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

Enable vectorization of pack_rhs with a column-major RHS.

Browse Source 
Rename and generalize Kernel<*> to PacketBlock<*,N>.
This commit is contained in:
Gael Guennebaud
2014-04-25 10:56:18 +02:00
parent b0e19db1cf
commit 3d8d0f6269
11 changed files with 64 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -452,7 +452,7 @@ PALIGN_NEON(3,Packet4i,vextq_s32)
#undef PALIGN_NEON
template<> EIGEN_DEVICE_FUNC inline void
ptranspose(Kernel<Packet4f>& kernel) {
ptranspose(PacketBlock<Packet4f,4>& kernel) {
float32x4x2_t tmp1 = vzipq_f32(kernel.packet[0], kernel.packet[1]);
float32x4x2_t tmp2 = vzipq_f32(kernel.packet[2], kernel.packet[3]);
@@ -463,7 +463,7 @@ ptranspose(Kernel<Packet4f>& kernel) {
}
template<> EIGEN_DEVICE_FUNC inline void
ptranspose(Kernel<Packet4i>& kernel) {
ptranspose(PacketBlock<Packet4i,4>& kernel) {
int32x4x2_t tmp1 = vzipq_s32(kernel.packet[0], kernel.packet[1]);
int32x4x2_t tmp2 = vzipq_s32(kernel.packet[2], kernel.packet[3]);
kernel.packet[0] = vcombine_s32(vget_low_s32(tmp1.val[0]), vget_low_s32(tmp2.val[0]));