Improve quality of f32/f64 generation. #103
Open
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 previous int-to-float conversion had a bias of probability 2^-24 / 2^-53 for types f32 / f64 respectively. The new conversion has a bias of 2^-64, which is the same bias as the underlying WyRand generator. The new conversion is a slightly shorter instruction sequence on x86 and ARM, but executes as 1 more uop on x86. Seems unlikely to harm performance much, if at all. https://rust.godbolt.org/z/q3zMxEc3T
This is unavailable on thumbv7m-none-eabi.
They both end up compiling to the same after inlining and simplification, but the former requires more work from the optimizer to get there.
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Outdated suggestions cannot be applied.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.
The previous int-to-float conversion had a bias of probability 2^-24 / 2^-53 for types f32 / f64 respectively. The new conversion has a bias of 2^-64, which is the same bias as the underlying WyRand generator. See comments in
lib.rsfor explanation.The new conversion is a slightly shorter instruction sequence on x86 and ARM, but executes as 1 more uop on x86. Seems unlikely to harm performance much, if at all. https://rust.godbolt.org/z/q3zMxEc3T.
The added tests in
smoke.rsfail with the old conversion and succeed with the new conversion.