Improve quantization utilities #598
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
meshopt_quantizeFloat and meshopt_quantizeHalf are moved to the new file. These are larger (esp. float->half), not as commonly used as unorm/snorm, don't benefit as much from inlining as it's more rare that they are used with constant arguments in perf-sensitive context, and also not used anywhere inside meshoptimizer itself, which means we still get to keep the independence between different translation units (we'll need to be careful with this in the future...). A new file makes for a convenient place to add more quantization utilities in the future as well. These are used in gltfpack, which doesn't see a noticeable perf degradation on geometry-heavy meshes that use quantizeFloat.
We now test both quantizeFloat and quantizeHalf fairly thoroughly via unit tests.
Also convert C comments to C++ - this is no longer part of meshoptimizer.h so it doesn't need to be C89 compatible.
This function reverses the transformation by meshopt_quantizeHalf; the reverse conversion is easier since exponent and mantissa just need to be expanded into the larger range. We still need to handle denormals and specials; unlike quantization, here we choose to preserve NaN payload because that allows us to fold inf and nan cases into a single conditional add.
meshopt_quantizeFloat does not require dequantization; meshopt_quantizeHalf now has a dequantize function; Snorm/Unorm variants have a trivial (divide-by-scale) dequantizer.
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Outdated suggestions cannot be applied.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.
meshopt_dequantizeHalfwhich can be used to reverse float->half quantization in cases when the result is needed on the CPUFixes #597.