR will use the correct accumulator type for mean and rstd. Note: Only AMD call into this call stack for backwards layer norm, so this was not an issue for NV. Pull Request resolved: pytorch#140259 Approved by: https://github.com/jianyuh (cherry picked from commit 001f736)
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.
It was raised that the backwards layer norm on AMD was slightly off the accuracy of the equivalent NVIDIA implementation.
On AMD we call into a helper kernel
cuLoadWriteStridedInputswhich processes strided input and accumulates the partial gradients into shared memory.In this kernel (pytorch#87635) we truncated
meanandrstdfrom T_ACC type to T which causes numerical issues in the warp buffers created in this kernel. This PR will use the correct accumulator type for mean and rstd.Note: Only AMD call into this call stack for backwards layer norm, so this was not an issue for NV.
Pull Request resolved: pytorch#140259
Approved by: https://github.com/jianyuh
(cherry picked from commit 001f736)