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

Created the ptranspose packet primitive that can transpose an array of N packets, where N is the number of words in each packet. This primitive will be used to complete the vectorization of the gemm_pack_lhs and gemm_pack_rhs functions.

Browse Source 
Implemented the primitive using SSE instructions.
This commit is contained in:
Benoit Steiner
2014-03-26 19:03:07 -07:00
parent 14bc4b9704
commit a419cea4a0
4 changed files with 61 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
Eigen/src/Core/arch/SSE/Complex.h
View File

@@ -435,6 +435,16 @@ EIGEN_STRONG_INLINE Packet1cd pcplxflip/*<Packet1cd>*/(const Packet1cd& x)
return Packet1cd(preverse(x.v));
}
template<> EIGEN_DEVICE_FUNC inline void
ptranspose(Kernel<Packet2cf>& kernel) {
__m128d w1 = _mm_castps_pd(kernel.packet[0].v);
__m128d w2 = _mm_castps_pd(kernel.packet[1].v);
__m128 tmp = _mm_castpd_ps(_mm_unpackhi_pd(w1, w2));
kernel.packet[0].v = _mm_castpd_ps(_mm_unpacklo_pd(w1, w2));
kernel.packet[1].v = tmp;
}
} // end namespace internal
} // end namespace Eigen