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

mixing types in product step 2:

Browse Source 
* pload* and pset1 are now templated on the packet type
* gemv routines are now embeded into a structure with
  a consistent API with respect to gemm
* some configurations of vector * matrix and matrix * matrix works fine,
  some need more work...
This commit is contained in:
Gael Guennebaud
2010-07-11 15:48:30 +02:00
parent 4161b8be67
commit ff96c94043
20 changed files with 600 additions and 549 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
Eigen/src/Core/arch/SSE/MathFunctions.h
View File

@@ -378,14 +378,14 @@ Packet4f ei_pcos<Packet4f>(const Packet4f& _x)
template<> EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
Packet4f ei_psqrt<Packet4f>(const Packet4f& _x)
{
Packet4f half = ei_pmul(_x, ei_pset1(.5f));
/* select only the inverse sqrt of non-zero inputs */
Packet4f non_zero_mask = _mm_cmpgt_ps(_x, ei_pset1(std::numeric_limits<float>::epsilon()));
Packet4f x = _mm_and_ps(non_zero_mask, _mm_rsqrt_ps(_x));
Packet4f half = ei_pmul(_x, ei_pset1<Packet4f>(.5f));
x = ei_pmul(x, ei_psub(ei_pset1(1.5f), ei_pmul(half, ei_pmul(x,x))));
return ei_pmul(_x,x);
/* select only the inverse sqrt of non-zero inputs */
Packet4f non_zero_mask = _mm_cmpgt_ps(_x, ei_pset1<Packet4f>(std::numeric_limits<float>::epsilon()));
Packet4f x = _mm_and_ps(non_zero_mask, _mm_rsqrt_ps(_x));
x = ei_pmul(x, ei_psub(ei_pset1<Packet4f>(1.5f), ei_pmul(half, ei_pmul(x,x))));
return ei_pmul(_x,x);
}
#endif // EIGEN_MATH_FUNCTIONS_SSE_H