1 | // Copyright 2009-2021 Intel Corporation |
2 | // SPDX-License-Identifier: Apache-2.0 |
3 | |
4 | #pragma once |
5 | |
6 | #include "math.h" |
7 | #include "vec3.h" |
8 | |
9 | namespace embree |
10 | { |
11 | //////////////////////////////////////////////////////////////////////////////// |
12 | /// Generic 4D vector Class |
13 | //////////////////////////////////////////////////////////////////////////////// |
14 | |
15 | template<typename T> struct Vec4 |
16 | { |
17 | enum { N = 4 }; |
18 | union { |
19 | struct { T x, y, z, w; }; |
20 | #if !(defined(__WIN32__) && _MSC_VER == 1800) // workaround for older VS 2013 compiler |
21 | T components[N]; |
22 | #endif |
23 | }; |
24 | |
25 | typedef T Scalar; |
26 | |
27 | //////////////////////////////////////////////////////////////////////////////// |
28 | /// Construction |
29 | //////////////////////////////////////////////////////////////////////////////// |
30 | |
31 | __forceinline Vec4( ) {} |
32 | __forceinline explicit Vec4( const T& a ) : x(a), y(a), z(a), w(a) {} |
33 | __forceinline Vec4( const T& x, const T& y, const T& z, const T& w ) : x(x), y(y), z(z), w(w) {} |
34 | __forceinline Vec4( const Vec3<T>& xyz, const T& w ) : x(xyz.x), y(xyz.y), z(xyz.z), w(w) {} |
35 | |
36 | __forceinline Vec4( const Vec4& other ) { x = other.x; y = other.y; z = other.z; w = other.w; } |
37 | __forceinline Vec4( const Vec3fx& other ); |
38 | |
39 | template<typename T1> __forceinline Vec4( const Vec4<T1>& a ) : x(T(a.x)), y(T(a.y)), z(T(a.z)), w(T(a.w)) {} |
40 | template<typename T1> __forceinline Vec4& operator =(const Vec4<T1>& other) { x = other.x; y = other.y; z = other.z; w = other.w; return *this; } |
41 | |
42 | __forceinline Vec4& operator =(const Vec4& other) { x = other.x; y = other.y; z = other.z; w = other.w; return *this; } |
43 | |
44 | __forceinline operator Vec3<T> () const { return Vec3<T>(x,y,z); } |
45 | |
46 | //////////////////////////////////////////////////////////////////////////////// |
47 | /// Constants |
48 | //////////////////////////////////////////////////////////////////////////////// |
49 | |
50 | __forceinline Vec4( ZeroTy ) : x(zero), y(zero), z(zero), w(zero) {} |
51 | __forceinline Vec4( OneTy ) : x(one), y(one), z(one), w(one) {} |
52 | __forceinline Vec4( PosInfTy ) : x(pos_inf), y(pos_inf), z(pos_inf), w(pos_inf) {} |
53 | __forceinline Vec4( NegInfTy ) : x(neg_inf), y(neg_inf), z(neg_inf), w(neg_inf) {} |
54 | |
55 | #if defined(__WIN32__) && (_MSC_VER == 1800) // workaround for older VS 2013 compiler |
56 | __forceinline const T& operator [](const size_t axis) const { assert(axis < 4); return (&x)[axis]; } |
57 | __forceinline T& operator [](const size_t axis) { assert(axis < 4); return (&x)[axis]; } |
58 | #else |
59 | __forceinline const T& operator [](const size_t axis ) const { assert(axis < 4); return components[axis]; } |
60 | __forceinline T& operator [](const size_t axis) { assert(axis < 4); return components[axis]; } |
61 | #endif |
62 | |
63 | //////////////////////////////////////////////////////////////////////////////// |
64 | /// Swizzles |
65 | //////////////////////////////////////////////////////////////////////////////// |
66 | |
67 | __forceinline Vec3<T> xyz() const { return Vec3<T>(x, y, z); } |
68 | }; |
69 | |
70 | //////////////////////////////////////////////////////////////////////////////// |
71 | /// Unary Operators |
72 | //////////////////////////////////////////////////////////////////////////////// |
73 | |
74 | template<typename T> __forceinline Vec4<T> operator +( const Vec4<T>& a ) { return Vec4<T>(+a.x, +a.y, +a.z, +a.w); } |
75 | template<typename T> __forceinline Vec4<T> operator -( const Vec4<T>& a ) { return Vec4<T>(-a.x, -a.y, -a.z, -a.w); } |
76 | template<typename T> __forceinline Vec4<T> abs ( const Vec4<T>& a ) { return Vec4<T>(abs (a.x), abs (a.y), abs (a.z), abs (a.w)); } |
77 | template<typename T> __forceinline Vec4<T> rcp ( const Vec4<T>& a ) { return Vec4<T>(rcp (a.x), rcp (a.y), rcp (a.z), rcp (a.w)); } |
78 | template<typename T> __forceinline Vec4<T> rsqrt ( const Vec4<T>& a ) { return Vec4<T>(rsqrt(a.x), rsqrt(a.y), rsqrt(a.z), rsqrt(a.w)); } |
79 | template<typename T> __forceinline Vec4<T> sqrt ( const Vec4<T>& a ) { return Vec4<T>(sqrt (a.x), sqrt (a.y), sqrt (a.z), sqrt (a.w)); } |
80 | |
81 | //////////////////////////////////////////////////////////////////////////////// |
82 | /// Binary Operators |
83 | //////////////////////////////////////////////////////////////////////////////// |
84 | |
85 | template<typename T> __forceinline Vec4<T> operator +( const Vec4<T>& a, const Vec4<T>& b ) { return Vec4<T>(a.x + b.x, a.y + b.y, a.z + b.z, a.w + b.w); } |
86 | template<typename T> __forceinline Vec4<T> operator -( const Vec4<T>& a, const Vec4<T>& b ) { return Vec4<T>(a.x - b.x, a.y - b.y, a.z - b.z, a.w - b.w); } |
87 | template<typename T> __forceinline Vec4<T> operator *( const Vec4<T>& a, const Vec4<T>& b ) { return Vec4<T>(a.x * b.x, a.y * b.y, a.z * b.z, a.w * b.w); } |
88 | template<typename T> __forceinline Vec4<T> operator *( const T& a, const Vec4<T>& b ) { return Vec4<T>(a * b.x, a * b.y, a * b.z, a * b.w); } |
89 | template<typename T> __forceinline Vec4<T> operator *( const Vec4<T>& a, const T& b ) { return Vec4<T>(a.x * b , a.y * b , a.z * b , a.w * b ); } |
90 | template<typename T> __forceinline Vec4<T> operator /( const Vec4<T>& a, const Vec4<T>& b ) { return Vec4<T>(a.x / b.x, a.y / b.y, a.z / b.z, a.w / b.w); } |
91 | template<typename T> __forceinline Vec4<T> operator /( const Vec4<T>& a, const T& b ) { return Vec4<T>(a.x / b , a.y / b , a.z / b , a.w / b ); } |
92 | template<typename T> __forceinline Vec4<T> operator /( const T& a, const Vec4<T>& b ) { return Vec4<T>(a / b.x, a / b.y, a / b.z, a / b.w); } |
93 | |
94 | template<typename T> __forceinline Vec4<T> min(const Vec4<T>& a, const Vec4<T>& b) { return Vec4<T>(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z), min(a.w, b.w)); } |
95 | template<typename T> __forceinline Vec4<T> max(const Vec4<T>& a, const Vec4<T>& b) { return Vec4<T>(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z), max(a.w, b.w)); } |
96 | |
97 | //////////////////////////////////////////////////////////////////////////////// |
98 | /// Ternary Operators |
99 | //////////////////////////////////////////////////////////////////////////////// |
100 | |
101 | template<typename T> __forceinline Vec4<T> madd ( const Vec4<T>& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>( madd(a.x,b.x,c.x), madd(a.y,b.y,c.y), madd(a.z,b.z,c.z), madd(a.w,b.w,c.w)); } |
102 | template<typename T> __forceinline Vec4<T> msub ( const Vec4<T>& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>( msub(a.x,b.x,c.x), msub(a.y,b.y,c.y), msub(a.z,b.z,c.z), msub(a.w,b.w,c.w)); } |
103 | template<typename T> __forceinline Vec4<T> nmadd ( const Vec4<T>& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>(nmadd(a.x,b.x,c.x),nmadd(a.y,b.y,c.y),nmadd(a.z,b.z,c.z),nmadd(a.w,b.w,c.w)); } |
104 | template<typename T> __forceinline Vec4<T> nmsub ( const Vec4<T>& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>(nmsub(a.x,b.x,c.x),nmsub(a.y,b.y,c.y),nmsub(a.z,b.z,c.z),nmsub(a.w,b.w,c.w)); } |
105 | |
106 | template<typename T> __forceinline Vec4<T> madd ( const T& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>( madd(a,b.x,c.x), madd(a,b.y,c.y), madd(a,b.z,c.z), madd(a,b.w,c.w)); } |
107 | template<typename T> __forceinline Vec4<T> msub ( const T& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>( msub(a,b.x,c.x), msub(a,b.y,c.y), msub(a,b.z,c.z), msub(a,b.w,c.w)); } |
108 | template<typename T> __forceinline Vec4<T> nmadd ( const T& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>(nmadd(a,b.x,c.x),nmadd(a,b.y,c.y),nmadd(a,b.z,c.z),nmadd(a,b.w,c.w)); } |
109 | template<typename T> __forceinline Vec4<T> nmsub ( const T& a, const Vec4<T>& b, const Vec4<T>& c) { return Vec4<T>(nmsub(a,b.x,c.x),nmsub(a,b.y,c.y),nmsub(a,b.z,c.z),nmsub(a,b.w,c.w)); } |
110 | |
111 | //////////////////////////////////////////////////////////////////////////////// |
112 | /// Assignment Operators |
113 | //////////////////////////////////////////////////////////////////////////////// |
114 | |
115 | template<typename T> __forceinline Vec4<T>& operator +=( Vec4<T>& a, const Vec4<T>& b ) { a.x += b.x; a.y += b.y; a.z += b.z; a.w += b.w; return a; } |
116 | template<typename T> __forceinline Vec4<T>& operator -=( Vec4<T>& a, const Vec4<T>& b ) { a.x -= b.x; a.y -= b.y; a.z -= b.z; a.w -= b.w; return a; } |
117 | template<typename T> __forceinline Vec4<T>& operator *=( Vec4<T>& a, const T& b ) { a.x *= b ; a.y *= b ; a.z *= b ; a.w *= b ; return a; } |
118 | template<typename T> __forceinline Vec4<T>& operator /=( Vec4<T>& a, const T& b ) { a.x /= b ; a.y /= b ; a.z /= b ; a.w /= b ; return a; } |
119 | |
120 | //////////////////////////////////////////////////////////////////////////////// |
121 | /// Reduction Operators |
122 | //////////////////////////////////////////////////////////////////////////////// |
123 | |
124 | template<typename T> __forceinline T reduce_add( const Vec4<T>& a ) { return a.x + a.y + a.z + a.w; } |
125 | template<typename T> __forceinline T reduce_mul( const Vec4<T>& a ) { return a.x * a.y * a.z * a.w; } |
126 | template<typename T> __forceinline T reduce_min( const Vec4<T>& a ) { return min(a.x, a.y, a.z, a.w); } |
127 | template<typename T> __forceinline T reduce_max( const Vec4<T>& a ) { return max(a.x, a.y, a.z, a.w); } |
128 | |
129 | //////////////////////////////////////////////////////////////////////////////// |
130 | /// Comparison Operators |
131 | //////////////////////////////////////////////////////////////////////////////// |
132 | |
133 | template<typename T> __forceinline bool operator ==( const Vec4<T>& a, const Vec4<T>& b ) { return a.x == b.x && a.y == b.y && a.z == b.z && a.w == b.w; } |
134 | template<typename T> __forceinline bool operator !=( const Vec4<T>& a, const Vec4<T>& b ) { return a.x != b.x || a.y != b.y || a.z != b.z || a.w != b.w; } |
135 | template<typename T> __forceinline bool operator < ( const Vec4<T>& a, const Vec4<T>& b ) { |
136 | if (a.x != b.x) return a.x < b.x; |
137 | if (a.y != b.y) return a.y < b.y; |
138 | if (a.z != b.z) return a.z < b.z; |
139 | if (a.w != b.w) return a.w < b.w; |
140 | return false; |
141 | } |
142 | |
143 | //////////////////////////////////////////////////////////////////////////////// |
144 | /// Shift Operators |
145 | //////////////////////////////////////////////////////////////////////////////// |
146 | |
147 | template<typename T> __forceinline Vec4<T> shift_right_1( const Vec4<T>& a ) { |
148 | return Vec4<T>(shift_right_1(a.x),shift_right_1(a.y),shift_right_1(a.z),shift_right_1(a.w)); |
149 | } |
150 | |
151 | //////////////////////////////////////////////////////////////////////////////// |
152 | /// Euclidean Space Operators |
153 | //////////////////////////////////////////////////////////////////////////////// |
154 | |
155 | template<typename T> __forceinline T dot ( const Vec4<T>& a, const Vec4<T>& b ) { return madd(a.x,b.x,madd(a.y,b.y,madd(a.z,b.z,a.w*b.w))); } |
156 | |
157 | template<typename T> __forceinline T length ( const Vec4<T>& a ) { return sqrt(dot(a,a)); } |
158 | template<typename T> __forceinline Vec4<T> normalize( const Vec4<T>& a ) { return a*rsqrt(dot(a,a)); } |
159 | template<typename T> __forceinline T distance ( const Vec4<T>& a, const Vec4<T>& b ) { return length(a-b); } |
160 | |
161 | //////////////////////////////////////////////////////////////////////////////// |
162 | /// Select |
163 | //////////////////////////////////////////////////////////////////////////////// |
164 | |
165 | template<typename T> __forceinline Vec4<T> select ( bool s, const Vec4<T>& t, const Vec4<T>& f ) { |
166 | return Vec4<T>(select(s,t.x,f.x),select(s,t.y,f.y),select(s,t.z,f.z),select(s,t.w,f.w)); |
167 | } |
168 | |
169 | template<typename T> __forceinline Vec4<T> select ( const Vec4<bool>& s, const Vec4<T>& t, const Vec4<T>& f ) { |
170 | return Vec4<T>(select(s.x,t.x,f.x),select(s.y,t.y,f.y),select(s.z,t.z,f.z),select(s.w,t.w,f.w)); |
171 | } |
172 | |
173 | template<typename T> __forceinline Vec4<T> select ( const typename T::Bool& s, const Vec4<T>& t, const Vec4<T>& f ) { |
174 | return Vec4<T>(select(s,t.x,f.x),select(s,t.y,f.y),select(s,t.z,f.z),select(s,t.w,f.w)); |
175 | } |
176 | |
177 | template<typename T> |
178 | __forceinline Vec4<T> lerp(const Vec4<T>& v0, const Vec4<T>& v1, const T& t) { |
179 | return madd(Vec4<T>(T(1.0f)-t),v0,t*v1); |
180 | } |
181 | |
182 | //////////////////////////////////////////////////////////////////////////////// |
183 | /// Output Operators |
184 | //////////////////////////////////////////////////////////////////////////////// |
185 | |
186 | template<typename T> __forceinline embree_ostream operator<<(embree_ostream cout, const Vec4<T>& a) { |
187 | return cout << "(" << a.x << ", " << a.y << ", " << a.z << ", " << a.w << ")" ; |
188 | } |
189 | |
190 | //////////////////////////////////////////////////////////////////////////////// |
191 | /// Default template instantiations |
192 | //////////////////////////////////////////////////////////////////////////////// |
193 | |
194 | typedef Vec4<bool > Vec4b; |
195 | typedef Vec4<unsigned char> Vec4uc; |
196 | typedef Vec4<int > Vec4i; |
197 | typedef Vec4<float > Vec4f; |
198 | } |
199 | |
200 | #include "vec3ba.h" |
201 | #include "vec3ia.h" |
202 | #include "vec3fa.h" |
203 | |
204 | //////////////////////////////////////////////////////////////////////////////// |
205 | /// SSE / AVX / MIC specializations |
206 | //////////////////////////////////////////////////////////////////////////////// |
207 | |
208 | #if defined(__SSE__) || defined(__ARM_NEON) |
209 | #include "../simd/sse.h" |
210 | #endif |
211 | |
212 | #if defined __AVX__ |
213 | #include "../simd/avx.h" |
214 | #endif |
215 | |
216 | #if defined __AVX512F__ |
217 | #include "../simd/avx512.h" |
218 | #endif |
219 | |
220 | namespace embree |
221 | { |
222 | template<> __forceinline Vec4<float>::Vec4( const Vec3fx& a ) { x = a.x; y = a.y; z = a.z; w = a.w; } |
223 | |
224 | #if defined(__AVX__) |
225 | template<> __forceinline Vec4<vfloat4>::Vec4( const Vec3fx& a ) { |
226 | x = a.x; y = a.y; z = a.z; w = a.w; |
227 | } |
228 | #elif defined(__SSE__) || defined(__ARM_NEON) |
229 | template<> __forceinline Vec4<vfloat4>::Vec4( const Vec3fx& a ) { |
230 | const vfloat4 v = vfloat4(a.m128); x = shuffle<0,0,0,0>(v); y = shuffle<1,1,1,1>(v); z = shuffle<2,2,2,2>(v); w = shuffle<3,3,3,3>(v); |
231 | } |
232 | #endif |
233 | |
234 | #if defined(__AVX__) |
235 | template<> __forceinline Vec4<vfloat8>::Vec4( const Vec3fx& a ) { |
236 | x = a.x; y = a.y; z = a.z; w = a.w; |
237 | } |
238 | #endif |
239 | |
240 | #if defined(__AVX512F__) |
241 | template<> __forceinline Vec4<vfloat16>::Vec4( const Vec3fx& a ) : x(a.x), y(a.y), z(a.z), w(a.w) {} |
242 | #endif |
243 | } |
244 | |