Make linearstep definitely return 1 for x > edge1 #2064
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
|
The division (xclamped - edge0) / (edge1 - edge0) is not guaranteed to be exactly 1 for xclamped == edge1 if the backend floating point math has higher than expected error bounds. Eg. it is common on GPU backends to use faster but less precise instructions for division. The modified code is mathematically equivalent and it logically guarantees the returned value is exactly 1.0 for x > edge1 > edge0.
roelandschoukens
force-pushed
the
linearstep-01
branch
from
eefacef to
19e4b96
Compare
Collaborator
|
You seem to still need DCO and CLA fixed up |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
Description
The division (xclamped - edge0) / (edge1 - edge0) is not guaranteed to be exactly 1 for xclamped == edge1 if the backend floating point math has higher than expected error bounds. Eg. it is common on GPU backends to use faster but less precise instructions for division.
The modified code is mathematically equivalent and it logically guarantees the returned value is exactly 1.0 for x > edge1 > edge0.
Fixes #2063
Tests
This should not break the existing test for linearstep. A difference in behaviour will be observed only on back-ends that implement some form of fast math.
Checklist:
already run clang-format v17 before submitting, I definitely will look at
the CI test that runs clang-format and fix anything that it highlights as
being nonconforming.