John Bauman | 8940182 | 2014-05-06 15:04:28 -0400 | [diff] [blame^] | 1 | // SwiftShader Software Renderer |
| 2 | // |
| 3 | // Copyright(c) 2005-2011 TransGaming Inc. |
| 4 | // |
| 5 | // All rights reserved. No part of this software may be copied, distributed, transmitted, |
| 6 | // transcribed, stored in a retrieval system, translated into any human or computer |
| 7 | // language by any means, or disclosed to third parties without the explicit written |
| 8 | // agreement of TransGaming Inc. Without such an agreement, no rights or licenses, express |
| 9 | // or implied, including but not limited to any patent rights, are granted to you. |
| 10 | // |
| 11 | |
| 12 | #include "ShaderCore.hpp" |
| 13 | |
| 14 | #include "Debug.hpp" |
| 15 | |
| 16 | namespace sw |
| 17 | { |
| 18 | void ShaderCore::mov(Color4f &dst, Color4f &src, bool floorToInteger) |
| 19 | { |
| 20 | if(floorToInteger) |
| 21 | { |
| 22 | dst.x = Floor(src.x); |
| 23 | } |
| 24 | else |
| 25 | { |
| 26 | dst = src; |
| 27 | } |
| 28 | } |
| 29 | |
| 30 | void ShaderCore::add(Color4f &dst, Color4f &src0, Color4f &src1) |
| 31 | { |
| 32 | dst.x = src0.x + src1.x; |
| 33 | dst.y = src0.y + src1.y; |
| 34 | dst.z = src0.z + src1.z; |
| 35 | dst.w = src0.w + src1.w; |
| 36 | } |
| 37 | |
| 38 | void ShaderCore::sub(Color4f &dst, Color4f &src0, Color4f &src1) |
| 39 | { |
| 40 | dst.x = src0.x - src1.x; |
| 41 | dst.y = src0.y - src1.y; |
| 42 | dst.z = src0.z - src1.z; |
| 43 | dst.w = src0.w - src1.w; |
| 44 | } |
| 45 | |
| 46 | void ShaderCore::mad(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2) |
| 47 | { |
| 48 | dst.x = src0.x * src1.x + src2.x; |
| 49 | dst.y = src0.y * src1.y + src2.y; |
| 50 | dst.z = src0.z * src1.z + src2.z; |
| 51 | dst.w = src0.w * src1.w + src2.w; |
| 52 | } |
| 53 | |
| 54 | void ShaderCore::mul(Color4f &dst, Color4f &src0, Color4f &src1) |
| 55 | { |
| 56 | dst.x = src0.x * src1.x; |
| 57 | dst.y = src0.y * src1.y; |
| 58 | dst.z = src0.z * src1.z; |
| 59 | dst.w = src0.w * src1.w; |
| 60 | } |
| 61 | |
| 62 | void ShaderCore::rcp(Color4f &dst, Color4f &src, bool pp) |
| 63 | { |
| 64 | Float4 rcp = reciprocal(src.x, pp, true); |
| 65 | |
| 66 | dst.x = rcp; |
| 67 | dst.y = rcp; |
| 68 | dst.z = rcp; |
| 69 | dst.w = rcp; |
| 70 | } |
| 71 | |
| 72 | void ShaderCore::rsq(Color4f &dst, Color4f &src, bool pp) |
| 73 | { |
| 74 | Float4 rsq = reciprocalSquareRoot(src.x, true, pp); |
| 75 | |
| 76 | dst.r = rsq; |
| 77 | dst.g = rsq; |
| 78 | dst.b = rsq; |
| 79 | dst.a = rsq; |
| 80 | } |
| 81 | |
| 82 | void ShaderCore::dp3(Color4f &dst, Color4f &src0, Color4f &src1) |
| 83 | { |
| 84 | Float4 dot = dot3(src0, src1); |
| 85 | |
| 86 | dst.x = dot; |
| 87 | dst.y = dot; |
| 88 | dst.z = dot; |
| 89 | dst.w = dot; |
| 90 | } |
| 91 | |
| 92 | void ShaderCore::dp4(Color4f &dst, Color4f &src0, Color4f &src1) |
| 93 | { |
| 94 | Float4 dot = dot4(src0, src1); |
| 95 | |
| 96 | dst.x = dot; |
| 97 | dst.y = dot; |
| 98 | dst.z = dot; |
| 99 | dst.w = dot; |
| 100 | } |
| 101 | |
| 102 | void ShaderCore::min(Color4f &dst, Color4f &src0, Color4f &src1) |
| 103 | { |
| 104 | dst.x = Min(src0.x, src1.x); |
| 105 | dst.y = Min(src0.y, src1.y); |
| 106 | dst.z = Min(src0.z, src1.z); |
| 107 | dst.w = Min(src0.w, src1.w); |
| 108 | } |
| 109 | |
| 110 | void ShaderCore::max(Color4f &dst, Color4f &src0, Color4f &src1) |
| 111 | { |
| 112 | dst.x = Max(src0.x, src1.x); |
| 113 | dst.y = Max(src0.y, src1.y); |
| 114 | dst.z = Max(src0.z, src1.z); |
| 115 | dst.w = Max(src0.w, src1.w); |
| 116 | } |
| 117 | |
| 118 | void ShaderCore::slt(Color4f &dst, Color4f &src0, Color4f &src1) |
| 119 | { |
| 120 | Int4 xMask = As<Int4>(CmpLT(src0.x, src1.x)); |
| 121 | Int4 yMask = As<Int4>(CmpLT(src0.y, src1.y)); |
| 122 | Int4 zMask = As<Int4>(CmpLT(src0.z, src1.z)); |
| 123 | Int4 wMask = As<Int4>(CmpLT(src0.w, src1.w)); |
| 124 | |
| 125 | Int4 iOne = As<Int4>(Float4(1, 1, 1, 1)); |
| 126 | |
| 127 | dst.x = As<Float4>(xMask & iOne); |
| 128 | dst.y = As<Float4>(yMask & iOne); |
| 129 | dst.z = As<Float4>(zMask & iOne); |
| 130 | dst.w = As<Float4>(wMask & iOne); |
| 131 | } |
| 132 | |
| 133 | void ShaderCore::sge(Color4f &dst, Color4f &src0, Color4f &src1) |
| 134 | { |
| 135 | Int4 xMask = As<Int4>(CmpNLT(src0.x, src1.x)); |
| 136 | Int4 yMask = As<Int4>(CmpNLT(src0.y, src1.y)); |
| 137 | Int4 zMask = As<Int4>(CmpNLT(src0.z, src1.z)); |
| 138 | Int4 wMask = As<Int4>(CmpNLT(src0.w, src1.w)); |
| 139 | |
| 140 | Int4 iOne = As<Int4>(Float4(1, 1, 1, 1)); |
| 141 | |
| 142 | dst.x = As<Float4>(xMask & iOne); |
| 143 | dst.y = As<Float4>(yMask & iOne); |
| 144 | dst.z = As<Float4>(zMask & iOne); |
| 145 | dst.w = As<Float4>(wMask & iOne); |
| 146 | } |
| 147 | |
| 148 | void ShaderCore::exp(Color4f &dst, Color4f &src, bool pp) |
| 149 | { |
| 150 | Float4 exp = exponential(src.x, pp); |
| 151 | |
| 152 | dst.x = exp; |
| 153 | dst.y = exp; |
| 154 | dst.z = exp; |
| 155 | dst.w = exp; |
| 156 | } |
| 157 | |
| 158 | void ShaderCore::log(Color4f &dst, Color4f &src, bool pp) |
| 159 | { |
| 160 | Float4 log = logarithm(src.x, true, pp); |
| 161 | |
| 162 | dst.x = log; |
| 163 | dst.y = log; |
| 164 | dst.z = log; |
| 165 | dst.w = log; |
| 166 | } |
| 167 | |
| 168 | void ShaderCore::lit(Color4f &dst, Color4f &src) |
| 169 | { |
| 170 | dst.x = Float4(1.0f, 1.0f, 1.0f, 1.0f); |
| 171 | dst.y = Max(src.x, Float4(0.0f, 0.0f, 0.0f, 0.0f)); |
| 172 | |
| 173 | Float4 pow; |
| 174 | |
| 175 | pow = src.w; |
| 176 | pow = Min(pow, Float4(127.9961f, 127.9961f, 127.9961f, 127.9961f)); |
| 177 | pow = Max(pow, Float4(-127.9961f, -127.9961f, -127.9961f, -127.9961f)); |
| 178 | |
| 179 | dst.z = power(src.y, pow); |
| 180 | dst.z = As<Float4>(As<Int4>(dst.z) & CmpNLT(src.x, Float4(0.0f, 0.0f, 0.0f, 0.0f))); |
| 181 | dst.z = As<Float4>(As<Int4>(dst.z) & CmpNLT(src.y, Float4(0.0f, 0.0f, 0.0f, 0.0f))); |
| 182 | |
| 183 | dst.w = Float4(1.0f, 1.0f, 1.0f, 1.0f); |
| 184 | } |
| 185 | |
| 186 | void ShaderCore::dst(Color4f &dst, Color4f &src0, Color4f &src1) |
| 187 | { |
| 188 | dst.x = 1; |
| 189 | dst.y = src0.y * src1.y; |
| 190 | dst.z = src0.z; |
| 191 | dst.w = src1.w; |
| 192 | } |
| 193 | |
| 194 | void ShaderCore::lrp(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2) |
| 195 | { |
| 196 | dst.x = src0.x * (src1.x - src2.x) + src2.x; |
| 197 | dst.y = src0.y * (src1.y - src2.y) + src2.y; |
| 198 | dst.z = src0.z * (src1.z - src2.z) + src2.z; |
| 199 | dst.w = src0.w * (src1.w - src2.w) + src2.w; |
| 200 | } |
| 201 | |
| 202 | void ShaderCore::frc(Color4f &dst, Color4f &src) |
| 203 | { |
| 204 | dst.x = Fraction(src.x); |
| 205 | dst.y = Fraction(src.y); |
| 206 | dst.z = Fraction(src.z); |
| 207 | dst.w = Fraction(src.w); |
| 208 | } |
| 209 | |
| 210 | void ShaderCore::pow(Color4f &dst, Color4f &src0, Color4f &src1, bool pp) |
| 211 | { |
| 212 | Float4 pow = power(src0.x, src1.x, pp); |
| 213 | |
| 214 | dst.x = pow; |
| 215 | dst.y = pow; |
| 216 | dst.z = pow; |
| 217 | dst.w = pow; |
| 218 | } |
| 219 | |
| 220 | void ShaderCore::crs(Color4f &dst, Color4f &src0, Color4f &src1) |
| 221 | { |
| 222 | dst.x = src0.y * src1.z - src0.z * src1.y; |
| 223 | dst.y = src0.z * src1.x - src0.x * src1.z; |
| 224 | dst.z = src0.x * src1.y - src0.y * src1.x; |
| 225 | } |
| 226 | |
| 227 | void ShaderCore::sgn(Color4f &dst, Color4f &src) |
| 228 | { |
| 229 | sgn(dst.x, src.x); |
| 230 | sgn(dst.y, src.y); |
| 231 | sgn(dst.z, src.z); |
| 232 | sgn(dst.w, src.w); |
| 233 | } |
| 234 | |
| 235 | void ShaderCore::abs(Color4f &dst, Color4f &src) |
| 236 | { |
| 237 | dst.x = Abs(src.x); |
| 238 | dst.y = Abs(src.y); |
| 239 | dst.z = Abs(src.z); |
| 240 | dst.w = Abs(src.w); |
| 241 | } |
| 242 | |
| 243 | void ShaderCore::nrm(Color4f &dst, Color4f &src, bool pp) |
| 244 | { |
| 245 | Float4 dot = dot3(src, src); |
| 246 | Float4 rsq = reciprocalSquareRoot(dot, false, pp); |
| 247 | |
| 248 | dst.x = src.x * rsq; |
| 249 | dst.y = src.y * rsq; |
| 250 | dst.z = src.z * rsq; |
| 251 | dst.w = src.w * rsq; |
| 252 | } |
| 253 | |
| 254 | void ShaderCore::sincos(Color4f &dst, Color4f &src, bool pp) |
| 255 | { |
| 256 | Float4 tmp0; |
| 257 | Float4 tmp1; |
| 258 | |
| 259 | tmp0 = src.x; |
| 260 | |
| 261 | // cos(x) = sin(x + pi/2) |
| 262 | tmp0 += Float4(1.57079632e+0f); |
| 263 | tmp1 = As<Float4>(CmpNLT(tmp0, Float4(3.14159265e+0f)) & As<Int4>(Float4(6.28318530e+0f))); |
| 264 | tmp0 -= tmp1; |
| 265 | |
| 266 | dst.x = sine(tmp0, pp); |
| 267 | dst.y = sine(src.x, pp); |
| 268 | } |
| 269 | |
| 270 | void ShaderCore::expp(Color4f &dst, Color4f &src, unsigned short version) |
| 271 | { |
| 272 | if(version < 0x0200) |
| 273 | { |
| 274 | Float4 frc = Fraction(src.x); |
| 275 | Float4 floor = src.x - frc; |
| 276 | |
| 277 | dst.x = exponential(floor, true); |
| 278 | dst.y = frc; |
| 279 | dst.z = exponential(src.x, true); |
| 280 | dst.w = Float4(1.0f, 1.0f, 1.0f, 1.0f); |
| 281 | } |
| 282 | else // Version >= 2.0 |
| 283 | { |
| 284 | exp(dst, src, true); // FIXME: 10-bit precision suffices |
| 285 | } |
| 286 | } |
| 287 | |
| 288 | void ShaderCore::logp(Color4f &dst, Color4f &src, unsigned short version) |
| 289 | { |
| 290 | if(version < 0x0200) |
| 291 | { |
| 292 | Float4 tmp0; |
| 293 | Float4 tmp1; |
| 294 | Float4 t; |
| 295 | Int4 r; |
| 296 | |
| 297 | tmp0 = Abs(src.x); |
| 298 | tmp1 = tmp0; |
| 299 | |
| 300 | // X component |
| 301 | r = As<Int4>(As<UInt4>(tmp0) >> 23) - Int4(127, 127, 127, 127); |
| 302 | dst.x = Float4(r); |
| 303 | |
| 304 | // Y component |
| 305 | dst.y = As<Float4>((As<Int4>(tmp1) & Int4(0x007FFFFF)) | As<Int4>(Float4(1.0f))); |
| 306 | |
| 307 | // Z component |
| 308 | dst.z = logarithm(src.x, true, true); |
| 309 | |
| 310 | // W component |
| 311 | dst.w = 1.0f; |
| 312 | } |
| 313 | else |
| 314 | { |
| 315 | log(dst, src, true); |
| 316 | } |
| 317 | } |
| 318 | |
| 319 | void ShaderCore::cmp(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2) |
| 320 | { |
| 321 | cmp(dst.x, src0.x, src1.x, src2.x); |
| 322 | cmp(dst.y, src0.y, src1.y, src2.y); |
| 323 | cmp(dst.z, src0.z, src1.z, src2.z); |
| 324 | cmp(dst.w, src0.w, src1.w, src2.w); |
| 325 | } |
| 326 | |
| 327 | void ShaderCore::dp2add(Color4f &dst, Color4f &src0, Color4f &src1, Color4f &src2) |
| 328 | { |
| 329 | Float4 t = src0.x * src1.x + src0.y * src1.y + src2.x; |
| 330 | |
| 331 | dst.x = t; |
| 332 | dst.y = t; |
| 333 | dst.z = t; |
| 334 | dst.w = t; |
| 335 | } |
| 336 | |
| 337 | void ShaderCore::sgn(Float4 &dst, Float4 &src) |
| 338 | { |
| 339 | Int4 neg = As<Int4>(CmpLT(src, Float4(0, 0, 0, 0))) & As<Int4>(Float4(-1, -1, -1, -1)); |
| 340 | Int4 pos = As<Int4>(CmpNLT(src, Float4(0, 0, 0, 0))) & As<Int4>(Float4(1, 1, 1, 1)); |
| 341 | dst = As<Float4>(neg | pos); |
| 342 | } |
| 343 | |
| 344 | void ShaderCore::cmp(Float4 &dst, Float4 &src0, Float4 &src1, Float4 &src2) |
| 345 | { |
| 346 | Int4 pos = CmpNLE(Float4(0.0f, 0.0f, 0.0f, 0.0f), src0); |
| 347 | Int4 t0 = pos & As<Int4>(src2); |
| 348 | Int4 t1 = ~pos & As<Int4>(src1); |
| 349 | dst = As<Float4>(t0 | t1); |
| 350 | } |
| 351 | |
| 352 | void ShaderCore::setp(Color4f &dst, Color4f &src0, Color4f &src1, Control control) |
| 353 | { |
| 354 | switch(control) |
| 355 | { |
| 356 | case Op::CONTROL_GT: |
| 357 | dst.x = As<Float4>(CmpNLE(src0.x, src1.x)); |
| 358 | dst.y = As<Float4>(CmpNLE(src0.y, src1.y)); |
| 359 | dst.z = As<Float4>(CmpNLE(src0.z, src1.z)); |
| 360 | dst.w = As<Float4>(CmpNLE(src0.w, src1.w)); |
| 361 | break; |
| 362 | case Op::CONTROL_EQ: |
| 363 | dst.x = As<Float4>(CmpEQ(src0.x, src1.x)); |
| 364 | dst.y = As<Float4>(CmpEQ(src0.y, src1.y)); |
| 365 | dst.z = As<Float4>(CmpEQ(src0.z, src1.z)); |
| 366 | dst.w = As<Float4>(CmpEQ(src0.w, src1.w)); |
| 367 | break; |
| 368 | case Op::CONTROL_GE: |
| 369 | dst.x = As<Float4>(CmpNLT(src0.x, src1.x)); |
| 370 | dst.y = As<Float4>(CmpNLT(src0.y, src1.y)); |
| 371 | dst.z = As<Float4>(CmpNLT(src0.z, src1.z)); |
| 372 | dst.w = As<Float4>(CmpNLT(src0.w, src1.w)); |
| 373 | break; |
| 374 | case Op::CONTROL_LT: |
| 375 | dst.x = As<Float4>(CmpLT(src0.x, src1.x)); |
| 376 | dst.y = As<Float4>(CmpLT(src0.y, src1.y)); |
| 377 | dst.z = As<Float4>(CmpLT(src0.z, src1.z)); |
| 378 | dst.w = As<Float4>(CmpLT(src0.w, src1.w)); |
| 379 | break; |
| 380 | case Op::CONTROL_NE: |
| 381 | dst.x = As<Float4>(CmpNEQ(src0.x, src1.x)); |
| 382 | dst.y = As<Float4>(CmpNEQ(src0.y, src1.y)); |
| 383 | dst.z = As<Float4>(CmpNEQ(src0.z, src1.z)); |
| 384 | dst.w = As<Float4>(CmpNEQ(src0.w, src1.w)); |
| 385 | break; |
| 386 | case Op::CONTROL_LE: |
| 387 | dst.x = As<Float4>(CmpLE(src0.x, src1.x)); |
| 388 | dst.y = As<Float4>(CmpLE(src0.y, src1.y)); |
| 389 | dst.z = As<Float4>(CmpLE(src0.z, src1.z)); |
| 390 | dst.w = As<Float4>(CmpLE(src0.w, src1.w)); |
| 391 | break; |
| 392 | default: |
| 393 | ASSERT(false); |
| 394 | } |
| 395 | } |
| 396 | } |