GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-08-14 04:35:57 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix plog(+inf) with AVX512

Browse Source
This commit is contained in:
Gael Guennebaud 2019-01-09 16:53:37 +01:00
parent 47810cf5b7
commit 3492a1ca74
1 changed files with 10 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
Eigen/src/Core/arch/AVX512/MathFunctions.h
View File

@ -47,6 +47,7 @@ plog<Packet16f>(const Packet16f& _x) {
// The smallest non denormalized float number.
_EIGEN_DECLARE_CONST_Packet16f_FROM_INT(min_norm_pos, 0x00800000);
_EIGEN_DECLARE_CONST_Packet16f_FROM_INT(minus_inf, 0xff800000);
_EIGEN_DECLARE_CONST_Packet16f_FROM_INT(pos_inf, 0x7f800000);
_EIGEN_DECLARE_CONST_Packet16f_FROM_INT(nan, 0x7fc00000);
// Polynomial coefficients.
@ -116,9 +117,15 @@ plog<Packet16f>(const Packet16f& _x) {
x = padd(x, y);
x = padd(x, y2);
// Filter out invalid inputs, i.e. negative arg will be NAN, 0 will be -INF.
__mmask16 pos_inf_mask = _mm512_cmp_ps_mask(_x,p16f_pos_inf,_CMP_EQ_OQ);
// Filter out invalid inputs, i.e.:
// - negative arg will be NAN,
// - 0 will be -INF.
// - +INF will be +INF
return _mm512_mask_blend_ps(iszero_mask,
_mm512_mask_blend_ps(invalid_mask, x, p16f_nan),
_mm512_mask_blend_ps(invalid_mask,
_mm512_mask_blend_ps(pos_inf_mask,x,p16f_pos_inf),
p16f_nan),
p16f_minus_inf);
}
#endif