1 // Copyright 2016 The SwiftShader Authors. All Rights Reserved. 2 // 3 // Licensed under the Apache License, Version 2.0 (the "License"); 4 // you may not use this file except in compliance with the License. 5 // You may obtain a copy of the License at 6 // 7 // http://www.apache.org/licenses/LICENSE-2.0 8 // 9 // Unless required by applicable law or agreed to in writing, software 10 // distributed under the License is distributed on an "AS IS" BASIS, 11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 // See the License for the specific language governing permissions and 13 // limitations under the License. 14 15 #ifndef sw_ShaderCore_hpp 16 #define sw_ShaderCore_hpp 17 18 #include "Shader.hpp" 19 #include "Reactor/Reactor.hpp" 20 #include "Common/Debug.hpp" 21 22 namespace sw 23 { 24 class Vector4s 25 { 26 public: 27 Vector4s(); 28 Vector4s(unsigned short x, unsigned short y, unsigned short z, unsigned short w); 29 Vector4s(const Vector4s &rhs); 30 31 Short4 &operator[](int i); 32 Vector4s &operator=(const Vector4s &rhs); 33 34 Short4 x; 35 Short4 y; 36 Short4 z; 37 Short4 w; 38 }; 39 40 class Vector4f 41 { 42 public: 43 Vector4f(); 44 Vector4f(float x, float y, float z, float w); 45 Vector4f(const Vector4f &rhs); 46 47 Float4 &operator[](int i); 48 Vector4f &operator=(const Vector4f &rhs); 49 50 Float4 x; 51 Float4 y; 52 Float4 z; 53 Float4 w; 54 }; 55 56 Float4 exponential2(RValue<Float4> x, bool pp = false); 57 Float4 logarithm2(RValue<Float4> x, bool abs, bool pp = false); 58 Float4 exponential(RValue<Float4> x, bool pp = false); 59 Float4 logarithm(RValue<Float4> x, bool abs, bool pp = false); 60 Float4 power(RValue<Float4> x, RValue<Float4> y, bool pp = false); 61 Float4 reciprocal(RValue<Float4> x, bool pp = false, bool finite = false, bool exactAtPow2 = false); 62 Float4 reciprocalSquareRoot(RValue<Float4> x, bool abs, bool pp = false); 63 Float4 modulo(RValue<Float4> x, RValue<Float4> y); 64 Float4 sine_pi(RValue<Float4> x, bool pp = false); // limited to [-pi, pi] range 65 Float4 cosine_pi(RValue<Float4> x, bool pp = false); // limited to [-pi, pi] range 66 Float4 sine(RValue<Float4> x, bool pp = false); 67 Float4 cosine(RValue<Float4> x, bool pp = false); 68 Float4 tangent(RValue<Float4> x, bool pp = false); 69 Float4 arccos(RValue<Float4> x, bool pp = false); 70 Float4 arcsin(RValue<Float4> x, bool pp = false); 71 Float4 arctan(RValue<Float4> x, bool pp = false); 72 Float4 arctan(RValue<Float4> y, RValue<Float4> x, bool pp = false); 73 Float4 sineh(RValue<Float4> x, bool pp = false); 74 Float4 cosineh(RValue<Float4> x, bool pp = false); 75 Float4 tangenth(RValue<Float4> x, bool pp = false); 76 Float4 arccosh(RValue<Float4> x, bool pp = false); // Limited to x >= 1 77 Float4 arcsinh(RValue<Float4> x, bool pp = false); 78 Float4 arctanh(RValue<Float4> x, bool pp = false); // Limited to ]-1, 1[ range 79 80 Float4 dot2(const Vector4f &v0, const Vector4f &v1); 81 Float4 dot3(const Vector4f &v0, const Vector4f &v1); 82 Float4 dot4(const Vector4f &v0, const Vector4f &v1); 83 84 void transpose4x4(Short4 &row0, Short4 &row1, Short4 &row2, Short4 &row3); 85 void transpose4x3(Short4 &row0, Short4 &row1, Short4 &row2, Short4 &row3); 86 void transpose4x4(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3); 87 void transpose4x3(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3); 88 void transpose4x2(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3); 89 void transpose4x1(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3); 90 void transpose2x4(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3); 91 void transpose4xN(Float4 &row0, Float4 &row1, Float4 &row2, Float4 &row3, int N); 92 93 class Register 94 { 95 public: 96 Register(const Reference<Float4> &x, const Reference<Float4> &y, const Reference<Float4> &z, const Reference<Float4> &w) : x(x), y(y), z(z), w(w) 97 { 98 } 99 100 Reference<Float4> &operator[](int i) 101 { 102 switch(i) 103 { 104 default: 105 case 0: return x; 106 case 1: return y; 107 case 2: return z; 108 case 3: return w; 109 } 110 } 111 112 Register &operator=(const Register &rhs) 113 { 114 x = rhs.x; 115 y = rhs.y; 116 z = rhs.z; 117 w = rhs.w; 118 119 return *this; 120 } 121 122 Register &operator=(const Vector4f &rhs) 123 { 124 x = rhs.x; 125 y = rhs.y; 126 z = rhs.z; 127 w = rhs.w; 128 129 return *this; 130 } 131 132 operator Vector4f() 133 { 134 Vector4f v; 135 136 v.x = x; 137 v.y = y; 138 v.z = z; 139 v.w = w; 140 141 return v; 142 } 143 144 Reference<Float4> x; 145 Reference<Float4> y; 146 Reference<Float4> z; 147 Reference<Float4> w; 148 }; 149 150 template<int S, bool D = false> 151 class RegisterArray 152 { 153 public: 154 RegisterArray(bool dynamic = D) : dynamic(dynamic) 155 { 156 if(dynamic) 157 { 158 x = new Array<Float4>(S); 159 y = new Array<Float4>(S); 160 z = new Array<Float4>(S); 161 w = new Array<Float4>(S); 162 } 163 else 164 { 165 x = new Array<Float4>[S]; 166 y = new Array<Float4>[S]; 167 z = new Array<Float4>[S]; 168 w = new Array<Float4>[S]; 169 } 170 } 171 172 ~RegisterArray() 173 { 174 if(dynamic) 175 { 176 delete x; 177 delete y; 178 delete z; 179 delete w; 180 } 181 else 182 { 183 delete[] x; 184 delete[] y; 185 delete[] z; 186 delete[] w; 187 } 188 } 189 190 Register operator[](int i) 191 { 192 if(dynamic) 193 { 194 return Register(x[0][i], y[0][i], z[0][i], w[0][i]); 195 } 196 else 197 { 198 return Register(x[i][0], y[i][0], z[i][0], w[i][0]); 199 } 200 } 201 202 Register operator[](RValue<Int> i) 203 { 204 ASSERT(dynamic); 205 206 return Register(x[0][i], y[0][i], z[0][i], w[0][i]); 207 } 208 209 private: 210 const bool dynamic; 211 Array<Float4> *x; 212 Array<Float4> *y; 213 Array<Float4> *z; 214 Array<Float4> *w; 215 }; 216 217 class ShaderCore 218 { 219 typedef Shader::Control Control; 220 221 public: 222 void mov(Vector4f &dst, const Vector4f &src, bool integerDestination = false); 223 void neg(Vector4f &dst, const Vector4f &src); 224 void ineg(Vector4f &dst, const Vector4f &src); 225 void f2b(Vector4f &dst, const Vector4f &src); 226 void b2f(Vector4f &dst, const Vector4f &src); 227 void f2i(Vector4f &dst, const Vector4f &src); 228 void i2f(Vector4f &dst, const Vector4f &src); 229 void f2u(Vector4f &dst, const Vector4f &src); 230 void u2f(Vector4f &dst, const Vector4f &src); 231 void i2b(Vector4f &dst, const Vector4f &src); 232 void b2i(Vector4f &dst, const Vector4f &src); 233 void add(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 234 void iadd(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 235 void sub(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 236 void isub(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 237 void mad(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 238 void imad(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 239 void mul(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 240 void imul(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 241 void rcpx(Vector4f &dst, const Vector4f &src, bool pp = false); 242 void div(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 243 void idiv(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 244 void udiv(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 245 void mod(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 246 void imod(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 247 void umod(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 248 void shl(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 249 void ishr(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 250 void ushr(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 251 void rsqx(Vector4f &dst, const Vector4f &src, bool pp = false); 252 void sqrt(Vector4f &dst, const Vector4f &src, bool pp = false); 253 void rsq(Vector4f &dst, const Vector4f &src, bool pp = false); 254 void len2(Float4 &dst, const Vector4f &src, bool pp = false); 255 void len3(Float4 &dst, const Vector4f &src, bool pp = false); 256 void len4(Float4 &dst, const Vector4f &src, bool pp = false); 257 void dist1(Float4 &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 258 void dist2(Float4 &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 259 void dist3(Float4 &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 260 void dist4(Float4 &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 261 void dp1(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 262 void dp2(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 263 void dp2add(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 264 void dp3(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 265 void dp4(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 266 void det2(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 267 void det3(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 268 void det4(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2, const Vector4f &src3); 269 void min(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 270 void imin(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 271 void umin(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 272 void max(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 273 void imax(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 274 void umax(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 275 void slt(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 276 void step(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 277 void exp2x(Vector4f &dst, const Vector4f &src, bool pp = false); 278 void exp2(Vector4f &dst, const Vector4f &src, bool pp = false); 279 void exp(Vector4f &dst, const Vector4f &src, bool pp = false); 280 void log2x(Vector4f &dst, const Vector4f &src, bool pp = false); 281 void log2(Vector4f &dst, const Vector4f &src, bool pp = false); 282 void log(Vector4f &dst, const Vector4f &src, bool pp = false); 283 void lit(Vector4f &dst, const Vector4f &src); 284 void att(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 285 void lrp(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 286 void isinf(Vector4f &dst, const Vector4f &src); 287 void isnan(Vector4f &dst, const Vector4f &src); 288 void smooth(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 289 void packHalf2x16(Vector4f &dst, const Vector4f &src); 290 void unpackHalf2x16(Vector4f &dst, const Vector4f &src); 291 void packSnorm2x16(Vector4f &dst, const Vector4f &src); 292 void packUnorm2x16(Vector4f &dst, const Vector4f &src); 293 void unpackSnorm2x16(Vector4f &dst, const Vector4f &src); 294 void unpackUnorm2x16(Vector4f &dst, const Vector4f &src); 295 void frc(Vector4f &dst, const Vector4f &src); 296 void trunc(Vector4f &dst, const Vector4f &src); 297 void floor(Vector4f &dst, const Vector4f &src); 298 void round(Vector4f &dst, const Vector4f &src); 299 void roundEven(Vector4f &dst, const Vector4f &src); 300 void ceil(Vector4f &dst, const Vector4f &src); 301 void powx(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 302 void pow(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 303 void crs(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 304 void forward1(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 305 void forward2(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 306 void forward3(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 307 void forward4(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 308 void reflect1(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 309 void reflect2(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 310 void reflect3(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 311 void reflect4(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 312 void refract1(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Float4 &src2); 313 void refract2(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Float4 &src2); 314 void refract3(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Float4 &src2); 315 void refract4(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Float4 &src2); 316 void sgn(Vector4f &dst, const Vector4f &src); 317 void isgn(Vector4f &dst, const Vector4f &src); 318 void abs(Vector4f &dst, const Vector4f &src); 319 void iabs(Vector4f &dst, const Vector4f &src); 320 void nrm2(Vector4f &dst, const Vector4f &src, bool pp = false); 321 void nrm3(Vector4f &dst, const Vector4f &src, bool pp = false); 322 void nrm4(Vector4f &dst, const Vector4f &src, bool pp = false); 323 void sincos(Vector4f &dst, const Vector4f &src, bool pp = false); 324 void cos(Vector4f &dst, const Vector4f &src, bool pp = false); 325 void sin(Vector4f &dst, const Vector4f &src, bool pp = false); 326 void tan(Vector4f &dst, const Vector4f &src, bool pp = false); 327 void acos(Vector4f &dst, const Vector4f &src, bool pp = false); 328 void asin(Vector4f &dst, const Vector4f &src, bool pp = false); 329 void atan(Vector4f &dst, const Vector4f &src, bool pp = false); 330 void atan2(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, bool pp = false); 331 void cosh(Vector4f &dst, const Vector4f &src, bool pp = false); 332 void sinh(Vector4f &dst, const Vector4f &src, bool pp = false); 333 void tanh(Vector4f &dst, const Vector4f &src, bool pp = false); 334 void acosh(Vector4f &dst, const Vector4f &src, bool pp = false); 335 void asinh(Vector4f &dst, const Vector4f &src, bool pp = false); 336 void atanh(Vector4f &dst, const Vector4f &src, bool pp = false); 337 void expp(Vector4f &dst, const Vector4f &src, unsigned short shaderModel); 338 void logp(Vector4f &dst, const Vector4f &src, unsigned short shaderModel); 339 void cmp0(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 340 void cmp(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, Control control); 341 void icmp(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, Control control); 342 void ucmp(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, Control control); 343 void select(Vector4f &dst, const Vector4f &src0, const Vector4f &src1, const Vector4f &src2); 344 void extract(Float4 &dst, const Vector4f &src0, const Float4 &src1); 345 void insert(Vector4f &dst, const Vector4f &src, const Float4 &element, const Float4 &index); 346 void all(Float4 &dst, const Vector4f &src); 347 void any(Float4 &dst, const Vector4f &src); 348 void bitwise_not(Vector4f &dst, const Vector4f &src); 349 void bitwise_or(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 350 void bitwise_xor(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 351 void bitwise_and(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 352 void equal(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 353 void notEqual(Vector4f &dst, const Vector4f &src0, const Vector4f &src1); 354 355 private: 356 void sgn(Float4 &dst, const Float4 &src); 357 void isgn(Float4 &dst, const Float4 &src); 358 void cmp0(Float4 &dst, const Float4 &src0, const Float4 &src1, const Float4 &src2); 359 void cmp0i(Float4 &dst, const Float4 &src0, const Float4 &src1, const Float4 &src2); 360 void select(Float4 &dst, RValue<Int4> src0, const Float4 &src1, const Float4 &src2); 361 void floatToHalfBits(Float4& dst, const Float4& floatBits, bool storeInUpperBits); 362 void halfToFloatBits(Float4& dst, const Float4& halfBits); 363 }; 364 } 365 366 #endif // sw_ShaderCore_hpp 367
