diff options
| author | Sergey Sharybin <sergey.vfx@gmail.com> | 2015-03-13 10:14:43 +0300 |
|---|---|---|
| committer | Sergey Sharybin <sergey.vfx@gmail.com> | 2015-03-13 10:38:14 +0300 |
| commit | 61eab743f1377fdfcf44f2e4928290a3fc4ccfea (patch) | |
| tree | 6dff417678cc61e7096c03f2a6c05dd5e33d42c5 /build_files | |
| parent | aa4cb95a5c8569704f166cfd6d8f65606502ea40 (diff) | |
Cycles: Optimization for CMJ in CUDA kernels
Two things:
- Use intrinsics for clz/ctz (ctz is implemented via ffs()).
- Use faster sqrt() function which precision is enough for
integer values.
Diffstat (limited to 'build_files')
0 files changed, 0 insertions, 0 deletions