Use a constant with clearly-defined type for log2e in fwd_kernel_splitK #1181
Conversation
This file contains 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
facebook-github-bot
added
the
CLA Signed
bottler
approved these changes
Dec 20, 2024
lw
approved these changes
Dec 20, 2024
| @@ -341,7 +341,8 @@ def _fwd_kernel_splitK( | |||
| # scale sm_scale by log_2(e) and use | |||
| # 2^x instead of exp in the loop because CSE and LICM | |||
| # don't work as expected with `exp` in the loop | |||
| qk_scale = sm_scale * 1.44269504 | |||
| log2e = tl.full((), 1.44269504, tl.float32) | |||
There was a problem hiding this comment.
Can we add a comment with an explanation/reference for why the code is this way?
There was a problem hiding this comment.
Yes definitely, good call
Summary: Triton 3.2 made some changes to its interpretation of constants (triton-lang/triton#4613) which makes Triton more consistent with pytorch/numpy, but cause some surprising issues with this kernel. Specifically it seems like log2e is interpreted as float32 in one instance and float64 in another, which leads to reduced prediction accuracy in some cases. To prevent this, let's make log2e a constant and define it as float32.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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:
Triton 3.2 made some changes to its interpretation of constants (triton-lang/triton#4613) which makes Triton more consistent with pytorch/numpy, but cause some surprising issues with this kernel. Specifically it seems like log2e is interpreted as float32 in one instance and float64 in another, which leads to reduced prediction accuracy in some cases.
To prevent this, let's make log2e a constant and define it as float32.