devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

replaced _mm_prefetch in GeneralBlockPanelKernel.h, with ei_prefetch() inline function.

Browse Source 
Implemented NEON and AltiVec versions, copied SSE version over from GeneralBlockPanelKernel.h.
Also in GCC case (or rather !_MSC_VER) it's implemented using __builtin_prefetch().
NEON managed to give a small but welcome boost, 0.88GFLOPS -> 0.91GFLOPS.
This commit is contained in:
oem
2010-04-24 00:44:14 +03:00
parent e3e34b5920
commit 6972c140f7
5 changed files with 36 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
Eigen/src/Core/arch/SSE/PacketMath.h
View File

@@ -233,6 +233,10 @@ template<> EIGEN_STRONG_INLINE void ei_pstoreu<double>(double* to, const Packet2
template<> EIGEN_STRONG_INLINE void ei_pstoreu<float>(float* to, const Packet4f& from) { EIGEN_DEBUG_UNALIGNED_STORE ei_pstoreu((double*)to, _mm_castps_pd(from)); }
template<> EIGEN_STRONG_INLINE void ei_pstoreu<int>(int* to, const Packet4i& from) { EIGEN_DEBUG_UNALIGNED_STORE ei_pstoreu((double*)to, _mm_castsi128_pd(from)); }
template<> EIGEN_STRONG_INLINE void ei_prefetch<float>(const float* addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); }
template<> EIGEN_STRONG_INLINE void ei_prefetch<double>(const double* addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); }
template<> EIGEN_STRONG_INLINE void ei_prefetch<int>(const int* addr) { _mm_prefetch((const char*)(addr), _MM_HINT_T0); }
#if defined(_MSC_VER) && (_MSC_VER <= 1500) && defined(_WIN64)
// The temporary variable fixes an internal compilation error.
// Direct of the struct members fixed bug #62.