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

Some fixes/cleanups for numeric_limits & fix for related bug in psqrt

Browse Source
This commit is contained in:
Rasmus Munk Larsen
2022-01-07 01:10:17 +00:00
parent ed27d988c1
commit 96dc37a03b
6 changed files with 43 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -75,8 +75,6 @@ Packet4f pcos<Packet4f>(const Packet4f& _x)
return pcos_float(_x);
}
#if EIGEN_FAST_MATH
// Functions for sqrt.
// The EIGEN_FAST_MATH version uses the _mm_rsqrt_ps approximation and one step
// of Newton's method, at a cost of 1-2 bits of precision as opposed to the
@@ -85,11 +83,13 @@ Packet4f pcos<Packet4f>(const Packet4f& _x)
// it can be inlined and pipelined with other computations, further reducing its
// effective latency. This is similar to Quake3's fast inverse square root.
// For detail see here: http://www.beyond3d.com/content/articles/8/
template<> EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
#if EIGEN_FAST_MATH
template<>
EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
Packet4f psqrt<Packet4f>(const Packet4f& _x)
{
Packet4f minus_half_x = pmul(_x, pset1<Packet4f>(-0.5f));
Packet4f denormal_mask = pandnot(
const Packet4f minus_half_x = pmul(_x, pset1<Packet4f>(-0.5f));
const Packet4f denormal_mask = pandnot(
pcmp_lt(_x, pset1<Packet4f>((std::numeric_limits<float>::min)())),
pcmp_lt(_x, pzero(_x)));