ENH: Add max values comparison for floating point #18863
Merged
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
charris
changed the title
|
See |
charris
added
the
component: SIMD
seiko2plus
requested changes
Apr 28, 2021
There was a problem hiding this comment.
Your code here is following x86 behavior rather than IEEE, since If "one" of the two arguments is NaN, the value of the other argument is returned.
Our SIMD interface tries not to differentiate between one architecture and another. in other words, it tries to provide the maximum possible performance for all architectures.
Therefore, I suggest the following:
- make
npyv_max_[f32, f64]map directly to vmaxq_[f32, f64], _mm[,256,512]max[ps, pd], vec_max(VSX) which not oby the standards and also not provide a standardized behavior for NaN. - add new intrinics npyv_maxp_[f32, f64] that follow IEEE behavior.
note: extra p for precise
You will also need to add a testing unit case for both intrinsic which requires the following:
- bring the new intrinics to Python world through SIMD module, check _simd/_simd.dispatch.c.src
- add the new test cases into core/tests/test_simd.py
Thank you!
howjmay
force-pushed
the
simd-max
branch
17 times, most recently
from
6f2bf34
to
190f2b9
Compare
seiko2plus
requested changes
Apr 30, 2021
|
Thank you @seiko2plus ! I learned a lot from your reviews. I have fixed them, please take a look. |
howjmay
force-pushed
the
simd-max
branch
2 times, most recently
from
8c861e2
to
49948f9
Compare
howjmay
force-pushed
the
simd-max
branch
3 times, most recently
from
95bcd72
to
7e5669b
Compare
seiko2plus
reviewed
Apr 30, 2021
SIMD intrinsics for comparing double/single precision has been added. Now only NEON and SSE ones are added.
|
Thank you @howjmay |
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.
SIMD intrinsics for comparing double/single precision has been
added. Now only NEON and SSE ones are added.