1 #include "rs_types.rsh" 2 3 extern float2 __attribute__((overloadable)) convert_float2(int2 c); 4 extern float3 __attribute__((overloadable)) convert_float3(int3 c); 5 extern float4 __attribute__((overloadable)) convert_float4(int4 c); 6 7 // Float ops, 6.11.2 8 9 #define FN_FUNC_FN(fnc) \ 10 extern float2 __attribute__((overloadable)) fnc(float2 v) { \ 11 float2 r; \ 12 r.x = fnc(v.x); \ 13 r.y = fnc(v.y); \ 14 return r; \ 15 } \ 16 extern float3 __attribute__((overloadable)) fnc(float3 v) { \ 17 float3 r; \ 18 r.x = fnc(v.x); \ 19 r.y = fnc(v.y); \ 20 r.z = fnc(v.z); \ 21 return r; \ 22 } \ 23 extern float4 __attribute__((overloadable)) fnc(float4 v) { \ 24 float4 r; \ 25 r.x = fnc(v.x); \ 26 r.y = fnc(v.y); \ 27 r.z = fnc(v.z); \ 28 r.w = fnc(v.w); \ 29 return r; \ 30 } 31 32 #define IN_FUNC_FN(fnc) \ 33 extern int2 __attribute__((overloadable)) fnc(float2 v) { \ 34 int2 r; \ 35 r.x = fnc(v.x); \ 36 r.y = fnc(v.y); \ 37 return r; \ 38 } \ 39 extern int3 __attribute__((overloadable)) fnc(float3 v) { \ 40 int3 r; \ 41 r.x = fnc(v.x); \ 42 r.y = fnc(v.y); \ 43 r.z = fnc(v.z); \ 44 return r; \ 45 } \ 46 extern int4 __attribute__((overloadable)) fnc(float4 v) { \ 47 int4 r; \ 48 r.x = fnc(v.x); \ 49 r.y = fnc(v.y); \ 50 r.z = fnc(v.z); \ 51 r.w = fnc(v.w); \ 52 return r; \ 53 } 54 55 #define FN_FUNC_FN_FN(fnc) \ 56 extern float2 __attribute__((overloadable)) fnc(float2 v1, float2 v2) { \ 57 float2 r; \ 58 r.x = fnc(v1.x, v2.x); \ 59 r.y = fnc(v1.y, v2.y); \ 60 return r; \ 61 } \ 62 extern float3 __attribute__((overloadable)) fnc(float3 v1, float3 v2) { \ 63 float3 r; \ 64 r.x = fnc(v1.x, v2.x); \ 65 r.y = fnc(v1.y, v2.y); \ 66 r.z = fnc(v1.z, v2.z); \ 67 return r; \ 68 } \ 69 extern float4 __attribute__((overloadable)) fnc(float4 v1, float4 v2) { \ 70 float4 r; \ 71 r.x = fnc(v1.x, v2.x); \ 72 r.y = fnc(v1.y, v2.y); \ 73 r.z = fnc(v1.z, v2.z); \ 74 r.w = fnc(v1.w, v2.w); \ 75 return r; \ 76 } 77 78 #define FN_FUNC_FN_F(fnc) \ 79 extern float2 __attribute__((overloadable)) fnc(float2 v1, float v2) { \ 80 float2 r; \ 81 r.x = fnc(v1.x, v2); \ 82 r.y = fnc(v1.y, v2); \ 83 return r; \ 84 } \ 85 extern float3 __attribute__((overloadable)) fnc(float3 v1, float v2) { \ 86 float3 r; \ 87 r.x = fnc(v1.x, v2); \ 88 r.y = fnc(v1.y, v2); \ 89 r.z = fnc(v1.z, v2); \ 90 return r; \ 91 } \ 92 extern float4 __attribute__((overloadable)) fnc(float4 v1, float v2) { \ 93 float4 r; \ 94 r.x = fnc(v1.x, v2); \ 95 r.y = fnc(v1.y, v2); \ 96 r.z = fnc(v1.z, v2); \ 97 r.w = fnc(v1.w, v2); \ 98 return r; \ 99 } 100 101 #define FN_FUNC_FN_IN(fnc) \ 102 extern float2 __attribute__((overloadable)) fnc(float2 v1, int2 v2) { \ 103 float2 r; \ 104 r.x = fnc(v1.x, v2.x); \ 105 r.y = fnc(v1.y, v2.y); \ 106 return r; \ 107 } \ 108 extern float3 __attribute__((overloadable)) fnc(float3 v1, int3 v2) { \ 109 float3 r; \ 110 r.x = fnc(v1.x, v2.x); \ 111 r.y = fnc(v1.y, v2.y); \ 112 r.z = fnc(v1.z, v2.z); \ 113 return r; \ 114 } \ 115 extern float4 __attribute__((overloadable)) fnc(float4 v1, int4 v2) { \ 116 float4 r; \ 117 r.x = fnc(v1.x, v2.x); \ 118 r.y = fnc(v1.y, v2.y); \ 119 r.z = fnc(v1.z, v2.z); \ 120 r.w = fnc(v1.w, v2.w); \ 121 return r; \ 122 } 123 124 #define FN_FUNC_FN_I(fnc) \ 125 extern float2 __attribute__((overloadable)) fnc(float2 v1, int v2) { \ 126 float2 r; \ 127 r.x = fnc(v1.x, v2); \ 128 r.y = fnc(v1.y, v2); \ 129 return r; \ 130 } \ 131 extern float3 __attribute__((overloadable)) fnc(float3 v1, int v2) { \ 132 float3 r; \ 133 r.x = fnc(v1.x, v2); \ 134 r.y = fnc(v1.y, v2); \ 135 r.z = fnc(v1.z, v2); \ 136 return r; \ 137 } \ 138 extern float4 __attribute__((overloadable)) fnc(float4 v1, int v2) { \ 139 float4 r; \ 140 r.x = fnc(v1.x, v2); \ 141 r.y = fnc(v1.y, v2); \ 142 r.z = fnc(v1.z, v2); \ 143 r.w = fnc(v1.w, v2); \ 144 return r; \ 145 } 146 147 #define FN_FUNC_FN_PFN(fnc) \ 148 extern float2 __attribute__((overloadable)) \ 149 fnc(float2 v1, float2 *v2) { \ 150 float2 r; \ 151 float t[2]; \ 152 r.x = fnc(v1.x, &t[0]); \ 153 r.y = fnc(v1.y, &t[1]); \ 154 v2->x = t[0]; \ 155 v2->y = t[1]; \ 156 return r; \ 157 } \ 158 extern float3 __attribute__((overloadable)) \ 159 fnc(float3 v1, float3 *v2) { \ 160 float3 r; \ 161 float t[3]; \ 162 r.x = fnc(v1.x, &t[0]); \ 163 r.y = fnc(v1.y, &t[1]); \ 164 r.z = fnc(v1.z, &t[2]); \ 165 v2->x = t[0]; \ 166 v2->y = t[1]; \ 167 v2->z = t[2]; \ 168 return r; \ 169 } \ 170 extern float4 __attribute__((overloadable)) \ 171 fnc(float4 v1, float4 *v2) { \ 172 float4 r; \ 173 float t[4]; \ 174 r.x = fnc(v1.x, &t[0]); \ 175 r.y = fnc(v1.y, &t[1]); \ 176 r.z = fnc(v1.z, &t[2]); \ 177 r.w = fnc(v1.w, &t[3]); \ 178 v2->x = t[0]; \ 179 v2->y = t[1]; \ 180 v2->z = t[2]; \ 181 v2->w = t[3]; \ 182 return r; \ 183 } 184 185 #define FN_FUNC_FN_PIN(fnc) \ 186 extern float2 __attribute__((overloadable)) fnc(float2 v1, int2 *v2) { \ 187 float2 r; \ 188 int t[2]; \ 189 r.x = fnc(v1.x, &t[0]); \ 190 r.y = fnc(v1.y, &t[1]); \ 191 v2->x = t[0]; \ 192 v2->y = t[1]; \ 193 return r; \ 194 } \ 195 extern float3 __attribute__((overloadable)) fnc(float3 v1, int3 *v2) { \ 196 float3 r; \ 197 int t[3]; \ 198 r.x = fnc(v1.x, &t[0]); \ 199 r.y = fnc(v1.y, &t[1]); \ 200 r.z = fnc(v1.z, &t[2]); \ 201 v2->x = t[0]; \ 202 v2->y = t[1]; \ 203 v2->z = t[2]; \ 204 return r; \ 205 } \ 206 extern float4 __attribute__((overloadable)) fnc(float4 v1, int4 *v2) { \ 207 float4 r; \ 208 int t[4]; \ 209 r.x = fnc(v1.x, &t[0]); \ 210 r.y = fnc(v1.y, &t[1]); \ 211 r.z = fnc(v1.z, &t[2]); \ 212 r.w = fnc(v1.w, &t[3]); \ 213 v2->x = t[0]; \ 214 v2->y = t[1]; \ 215 v2->z = t[2]; \ 216 v2->w = t[3]; \ 217 return r; \ 218 } 219 220 #define FN_FUNC_FN_FN_FN(fnc) \ 221 extern float2 __attribute__((overloadable)) \ 222 fnc(float2 v1, float2 v2, float2 v3) { \ 223 float2 r; \ 224 r.x = fnc(v1.x, v2.x, v3.x); \ 225 r.y = fnc(v1.y, v2.y, v3.y); \ 226 return r; \ 227 } \ 228 extern float3 __attribute__((overloadable)) \ 229 fnc(float3 v1, float3 v2, float3 v3) { \ 230 float3 r; \ 231 r.x = fnc(v1.x, v2.x, v3.x); \ 232 r.y = fnc(v1.y, v2.y, v3.y); \ 233 r.z = fnc(v1.z, v2.z, v3.z); \ 234 return r; \ 235 } \ 236 extern float4 __attribute__((overloadable)) \ 237 fnc(float4 v1, float4 v2, float4 v3) { \ 238 float4 r; \ 239 r.x = fnc(v1.x, v2.x, v3.x); \ 240 r.y = fnc(v1.y, v2.y, v3.y); \ 241 r.z = fnc(v1.z, v2.z, v3.z); \ 242 r.w = fnc(v1.w, v2.w, v3.w); \ 243 return r; \ 244 } 245 246 #define FN_FUNC_FN_FN_PIN(fnc) \ 247 extern float2 __attribute__((overloadable)) \ 248 fnc(float2 v1, float2 v2, int2 *v3) { \ 249 float2 r; \ 250 int t[2]; \ 251 r.x = fnc(v1.x, v2.x, &t[0]); \ 252 r.y = fnc(v1.y, v2.y, &t[1]); \ 253 v3->x = t[0]; \ 254 v3->y = t[1]; \ 255 return r; \ 256 } \ 257 extern float3 __attribute__((overloadable)) \ 258 fnc(float3 v1, float3 v2, int3 *v3) { \ 259 float3 r; \ 260 int t[3]; \ 261 r.x = fnc(v1.x, v2.x, &t[0]); \ 262 r.y = fnc(v1.y, v2.y, &t[1]); \ 263 r.z = fnc(v1.z, v2.z, &t[2]); \ 264 v3->x = t[0]; \ 265 v3->y = t[1]; \ 266 v3->z = t[2]; \ 267 return r; \ 268 } \ 269 extern float4 __attribute__((overloadable)) \ 270 fnc(float4 v1, float4 v2, int4 *v3) { \ 271 float4 r; \ 272 int t[4]; \ 273 r.x = fnc(v1.x, v2.x, &t[0]); \ 274 r.y = fnc(v1.y, v2.y, &t[1]); \ 275 r.z = fnc(v1.z, v2.z, &t[2]); \ 276 r.w = fnc(v1.w, v2.w, &t[3]); \ 277 v3->x = t[0]; \ 278 v3->y = t[1]; \ 279 v3->z = t[2]; \ 280 v3->w = t[3]; \ 281 return r; \ 282 } 283 284 static const int iposinf = 0x7f800000; 285 static const int ineginf = 0xff800000; 286 287 static const float posinf() { 288 float f = *((float*)&iposinf); 289 return f; 290 } 291 292 static const float neginf() { 293 float f = *((float*)&ineginf); 294 return f; 295 } 296 297 static bool isinf(float f) { 298 int i = *((int*)(void*)&f); 299 return (i == iposinf) || (i == ineginf); 300 } 301 302 static bool isnan(float f) { 303 int i = *((int*)(void*)&f); 304 return (((i & 0x7f800000) == 0x7f800000) && (i & 0x007fffff)); 305 } 306 307 static bool isposzero(float f) { 308 int i = *((int*)(void*)&f); 309 return (i == 0x00000000); 310 } 311 312 static bool isnegzero(float f) { 313 int i = *((int*)(void*)&f); 314 return (i == 0x80000000); 315 } 316 317 static bool iszero(float f) { 318 return isposzero(f) || isnegzero(f); 319 } 320 321 322 extern float __attribute__((overloadable)) acos(float); 323 FN_FUNC_FN(acos) 324 325 extern float __attribute__((overloadable)) acosh(float); 326 FN_FUNC_FN(acosh) 327 328 329 extern float __attribute__((overloadable)) acospi(float v) { 330 return acos(v) / M_PI; 331 } 332 FN_FUNC_FN(acospi) 333 334 extern float __attribute__((overloadable)) asin(float); 335 FN_FUNC_FN(asin) 336 337 extern float __attribute__((overloadable)) asinh(float); 338 FN_FUNC_FN(asinh) 339 340 extern float __attribute__((overloadable)) asinpi(float v) { 341 return asin(v) / M_PI; 342 } 343 FN_FUNC_FN(asinpi) 344 345 extern float __attribute__((overloadable)) atan(float); 346 FN_FUNC_FN(atan) 347 348 extern float __attribute__((overloadable)) atan2(float, float); 349 FN_FUNC_FN_FN(atan2) 350 351 extern float __attribute__((overloadable)) atanh(float); 352 FN_FUNC_FN(atanh) 353 354 extern float __attribute__((overloadable)) atanpi(float v) { 355 return atan(v) / M_PI; 356 } 357 FN_FUNC_FN(atanpi) 358 359 360 extern float __attribute__((overloadable)) atan2pi(float y, float x) { 361 return atan2(y, x) / M_PI; 362 } 363 FN_FUNC_FN_FN(atan2pi) 364 365 extern float __attribute__((overloadable)) cbrt(float); 366 FN_FUNC_FN(cbrt) 367 368 extern float __attribute__((overloadable)) ceil(float); 369 FN_FUNC_FN(ceil) 370 371 extern float __attribute__((overloadable)) copysign(float, float); 372 FN_FUNC_FN_FN(copysign) 373 374 extern float __attribute__((overloadable)) cos(float); 375 FN_FUNC_FN(cos) 376 377 extern float __attribute__((overloadable)) cosh(float); 378 FN_FUNC_FN(cosh) 379 380 extern float __attribute__((overloadable)) cospi(float v) { 381 return cos(v * M_PI); 382 } 383 FN_FUNC_FN(cospi) 384 385 extern float __attribute__((overloadable)) erfc(float); 386 FN_FUNC_FN(erfc) 387 388 extern float __attribute__((overloadable)) erf(float); 389 FN_FUNC_FN(erf) 390 391 extern float __attribute__((overloadable)) exp(float); 392 FN_FUNC_FN(exp) 393 394 extern float __attribute__((overloadable)) exp2(float); 395 FN_FUNC_FN(exp2) 396 397 extern float __attribute__((overloadable)) pow(float, float); 398 399 extern float __attribute__((overloadable)) exp10(float v) { 400 return pow(10.f, v); 401 } 402 FN_FUNC_FN(exp10) 403 404 extern float __attribute__((overloadable)) expm1(float); 405 FN_FUNC_FN(expm1) 406 407 extern float __attribute__((overloadable)) fabs(float); 408 FN_FUNC_FN(fabs) 409 410 extern float __attribute__((overloadable)) fdim(float, float); 411 FN_FUNC_FN_FN(fdim) 412 413 extern float __attribute__((overloadable)) floor(float); 414 FN_FUNC_FN(floor) 415 416 extern float __attribute__((overloadable)) fma(float, float, float); 417 FN_FUNC_FN_FN_FN(fma) 418 419 extern float __attribute__((overloadable)) fmin(float, float); 420 421 extern float __attribute__((overloadable)) fmod(float, float); 422 FN_FUNC_FN_FN(fmod) 423 424 extern float __attribute__((overloadable)) fract(float v, float *iptr) { 425 int i = (int)floor(v); 426 iptr[0] = i; 427 return fmin(v - i, 0x1.fffffep-1f); 428 } 429 FN_FUNC_FN_PFN(fract) 430 431 extern float __attribute__((overloadable)) frexp(float, int *); 432 FN_FUNC_FN_PIN(frexp) 433 434 extern float __attribute__((overloadable)) hypot(float, float); 435 FN_FUNC_FN_FN(hypot) 436 437 extern int __attribute__((overloadable)) ilogb(float); 438 IN_FUNC_FN(ilogb) 439 440 extern float __attribute__((overloadable)) ldexp(float, int); 441 FN_FUNC_FN_IN(ldexp) 442 FN_FUNC_FN_I(ldexp) 443 444 extern float __attribute__((overloadable)) lgamma(float); 445 FN_FUNC_FN(lgamma) 446 extern float __attribute__((overloadable)) lgamma(float, int*); 447 FN_FUNC_FN_PIN(lgamma) 448 449 extern float __attribute__((overloadable)) log(float); 450 FN_FUNC_FN(log) 451 452 extern float __attribute__((overloadable)) log10(float); 453 FN_FUNC_FN(log10) 454 455 456 extern float __attribute__((overloadable)) log2(float v) { 457 return log10(v) / log10(2.f); 458 } 459 FN_FUNC_FN(log2) 460 461 extern float __attribute__((overloadable)) log1p(float); 462 FN_FUNC_FN(log1p) 463 464 extern float __attribute__((overloadable)) logb(float); 465 FN_FUNC_FN(logb) 466 467 extern float __attribute__((overloadable)) mad(float a, float b, float c) { 468 return a * b + c; 469 } 470 extern float2 __attribute__((overloadable)) mad(float2 a, float2 b, float2 c) { 471 return a * b + c; 472 } 473 extern float3 __attribute__((overloadable)) mad(float3 a, float3 b, float3 c) { 474 return a * b + c; 475 } 476 extern float4 __attribute__((overloadable)) mad(float4 a, float4 b, float4 c) { 477 return a * b + c; 478 } 479 480 extern float __attribute__((overloadable)) modf(float, float *); 481 FN_FUNC_FN_PFN(modf); 482 483 extern float __attribute__((overloadable)) nan(uint v) { 484 float f[1]; 485 uint32_t *ip = (uint32_t *)f; 486 *ip = v | 0x7fc00000; 487 return f[0]; 488 } 489 490 extern float __attribute__((overloadable)) nextafter(float, float); 491 FN_FUNC_FN_FN(nextafter) 492 493 FN_FUNC_FN_FN(pow) 494 495 extern float __attribute__((overloadable)) pown(float v, int p) { 496 return pow(v, (float)p); 497 } 498 extern float2 __attribute__((overloadable)) pown(float2 v, int2 p) { 499 float2 f2 = convert_float2(p); 500 return pow(v, f2); 501 } 502 extern float3 __attribute__((overloadable)) pown(float3 v, int3 p) { 503 float3 f3 = convert_float3(p); 504 return pow(v, f3); 505 } 506 extern float4 __attribute__((overloadable)) pown(float4 v, int4 p) { 507 float4 f4 = convert_float4(p); 508 return pow(v, f4); 509 } 510 511 extern float __attribute__((overloadable)) powr(float v, float p) { 512 return pow(v, p); 513 } 514 extern float2 __attribute__((overloadable)) powr(float2 v, float2 p) { 515 return pow(v, p); 516 } 517 extern float3 __attribute__((overloadable)) powr(float3 v, float3 p) { 518 return pow(v, p); 519 } 520 extern float4 __attribute__((overloadable)) powr(float4 v, float4 p) { 521 return pow(v, p); 522 } 523 524 extern float __attribute__((overloadable)) remainder(float, float); 525 FN_FUNC_FN_FN(remainder) 526 527 extern float __attribute__((overloadable)) remquo(float, float, int *); 528 FN_FUNC_FN_FN_PIN(remquo) 529 530 extern float __attribute__((overloadable)) rint(float); 531 FN_FUNC_FN(rint) 532 533 extern float __attribute__((overloadable)) rootn(float v, int r) { 534 if (r == 0) { 535 return nan(0); 536 } 537 538 if (iszero(v)) { 539 if (r < 0) { 540 if (r & 1) { 541 return copysign(posinf(), v); 542 } else { 543 return posinf(); 544 } 545 } else { 546 if (r & 1) { 547 return copysign(0.f, v); 548 } else { 549 return 0.f; 550 } 551 } 552 } 553 554 if (!isinf(v) && !isnan(v) && (v < 0.f)) { 555 if (r & 1) { 556 return (-1.f * pow(-1.f * v, 1.f / r)); 557 } else { 558 return nan(0); 559 } 560 } 561 562 return pow(v, 1.f / r); 563 } 564 FN_FUNC_FN_IN(rootn); 565 566 extern float __attribute__((overloadable)) round(float); 567 FN_FUNC_FN(round) 568 569 570 extern float __attribute__((overloadable)) sqrt(float); 571 extern float __attribute__((overloadable)) rsqrt(float v) { 572 return 1.f / sqrt(v); 573 } 574 FN_FUNC_FN(rsqrt) 575 576 extern float __attribute__((overloadable)) sin(float); 577 FN_FUNC_FN(sin) 578 579 extern float __attribute__((overloadable)) sincos(float v, float *cosptr) { 580 *cosptr = cos(v); 581 return sin(v); 582 } 583 extern float2 __attribute__((overloadable)) sincos(float2 v, float2 *cosptr) { 584 *cosptr = cos(v); 585 return sin(v); 586 } 587 extern float3 __attribute__((overloadable)) sincos(float3 v, float3 *cosptr) { 588 *cosptr = cos(v); 589 return sin(v); 590 } 591 extern float4 __attribute__((overloadable)) sincos(float4 v, float4 *cosptr) { 592 *cosptr = cos(v); 593 return sin(v); 594 } 595 596 extern float __attribute__((overloadable)) sinh(float); 597 FN_FUNC_FN(sinh) 598 599 extern float __attribute__((overloadable)) sinpi(float v) { 600 return sin(v * M_PI); 601 } 602 FN_FUNC_FN(sinpi) 603 604 FN_FUNC_FN(sqrt) 605 606 extern float __attribute__((overloadable)) tan(float); 607 FN_FUNC_FN(tan) 608 609 extern float __attribute__((overloadable)) tanh(float); 610 FN_FUNC_FN(tanh) 611 612 extern float __attribute__((overloadable)) tanpi(float v) { 613 return tan(v * M_PI); 614 } 615 FN_FUNC_FN(tanpi) 616 617 618 extern float __attribute__((overloadable)) tgamma(float); 619 FN_FUNC_FN(tgamma) 620 621 extern float __attribute__((overloadable)) trunc(float); 622 FN_FUNC_FN(trunc) 623 624 // Int ops (partial), 6.11.3 625 626 #define XN_FUNC_YN(typeout, fnc, typein) \ 627 extern typeout __attribute__((overloadable)) fnc(typein); \ 628 extern typeout##2 __attribute__((overloadable)) fnc(typein##2 v) { \ 629 typeout##2 r; \ 630 r.x = fnc(v.x); \ 631 r.y = fnc(v.y); \ 632 return r; \ 633 } \ 634 extern typeout##3 __attribute__((overloadable)) fnc(typein##3 v) { \ 635 typeout##3 r; \ 636 r.x = fnc(v.x); \ 637 r.y = fnc(v.y); \ 638 r.z = fnc(v.z); \ 639 return r; \ 640 } \ 641 extern typeout##4 __attribute__((overloadable)) fnc(typein##4 v) { \ 642 typeout##4 r; \ 643 r.x = fnc(v.x); \ 644 r.y = fnc(v.y); \ 645 r.z = fnc(v.z); \ 646 r.w = fnc(v.w); \ 647 return r; \ 648 } 649 650 651 #define UIN_FUNC_IN(fnc) \ 652 XN_FUNC_YN(uchar, fnc, char) \ 653 XN_FUNC_YN(ushort, fnc, short) \ 654 XN_FUNC_YN(uint, fnc, int) 655 656 #define IN_FUNC_IN(fnc) \ 657 XN_FUNC_YN(uchar, fnc, uchar) \ 658 XN_FUNC_YN(char, fnc, char) \ 659 XN_FUNC_YN(ushort, fnc, ushort) \ 660 XN_FUNC_YN(short, fnc, short) \ 661 XN_FUNC_YN(uint, fnc, uint) \ 662 XN_FUNC_YN(int, fnc, int) 663 664 665 #define XN_FUNC_XN_XN_BODY(type, fnc, body) \ 666 extern type __attribute__((overloadable)) \ 667 fnc(type v1, type v2) { \ 668 return body; \ 669 } \ 670 extern type##2 __attribute__((overloadable)) \ 671 fnc(type##2 v1, type##2 v2) { \ 672 type##2 r; \ 673 r.x = fnc(v1.x, v2.x); \ 674 r.y = fnc(v1.y, v2.y); \ 675 return r; \ 676 } \ 677 extern type##3 __attribute__((overloadable)) \ 678 fnc(type##3 v1, type##3 v2) { \ 679 type##3 r; \ 680 r.x = fnc(v1.x, v2.x); \ 681 r.y = fnc(v1.y, v2.y); \ 682 r.z = fnc(v1.z, v2.z); \ 683 return r; \ 684 } \ 685 extern type##4 __attribute__((overloadable)) \ 686 fnc(type##4 v1, type##4 v2) { \ 687 type##4 r; \ 688 r.x = fnc(v1.x, v2.x); \ 689 r.y = fnc(v1.y, v2.y); \ 690 r.z = fnc(v1.z, v2.z); \ 691 r.w = fnc(v1.w, v2.w); \ 692 return r; \ 693 } 694 695 #define IN_FUNC_IN_IN_BODY(fnc, body) \ 696 XN_FUNC_XN_XN_BODY(uchar, fnc, body) \ 697 XN_FUNC_XN_XN_BODY(char, fnc, body) \ 698 XN_FUNC_XN_XN_BODY(ushort, fnc, body) \ 699 XN_FUNC_XN_XN_BODY(short, fnc, body) \ 700 XN_FUNC_XN_XN_BODY(uint, fnc, body) \ 701 XN_FUNC_XN_XN_BODY(int, fnc, body) \ 702 XN_FUNC_XN_XN_BODY(float, fnc, body) 703 704 UIN_FUNC_IN(abs) 705 IN_FUNC_IN(clz) 706 707 708 // 6.11.4 709 710 711 extern float __attribute__((overloadable)) degrees(float radians) { 712 return radians * (180.f / M_PI); 713 } 714 extern float2 __attribute__((overloadable)) degrees(float2 radians) { 715 return radians * (180.f / M_PI); 716 } 717 extern float3 __attribute__((overloadable)) degrees(float3 radians) { 718 return radians * (180.f / M_PI); 719 } 720 extern float4 __attribute__((overloadable)) degrees(float4 radians) { 721 return radians * (180.f / M_PI); 722 } 723 724 extern float __attribute__((overloadable)) mix(float start, float stop, float amount) { 725 return start + (stop - start) * amount; 726 } 727 extern float2 __attribute__((overloadable)) mix(float2 start, float2 stop, float2 amount) { 728 return start + (stop - start) * amount; 729 } 730 extern float3 __attribute__((overloadable)) mix(float3 start, float3 stop, float3 amount) { 731 return start + (stop - start) * amount; 732 } 733 extern float4 __attribute__((overloadable)) mix(float4 start, float4 stop, float4 amount) { 734 return start + (stop - start) * amount; 735 } 736 extern float2 __attribute__((overloadable)) mix(float2 start, float2 stop, float amount) { 737 return start + (stop - start) * amount; 738 } 739 extern float3 __attribute__((overloadable)) mix(float3 start, float3 stop, float amount) { 740 return start + (stop - start) * amount; 741 } 742 extern float4 __attribute__((overloadable)) mix(float4 start, float4 stop, float amount) { 743 return start + (stop - start) * amount; 744 } 745 746 extern float __attribute__((overloadable)) radians(float degrees) { 747 return degrees * (M_PI / 180.f); 748 } 749 extern float2 __attribute__((overloadable)) radians(float2 degrees) { 750 return degrees * (M_PI / 180.f); 751 } 752 extern float3 __attribute__((overloadable)) radians(float3 degrees) { 753 return degrees * (M_PI / 180.f); 754 } 755 extern float4 __attribute__((overloadable)) radians(float4 degrees) { 756 return degrees * (M_PI / 180.f); 757 } 758 759 extern float __attribute__((overloadable)) step(float edge, float v) { 760 return (v < edge) ? 0.f : 1.f; 761 } 762 extern float2 __attribute__((overloadable)) step(float2 edge, float2 v) { 763 float2 r; 764 r.x = (v.x < edge.x) ? 0.f : 1.f; 765 r.y = (v.y < edge.y) ? 0.f : 1.f; 766 return r; 767 } 768 extern float3 __attribute__((overloadable)) step(float3 edge, float3 v) { 769 float3 r; 770 r.x = (v.x < edge.x) ? 0.f : 1.f; 771 r.y = (v.y < edge.y) ? 0.f : 1.f; 772 r.z = (v.z < edge.z) ? 0.f : 1.f; 773 return r; 774 } 775 extern float4 __attribute__((overloadable)) step(float4 edge, float4 v) { 776 float4 r; 777 r.x = (v.x < edge.x) ? 0.f : 1.f; 778 r.y = (v.y < edge.y) ? 0.f : 1.f; 779 r.z = (v.z < edge.z) ? 0.f : 1.f; 780 r.w = (v.w < edge.w) ? 0.f : 1.f; 781 return r; 782 } 783 extern float2 __attribute__((overloadable)) step(float2 edge, float v) { 784 float2 r; 785 r.x = (v < edge.x) ? 0.f : 1.f; 786 r.y = (v < edge.y) ? 0.f : 1.f; 787 return r; 788 } 789 extern float3 __attribute__((overloadable)) step(float3 edge, float v) { 790 float3 r; 791 r.x = (v < edge.x) ? 0.f : 1.f; 792 r.y = (v < edge.y) ? 0.f : 1.f; 793 r.z = (v < edge.z) ? 0.f : 1.f; 794 return r; 795 } 796 extern float4 __attribute__((overloadable)) step(float4 edge, float v) { 797 float4 r; 798 r.x = (v < edge.x) ? 0.f : 1.f; 799 r.y = (v < edge.y) ? 0.f : 1.f; 800 r.z = (v < edge.z) ? 0.f : 1.f; 801 r.w = (v < edge.w) ? 0.f : 1.f; 802 return r; 803 } 804 805 extern float __attribute__((overloadable)) smoothstep(float, float, float); 806 extern float2 __attribute__((overloadable)) smoothstep(float2, float2, float2); 807 extern float3 __attribute__((overloadable)) smoothstep(float3, float3, float3); 808 extern float4 __attribute__((overloadable)) smoothstep(float4, float4, float4); 809 extern float2 __attribute__((overloadable)) smoothstep(float, float, float2); 810 extern float3 __attribute__((overloadable)) smoothstep(float, float, float3); 811 extern float4 __attribute__((overloadable)) smoothstep(float, float, float4); 812 813 extern float __attribute__((overloadable)) sign(float v) { 814 if (v > 0) return 1.f; 815 if (v < 0) return -1.f; 816 return v; 817 } 818 FN_FUNC_FN(sign) 819 820 821 // 6.11.5 822 extern float3 __attribute__((overloadable)) cross(float3 lhs, float3 rhs) { 823 float3 r; 824 r.x = lhs.y * rhs.z - lhs.z * rhs.y; 825 r.y = lhs.z * rhs.x - lhs.x * rhs.z; 826 r.z = lhs.x * rhs.y - lhs.y * rhs.x; 827 return r; 828 } 829 830 extern float4 __attribute__((overloadable)) cross(float4 lhs, float4 rhs) { 831 float4 r; 832 r.x = lhs.y * rhs.z - lhs.z * rhs.y; 833 r.y = lhs.z * rhs.x - lhs.x * rhs.z; 834 r.z = lhs.x * rhs.y - lhs.y * rhs.x; 835 r.w = 0.f; 836 return r; 837 } 838 839 extern float __attribute__((overloadable)) dot(float lhs, float rhs) { 840 return lhs * rhs; 841 } 842 extern float __attribute__((overloadable)) dot(float2 lhs, float2 rhs) { 843 return lhs.x*rhs.x + lhs.y*rhs.y; 844 } 845 extern float __attribute__((overloadable)) dot(float3 lhs, float3 rhs) { 846 return lhs.x*rhs.x + lhs.y*rhs.y + lhs.z*rhs.z; 847 } 848 extern float __attribute__((overloadable)) dot(float4 lhs, float4 rhs) { 849 return lhs.x*rhs.x + lhs.y*rhs.y + lhs.z*rhs.z + lhs.w*rhs.w; 850 } 851 852 extern float __attribute__((overloadable)) length(float v) { 853 return fabs(v); 854 } 855 extern float __attribute__((overloadable)) length(float2 v) { 856 return sqrt(v.x*v.x + v.y*v.y); 857 } 858 extern float __attribute__((overloadable)) length(float3 v) { 859 return sqrt(v.x*v.x + v.y*v.y + v.z*v.z); 860 } 861 extern float __attribute__((overloadable)) length(float4 v) { 862 return sqrt(v.x*v.x + v.y*v.y + v.z*v.z + v.w*v.w); 863 } 864 865 extern float __attribute__((overloadable)) distance(float lhs, float rhs) { 866 return length(lhs - rhs); 867 } 868 extern float __attribute__((overloadable)) distance(float2 lhs, float2 rhs) { 869 return length(lhs - rhs); 870 } 871 extern float __attribute__((overloadable)) distance(float3 lhs, float3 rhs) { 872 return length(lhs - rhs); 873 } 874 extern float __attribute__((overloadable)) distance(float4 lhs, float4 rhs) { 875 return length(lhs - rhs); 876 } 877 878 extern float __attribute__((overloadable)) normalize(float v) { 879 return 1.f; 880 } 881 extern float2 __attribute__((overloadable)) normalize(float2 v) { 882 return v / length(v); 883 } 884 extern float3 __attribute__((overloadable)) normalize(float3 v) { 885 return v / length(v); 886 } 887 extern float4 __attribute__((overloadable)) normalize(float4 v) { 888 return v / length(v); 889 } 890 891 extern float __attribute__((overloadable)) half_sqrt(float); 892 893 extern float __attribute__((overloadable)) fast_length(float v) { 894 return v; 895 } 896 extern float __attribute__((overloadable)) fast_length(float2 v) { 897 return half_sqrt(v.x*v.x + v.y*v.y); 898 } 899 extern float __attribute__((overloadable)) fast_length(float3 v) { 900 return half_sqrt(v.x*v.x + v.y*v.y + v.z*v.z); 901 } 902 extern float __attribute__((overloadable)) fast_length(float4 v) { 903 return half_sqrt(v.x*v.x + v.y*v.y + v.z*v.z + v.w*v.w); 904 } 905 906 extern float __attribute__((overloadable)) fast_distance(float lhs, float rhs) { 907 return fast_length(lhs - rhs); 908 } 909 extern float __attribute__((overloadable)) fast_distance(float2 lhs, float2 rhs) { 910 return fast_length(lhs - rhs); 911 } 912 extern float __attribute__((overloadable)) fast_distance(float3 lhs, float3 rhs) { 913 return fast_length(lhs - rhs); 914 } 915 extern float __attribute__((overloadable)) fast_distance(float4 lhs, float4 rhs) { 916 return fast_length(lhs - rhs); 917 } 918 919 extern float __attribute__((overloadable)) half_rsqrt(float); 920 921 extern float __attribute__((overloadable)) fast_normalize(float v) { 922 return 1.f; 923 } 924 extern float2 __attribute__((overloadable)) fast_normalize(float2 v) { 925 return v * half_rsqrt(v.x*v.x + v.y*v.y); 926 } 927 extern float3 __attribute__((overloadable)) fast_normalize(float3 v) { 928 return v * half_rsqrt(v.x*v.x + v.y*v.y + v.z*v.z); 929 } 930 extern float4 __attribute__((overloadable)) fast_normalize(float4 v) { 931 return v * half_rsqrt(v.x*v.x + v.y*v.y + v.z*v.z + v.w*v.w); 932 } 933 934 extern float __attribute__((overloadable)) half_recip(float); 935 936 /* 937 extern float __attribute__((overloadable)) approx_atan(float x) { 938 if (x == 0.f) 939 return 0.f; 940 if (x < 0.f) 941 return -1.f * approx_atan(-1.f * x); 942 if (x > 1.f) 943 return M_PI_2 - approx_atan(approx_recip(x)); 944 return x * approx_recip(1.f + 0.28f * x*x); 945 } 946 FN_FUNC_FN(approx_atan) 947 */ 948 949 #undef FN_FUNC_FN 950 #undef IN_FUNC_FN 951 #undef FN_FUNC_FN_FN 952 #undef FN_FUNC_FN_F 953 #undef FN_FUNC_FN_IN 954 #undef FN_FUNC_FN_I 955 #undef FN_FUNC_FN_PFN 956 #undef FN_FUNC_FN_PIN 957 #undef FN_FUNC_FN_FN_FN 958 #undef FN_FUNC_FN_FN_PIN 959 #undef XN_FUNC_YN 960 #undef UIN_FUNC_IN 961 #undef IN_FUNC_IN 962 #undef XN_FUNC_XN_XN_BODY 963 #undef IN_FUNC_IN_IN_BODY 964