Adsk Contrib - Modify half-domain LUT1D GPU shader to improve zero handling#1981
Merged
doug-walker merged 2 commits intoSep 5, 2024
Merged
Conversation
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
zachlewis
approved these changes
Sep 4, 2024
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Made a small improvement in the GPU shader for half-domain LUT1D ops.
The half-domain LUT1D is structured like a half-float and so it has two entries for zero (positive and negative). The shader was looking up positive 0 at the negative 0 entry. Normally, a half-domain LUT1D should set +0 and -0 to the same value, so in practice this should not be an issue, but it is now more technically correct. I added a unit test to verify this (it failed prior to the shader adjustment, the GPU matched the CPU at -0 rather than +0).
In addition, I clarified the code that scales denorms in that shader. This is numerically equivalent but might be somewhat clearer to read.
Finally, I added an unrelated but potentially useful unit test to verify that the cacheID of a Processor containing a LUT3D changes if one of the LUT entries changes.