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