[pallas:triton] Fix atomic min/max lowering for uint and float types #26604
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
|
I think we should follow Triton when making these choices. It looks like it does allow min/max for floating types, so Pallas Triton should too. Not sure what it does for nans, though? |
superbobry
approved these changes
Feb 20, 2025
google-ml-butler
bot
added
kokoro:force-run
pull ready
max will effectively propagate nans w/o sign bit set while min will will do so with nans with sign bit set. So behavior is not really that consistent. I'm guessing one could say that when using nans for min and max the result is undefined? XLA has a clear IEEE 754 semantics here (sans SNANs). It as a tradeoff. |
draganmladjenovic
force-pushed
the
min_max_atomics
branch
from
a9645d7 to
ae7f6b6
Compare
draganmladjenovic
force-pushed
the
min_max_atomics
branch
from
ae7f6b6 to
5a59b1d
Compare
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Outdated suggestions cannot be applied.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.
Floating point implementation is lifted from triton frontend itself. It has somewhat inconsistent semantics with regards to nan values. Should we leave it as is? Or implement it via cas loop matching np.max or np.maxnum semantics? Or should we allow min and max on fp types at all?