Match numpy sign semantics #8330
Open
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
|
I've been looking more at the PTX code generated. I think the original method of |
kmaehashi
added
cat:bug
|
Note that numpy 2.0 redefines the complex sign to return |
prutschman-iv
force-pushed
the
match-numpy-sign-semantics
branch
from
1d27ab7 to
e1628c0
Compare
|
I have verified the changes with the main branch. I believe this is ready for merge. |
Once guarded by the isnan test, the (x > 0) - (x < 0) has the correct semantics, and produces tighter PTX code.
prutschman-iv
force-pushed
the
match-numpy-sign-semantics
branch
from
e1628c0 to
0c48c6f
Compare
|
This pull request is now in conflicts. Could you fix it @prutschman-iv? 🙏 |
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Outdated suggestions cannot be applied.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.
Here is my first pass at modifying the
signfunction to matchnumpy's NaN semantics. It includes a test case usingcheck_unary_inf_nan. This is to fix #8327I've confirmed that the tests pass on v12, and have rebased to main, but am currently having build difficulties on main, so this is a draft for now.
The non-complex logic is based directly on numpy's logic. The static_casts are there to avoid a compiler error about ambiguous int casting for the ternary operator when the template is used with integer types.