intrinsic_exponential.hpp
Go to the documentation of this file.
1 
29 #pragma once
30 
31 #include "setup.hpp"
32 
33 #if(!(GLM_ARCH & GLM_ARCH_SSE2))
34 # error "SSE2 instructions not supported or enabled"
35 #else
36 
37 namespace glm{
38 namespace detail
39 {
40 /*
41 GLM_FUNC_QUALIFIER __m128 sse_rsqrt_nr_ss(__m128 const x)
42 {
43  __m128 recip = _mm_rsqrt_ss( x ); // "estimate" opcode
44  const static __m128 three = { 3, 3, 3, 3 }; // aligned consts for fast load
45  const static __m128 half = { 0.5,0.5,0.5,0.5 };
46  __m128 halfrecip = _mm_mul_ss( half, recip );
47  __m128 threeminus_xrr = _mm_sub_ss( three, _mm_mul_ss( x, _mm_mul_ss ( recip, recip ) ) );
48  return _mm_mul_ss( halfrecip, threeminus_xrr );
49 }
50 
51 GLM_FUNC_QUALIFIER __m128 sse_normalize_fast_ps( float * RESTRICT vOut, float * RESTRICT vIn )
52 {
53  __m128 x = _mm_load_ss(&vIn[0]);
54  __m128 y = _mm_load_ss(&vIn[1]);
55  __m128 z = _mm_load_ss(&vIn[2]);
56 
57  const __m128 l = // compute x*x + y*y + z*z
58  _mm_add_ss(
59  _mm_add_ss( _mm_mul_ss(x,x),
60  _mm_mul_ss(y,y)
61  ),
62  _mm_mul_ss( z, z )
63  );
64 
65 
66  const __m128 rsqt = _mm_rsqrt_nr_ss( l );
67  _mm_store_ss( &vOut[0] , _mm_mul_ss( rsqt, x ) );
68  _mm_store_ss( &vOut[1] , _mm_mul_ss( rsqt, y ) );
69  _mm_store_ss( &vOut[2] , _mm_mul_ss( rsqt, z ) );
70 
71  return _mm_mul_ss( l , rsqt );
72 }
73 */
74 }//namespace detail
75 }//namespace glm
76 
77 #endif//GLM_ARCH


rtabmap
Author(s): Mathieu Labbe
autogenerated on Mon Dec 14 2020 03:34:59