GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-09-22 22:33:15 +08:00
Code Issues Packages Projects Releases Wiki Activity

Added the ability to divide a half float by an index

Browse Source
This commit is contained in:
Benoit Steiner 2016-03-23 09:46:42 -07:00
parent 6971146ca9
commit 0e68882604
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
Eigen/src/Core/arch/CUDA/Half.h
View File

@ -234,6 +234,12 @@ static inline EIGEN_DEVICE_FUNC bool operator > (const half& a, const half& b) {
#endif // Emulate support for half floats #endif // Emulate support for half floats
// Division by an index. Do it in full float precision to avoid accuracy
// issues in converting the denominator to half.
static inline EIGEN_DEVICE_FUNC half operator / (const half& a, Index b) {
return Eigen::half(static_cast<float>(a) / static_cast<float>(b));
}
// Conversion routines, including fallbacks for the host or older CUDA. // Conversion routines, including fallbacks for the host or older CUDA.
// Note that newer Intel CPUs (Haswell or newer) have vectorized versions of // Note that newer Intel CPUs (Haswell or newer) have vectorized versions of
// these in hardware. If we need more performance on older/other CPUs, they are // these in hardware. If we need more performance on older/other CPUs, they are