Jj/cumsum nvfuserex opinfo tolerance #2586
Open
+14
−0
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
…race" This reverts commit 0354e9d. the promotion logic is also wrong. besides, having a type promotion to alter the behavior seems to be the wrong thing to do.
… original tests do run with un-promoted_math
for more information, see https://pre-commit.ci
…ance' into jj/cumsum_nvfuserex_opinfo_tolerance
jjsjann123
requested review from
mruberry,
lantiga,
t-vi and
KaelanDt
as code owners
|
Numerics looks pretty nasty for bf16/fp16. |
|
@naoyam test passed for me locally on your nvfuser branch. |
|
Linking the related PR NVIDIA/Fuser#5312 |
crcrpar
approved these changes
Oct 3, 2025
Looks like Naoya added that right before your review. I added the link in the PR description. |
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Outdated suggestions cannot be applied.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.
nvfuserex's new codegen support for cumsum runs math in reduced precision. as pytorch does.
It's failing opinfo test, since reference implementation uses double. Bumping the tolerance to keep CI happy.
Linking the related PR NVIDIA/Fuser#5312