BUG: np.tan(np.inf) test failure in Apple silicon #19974
Merged
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
<
8000
/div>
Numpy has a test failure where `np.tan(np.inf)` is not raising "invalid". The bug is due to an underlying bug in Apple's Libm. The bug is only present on arm64 and does not affect x86_64. The changes here apply to sin, cos, and tan. If the input is a non-finite value, then return `(x - x)`, which will raise "invalid" for `x=INFINITY` as well as return `NAN` for both `INFINITY` and `NAN` as input. Testing: (Note, the testing below is on top of another branch that fixes divide-by-zero failures in reciprocal) Before change: ``` FAILED numpy/core/tests/test_umath.py::TestSpecialFloats::test_tan - AssertionError: FloatingPointError not raised by tan 1 failed, 15589 passed, 302 skipped, 1268 deselected, 31 xfailed, 3 xpassed in 70.99s (0:01:10) ``` After change: ``` 15590 passed, 302 skipped, 1268 deselected, 31 xfailed, 3 xpassed in 71.38s (0:01:11) ```
|
Looks like you should also update the blocklist in |
Thanks, taking a look. |
There are no alternative implementations available so the #undef approach doesn't look like it would work here |
|
Thanks @Developer-Ecosystem-Engineering . |
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Outdated suggestions cannot be applied.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.
Numpy has a test failure where
np.tan(np.inf)is not raising "invalid". The bug is due to an underlying bug in Apple's Libm. The bug is only present on arm64 and does not affect x86_64.The changes here apply to sin, cos, and tan. If the input is a non-finite value, then return
(x - x), which will raise "invalid" forx=INFINITYas well as returnNANfor bothINFINITYandNANas input.Testing:
On top of changes for #19926
Before change:
After change: