1 // This file is part of Eigen, a lightweight C++ template library 2 // for linear algebra. 3 // 4 // Copyright (C) 2009 Rohit Garg <rpg.314 (at) gmail.com> 5 // Copyright (C) 2009-2010 Gael Guennebaud <gael.guennebaud (at) inria.fr> 6 // 7 // This Source Code Form is subject to the terms of the Mozilla 8 // Public License v. 2.0. If a copy of the MPL was not distributed 9 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/. 10 11 #ifndef EIGEN_GEOMETRY_SSE_H 12 #define EIGEN_GEOMETRY_SSE_H 13 14 namespace Eigen { 15 16 namespace internal { 17 18 template<class Derived, class OtherDerived> 19 struct quat_product<Architecture::SSE, Derived, OtherDerived, float, Aligned> 20 { 21 static inline Quaternion<float> run(const QuaternionBase<Derived>& _a, const QuaternionBase<OtherDerived>& _b) 22 { 23 const __m128 mask = _mm_castsi128_ps(_mm_setr_epi32(0,0,0,0x80000000)); 24 Quaternion<float> res; 25 __m128 a = _a.coeffs().template packet<Aligned>(0); 26 __m128 b = _b.coeffs().template packet<Aligned>(0); 27 __m128 flip1 = _mm_xor_ps(_mm_mul_ps(vec4f_swizzle1(a,1,2,0,2), 28 vec4f_swizzle1(b,2,0,1,2)),mask); 29 __m128 flip2 = _mm_xor_ps(_mm_mul_ps(vec4f_swizzle1(a,3,3,3,1), 30 vec4f_swizzle1(b,0,1,2,1)),mask); 31 pstore(&res.x(), 32 _mm_add_ps(_mm_sub_ps(_mm_mul_ps(a,vec4f_swizzle1(b,3,3,3,3)), 33 _mm_mul_ps(vec4f_swizzle1(a,2,0,1,0), 34 vec4f_swizzle1(b,1,2,0,0))), 35 _mm_add_ps(flip1,flip2))); 36 return res; 37 } 38 }; 39 40 template<typename VectorLhs,typename VectorRhs> 41 struct cross3_impl<Architecture::SSE,VectorLhs,VectorRhs,float,true> 42 { 43 static inline typename plain_matrix_type<VectorLhs>::type 44 run(const VectorLhs& lhs, const VectorRhs& rhs) 45 { 46 __m128 a = lhs.template packet<VectorLhs::Flags&AlignedBit ? Aligned : Unaligned>(0); 47 __m128 b = rhs.template packet<VectorRhs::Flags&AlignedBit ? Aligned : Unaligned>(0); 48 __m128 mul1=_mm_mul_ps(vec4f_swizzle1(a,1,2,0,3),vec4f_swizzle1(b,2,0,1,3)); 49 __m128 mul2=_mm_mul_ps(vec4f_swizzle1(a,2,0,1,3),vec4f_swizzle1(b,1,2,0,3)); 50 typename plain_matrix_type<VectorLhs>::type res; 51 pstore(&res.x(),_mm_sub_ps(mul1,mul2)); 52 return res; 53 } 54 }; 55 56 57 58 59 template<class Derived, class OtherDerived> 60 struct quat_product<Architecture::SSE, Derived, OtherDerived, double, Aligned> 61 { 62 static inline Quaternion<double> run(const QuaternionBase<Derived>& _a, const QuaternionBase<OtherDerived>& _b) 63 { 64 const Packet2d mask = _mm_castsi128_pd(_mm_set_epi32(0x0,0x0,0x80000000,0x0)); 65 66 Quaternion<double> res; 67 68 const double* a = _a.coeffs().data(); 69 Packet2d b_xy = _b.coeffs().template packet<Aligned>(0); 70 Packet2d b_zw = _b.coeffs().template packet<Aligned>(2); 71 Packet2d a_xx = pset1<Packet2d>(a[0]); 72 Packet2d a_yy = pset1<Packet2d>(a[1]); 73 Packet2d a_zz = pset1<Packet2d>(a[2]); 74 Packet2d a_ww = pset1<Packet2d>(a[3]); 75 76 // two temporaries: 77 Packet2d t1, t2; 78 79 /* 80 * t1 = ww*xy + yy*zw 81 * t2 = zz*xy - xx*zw 82 * res.xy = t1 +/- swap(t2) 83 */ 84 t1 = padd(pmul(a_ww, b_xy), pmul(a_yy, b_zw)); 85 t2 = psub(pmul(a_zz, b_xy), pmul(a_xx, b_zw)); 86 #ifdef EIGEN_VECTORIZE_SSE3 87 EIGEN_UNUSED_VARIABLE(mask) 88 pstore(&res.x(), _mm_addsub_pd(t1, preverse(t2))); 89 #else 90 pstore(&res.x(), padd(t1, pxor(mask,preverse(t2)))); 91 #endif 92 93 /* 94 * t1 = ww*zw - yy*xy 95 * t2 = zz*zw + xx*xy 96 * res.zw = t1 -/+ swap(t2) = swap( swap(t1) +/- t2) 97 */ 98 t1 = psub(pmul(a_ww, b_zw), pmul(a_yy, b_xy)); 99 t2 = padd(pmul(a_zz, b_zw), pmul(a_xx, b_xy)); 100 #ifdef EIGEN_VECTORIZE_SSE3 101 EIGEN_UNUSED_VARIABLE(mask) 102 pstore(&res.z(), preverse(_mm_addsub_pd(preverse(t1), t2))); 103 #else 104 pstore(&res.z(), psub(t1, pxor(mask,preverse(t2)))); 105 #endif 106 107 return res; 108 } 109 }; 110 111 } // end namespace internal 112 113 } // end namespace Eigen 114 115 #endif // EIGEN_GEOMETRY_SSE_H 116
