| commit | 747f49ba67b8895a5831ab539de551b916f3738c | [log] [tgz] |
|---|---|---|
| author | Steven Fuerst <svfuerst@gmail.com> | Wed Aug 15 15:07:15 2012 -0700 |
| committer | Alex Deucher <alexander.deucher@amd.com> | Thu Sep 20 13:10:34 2012 -0400 |
| tree | b1b1561ae97f9f7fdab8ae44e019bd5dca0e62e4 | |
| parent | 7ff64fcaa7b7ba62d12758e49643f31dd9e90ece [diff] |
Replace int2float() with an optimized version. We use __fls() to find the most significant bit. Using that, the loop can be avoided. A second trick is to use the behaviour of the rotate instructions to expand the range of the unsigned int to float conversion to the full 32 bits in a branchless way. The routine is now exact up to 2^24. Above that, we truncate which is equivalent to rounding towards zero. Signed-off-by: Steven Fuerst <svfuerst@gmail.com>