1 /* 2 * Copyright (C) 2011-2012 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17 /** @file rs_cl.rsh 18 * \brief Basic math functions 19 * 20 * 21 */ 22 23 #ifndef __RS_CL_RSH__ 24 #define __RS_CL_RSH__ 25 26 // Conversions 27 #define CVT_FUNC_2(typeout, typein) \ 28 _RS_RUNTIME typeout##2 __attribute__((overloadable)) \ 29 convert_##typeout##2(typein##2 v); \ 30 _RS_RUNTIME typeout##3 __attribute__((overloadable)) \ 31 convert_##typeout##3(typein##3 v); \ 32 _RS_RUNTIME typeout##4 __attribute__((overloadable)) \ 33 convert_##typeout##4(typein##4 v); 34 35 36 #define CVT_FUNC(type) CVT_FUNC_2(type, uchar) \ 37 CVT_FUNC_2(type, char) \ 38 CVT_FUNC_2(type, ushort) \ 39 CVT_FUNC_2(type, short) \ 40 CVT_FUNC_2(type, uint) \ 41 CVT_FUNC_2(type, int) \ 42 CVT_FUNC_2(type, float) 43 44 /** 45 * Convert to char. 46 * 47 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 48 */ 49 CVT_FUNC(char) 50 51 /** 52 * Convert to unsigned char. 53 * 54 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 55 */ 56 CVT_FUNC(uchar) 57 58 /** 59 * Convert to short. 60 * 61 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 62 */ 63 CVT_FUNC(short) 64 65 /** 66 * Convert to unsigned short. 67 * 68 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 69 */ 70 CVT_FUNC(ushort) 71 72 /** 73 * Convert to int. 74 * 75 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 76 */ 77 CVT_FUNC(int) 78 79 /** 80 * Convert to unsigned int. 81 * 82 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 83 */ 84 CVT_FUNC(uint) 85 86 /** 87 * Convert to float. 88 * 89 * Supports 2,3,4 components of uchar, char, ushort, short, uint, int, float. 90 */ 91 CVT_FUNC(float) 92 93 // Float ops, 6.11.2 94 95 #define FN_FUNC_FN(fnc) \ 96 _RS_RUNTIME float2 __attribute__((overloadable)) fnc(float2 v); \ 97 _RS_RUNTIME float3 __attribute__((overloadable)) fnc(float3 v); \ 98 _RS_RUNTIME float4 __attribute__((overloadable)) fnc(float4 v); 99 100 #define F_FUNC_FN(fnc) \ 101 _RS_RUNTIME float __attribute__((overloadable)) fnc(float2 v); \ 102 _RS_RUNTIME float __attribute__((overloadable)) fnc(float3 v); \ 103 _RS_RUNTIME float __attribute__((overloadable)) fnc(float4 v); 104 105 #define IN_FUNC_FN(fnc) \ 106 _RS_RUNTIME int2 __attribute__((overloadable)) fnc(float2 v); \ 107 _RS_RUNTIME int3 __attribute__((overloadable)) fnc(float3 v); \ 108 _RS_RUNTIME int4 __attribute__((overloadable)) fnc(float4 v); 109 110 #define FN_FUNC_FN_FN(fnc) \ 111 _RS_RUNTIME float2 __attribute__((overloadable)) fnc(float2 v1, float2 v2); \ 112 _RS_RUNTIME float3 __attribute__((overloadable)) fnc(float3 v1, float3 v2); \ 113 _RS_RUNTIME float4 __attribute__((overloadable)) fnc(float4 v1, float4 v2); 114 115 #define F_FUNC_FN_FN(fnc) \ 116 _RS_RUNTIME float __attribute__((overloadable)) fnc(float2 v1, float2 v2); \ 117 _RS_RUNTIME float __attribute__((overloadable)) fnc(float3 v1, float3 v2); \ 118 _RS_RUNTIME float __attribute__((overloadable)) fnc(float4 v1, float4 v2); 119 120 #define FN_FUNC_FN_F(fnc) \ 121 _RS_RUNTIME float2 __attribute__((overloadable)) fnc(float2 v1, float v2); \ 122 _RS_RUNTIME float3 __attribute__((overloadable)) fnc(float3 v1, float v2); \ 123 _RS_RUNTIME float4 __attribute__((overloadable)) fnc(float4 v1, float v2); 124 125 #define FN_FUNC_FN_IN(fnc) \ 126 _RS_RUNTIME float2 __attribute__((overloadable)) fnc(float2 v1, int2 v2); \ 127 _RS_RUNTIME float3 __attribute__((overloadable)) fnc(float3 v1, int3 v2); \ 128 _RS_RUNTIME float4 __attribute__((overloadable)) fnc(float4 v1, int4 v2); \ 129 130 #define FN_FUNC_FN_I(fnc) \ 131 _RS_RUNTIME float2 __attribute__((overloadable)) fnc(float2 v1, int v2); \ 132 _RS_RUNTIME float3 __attribute__((overloadable)) fnc(float3 v1, int v2); \ 133 _RS_RUNTIME float4 __attribute__((overloadable)) fnc(float4 v1, int v2); 134 135 #define FN_FUNC_FN_PFN(fnc) \ 136 _RS_RUNTIME float2 __attribute__((overloadable)) \ 137 fnc(float2 v1, float2 *v2); \ 138 _RS_RUNTIME float3 __attribute__((overloadable)) \ 139 fnc(float3 v1, float3 *v2); \ 140 _RS_RUNTIME float4 __attribute__((overloadable)) \ 141 fnc(float4 v1, float4 *v2); 142 143 #define FN_FUNC_FN_PIN(fnc) \ 144 _RS_RUNTIME float2 __attribute__((overloadable)) fnc(float2 v1, int2 *v2); \ 145 _RS_RUNTIME float3 __attribute__((overloadable)) fnc(float3 v1, int3 *v2); \ 146 _RS_RUNTIME float4 __attribute__((overloadable)) fnc(float4 v1, int4 *v2); 147 148 #define FN_FUNC_FN_FN_FN(fnc) \ 149 _RS_RUNTIME float2 __attribute__((overloadable)) \ 150 fnc(float2 v1, float2 v2, float2 v3); \ 151 _RS_RUNTIME float3 __attribute__((overloadable)) \ 152 fnc(float3 v1, float3 v2, float3 v3); \ 153 _RS_RUNTIME float4 __attribute__((overloadable)) \ 154 fnc(float4 v1, float4 v2, float4 v3); 155 156 #define FN_FUNC_FN_FN_F(fnc) \ 157 _RS_RUNTIME float2 __attribute__((overloadable)) \ 158 fnc(float2 v1, float2 v2, float v3); \ 159 _RS_RUNTIME float3 __attribute__((overloadable)) \ 160 fnc(float3 v1, float3 v2, float v3); \ 161 _RS_RUNTIME float4 __attribute__((overloadable)) \ 162 fnc(float4 v1, float4 v2, float v3); 163 164 #define FN_FUNC_FN_F_F(fnc) \ 165 _RS_RUNTIME float2 __attribute__((overloadable)) \ 166 fnc(float2 v1, float v2, float v3); \ 167 _RS_RUNTIME float3 __attribute__((overloadable)) \ 168 fnc(float3 v1, float v2, float v3); \ 169 _RS_RUNTIME float4 __attribute__((overloadable)) \ 170 fnc(float4 v1, float v2, float v3); 171 172 #define FN_FUNC_FN_FN_PIN(fnc) \ 173 _RS_RUNTIME float2 __attribute__((overloadable)) \ 174 fnc(float2 v1, float2 v2, int2 *v3); \ 175 _RS_RUNTIME float3 __attribute__((overloadable)) \ 176 fnc(float3 v1, float3 v2, int3 *v3); \ 177 _RS_RUNTIME float4 __attribute__((overloadable)) \ 178 fnc(float4 v1, float4 v2, int4 *v3); 179 180 181 /** 182 * Return the inverse cosine. 183 * 184 * Supports float, float2, float3, float4 185 */ 186 extern float __attribute__((overloadable)) acos(float); 187 FN_FUNC_FN(acos) 188 189 /** 190 * Return the inverse hyperbolic cosine. 191 * 192 * Supports float, float2, float3, float4 193 */ 194 extern float __attribute__((overloadable)) acosh(float); 195 FN_FUNC_FN(acosh) 196 197 /** 198 * Return the inverse cosine divided by PI. 199 * 200 * Supports float, float2, float3, float4 201 */ 202 _RS_RUNTIME float __attribute__((overloadable)) acospi(float v); 203 FN_FUNC_FN(acospi) 204 205 /** 206 * Return the inverse sine. 207 * 208 * Supports float, float2, float3, float4 209 */ 210 extern float __attribute__((overloadable)) asin(float); 211 FN_FUNC_FN(asin) 212 213 /** 214 * Return the inverse hyperbolic sine. 215 * 216 * Supports float, float2, float3, float4 217 */ 218 extern float __attribute__((overloadable)) asinh(float); 219 FN_FUNC_FN(asinh) 220 221 222 /** 223 * Return the inverse sine divided by PI. 224 * 225 * Supports float, float2, float3, float4 226 */ 227 _RS_RUNTIME float __attribute__((overloadable)) asinpi(float v); 228 FN_FUNC_FN(asinpi) 229 230 /** 231 * Return the inverse tangent. 232 * 233 * Supports float, float2, float3, float4 234 */ 235 extern float __attribute__((overloadable)) atan(float); 236 FN_FUNC_FN(atan) 237 238 /** 239 * Return the inverse tangent of y / x. 240 * 241 * Supports float, float2, float3, float4. Both arguments must be of the same 242 * type. 243 * 244 * @param y 245 * @param x 246 */ 247 extern float __attribute__((overloadable)) atan2(float y, float x); 248 FN_FUNC_FN_FN(atan2) 249 250 /** 251 * Return the inverse hyperbolic tangent. 252 * 253 * Supports float, float2, float3, float4 254 */ 255 extern float __attribute__((overloadable)) atanh(float); 256 FN_FUNC_FN(atanh) 257 258 /** 259 * Return the inverse tangent divided by PI. 260 * 261 * Supports float, float2, float3, float4 262 */ 263 _RS_RUNTIME float __attribute__((overloadable)) atanpi(float v); 264 FN_FUNC_FN(atanpi) 265 266 /** 267 * Return the inverse tangent of y / x, divided by PI. 268 * 269 * Supports float, float2, float3, float4. Both arguments must be of the same 270 * type. 271 * 272 * @param y 273 * @param x 274 */ 275 _RS_RUNTIME float __attribute__((overloadable)) atan2pi(float y, float x); 276 FN_FUNC_FN_FN(atan2pi) 277 278 279 /** 280 * Return the cube root. 281 * 282 * Supports float, float2, float3, float4. 283 */ 284 extern float __attribute__((overloadable)) cbrt(float); 285 FN_FUNC_FN(cbrt) 286 287 /** 288 * Return the smallest integer not less than a value. 289 * 290 * Supports float, float2, float3, float4. 291 */ 292 extern float __attribute__((overloadable)) ceil(float); 293 FN_FUNC_FN(ceil) 294 295 /** 296 * Copy the sign bit from y to x. 297 * 298 * Supports float, float2, float3, float4. Both arguments must be of the same 299 * type. 300 * 301 * @param x 302 * @param y 303 */ 304 extern float __attribute__((overloadable)) copysign(float x, float y); 305 FN_FUNC_FN_FN(copysign) 306 307 /** 308 * Return the cosine. 309 * 310 * Supports float, float2, float3, float4. 311 */ 312 extern float __attribute__((overloadable)) cos(float); 313 FN_FUNC_FN(cos) 314 315 /** 316 * Return the hypebolic cosine. 317 * 318 * Supports float, float2, float3, float4. 319 */ 320 extern float __attribute__((overloadable)) cosh(float); 321 FN_FUNC_FN(cosh) 322 323 /** 324 * Return the cosine of the value * PI. 325 * 326 * Supports float, float2, float3, float4. 327 */ 328 _RS_RUNTIME float __attribute__((overloadable)) cospi(float v); 329 FN_FUNC_FN(cospi) 330 331 /** 332 * Return the complementary error function. 333 * 334 * Supports float, float2, float3, float4. 335 */ 336 extern float __attribute__((overloadable)) erfc(float); 337 FN_FUNC_FN(erfc) 338 339 /** 340 * Return the error function. 341 * 342 * Supports float, float2, float3, float4. 343 */ 344 extern float __attribute__((overloadable)) erf(float); 345 FN_FUNC_FN(erf) 346 347 /** 348 * Return e ^ value. 349 * 350 * Supports float, float2, float3, float4. 351 */ 352 extern float __attribute__((overloadable)) exp(float); 353 FN_FUNC_FN(exp) 354 355 /** 356 * Return 2 ^ value. 357 * 358 * Supports float, float2, float3, float4. 359 */ 360 extern float __attribute__((overloadable)) exp2(float); 361 FN_FUNC_FN(exp2) 362 363 /** 364 * Return x ^ y. 365 * 366 * Supports float, float2, float3, float4. Both arguments must be of the same 367 * type. 368 */ 369 extern float __attribute__((overloadable)) pow(float x, float y); 370 FN_FUNC_FN_FN(pow) 371 372 /** 373 * Return 10 ^ value. 374 * 375 * Supports float, float2, float3, float4. 376 */ 377 _RS_RUNTIME float __attribute__((overloadable)) exp10(float v); 378 FN_FUNC_FN(exp10) 379 380 /** 381 * Return (e ^ value) - 1. 382 * 383 * Supports float, float2, float3, float4. 384 */ 385 extern float __attribute__((overloadable)) expm1(float); 386 FN_FUNC_FN(expm1) 387 388 /** 389 * Return the absolute value of a value. 390 * 391 * Supports float, float2, float3, float4. 392 */ 393 extern float __attribute__((overloadable)) fabs(float); 394 FN_FUNC_FN(fabs) 395 396 /** 397 * Return the positive difference between two values. 398 * 399 * Supports float, float2, float3, float4. Both arguments must be of the same 400 * type. 401 */ 402 extern float __attribute__((overloadable)) fdim(float, float); 403 FN_FUNC_FN_FN(fdim) 404 405 /** 406 * Return the smallest integer not greater than a value. 407 * 408 * Supports float, float2, float3, float4. 409 */ 410 extern float __attribute__((overloadable)) floor(float); 411 FN_FUNC_FN(floor) 412 413 /** 414 * Return a*b + c. 415 * 416 * Supports float, float2, float3, float4. 417 */ 418 extern float __attribute__((overloadable)) fma(float a, float b, float c); 419 FN_FUNC_FN_FN_FN(fma) 420 421 /** 422 * Return (x < y ? y : x) 423 * 424 * Supports float, float2, float3, float4. 425 * @param x: may be float, float2, float3, float4 426 * @param y: may be float or vector. If vector must match type of x. 427 */ 428 extern float __attribute__((overloadable)) fmax(float x, float y); 429 FN_FUNC_FN_FN(fmax); 430 FN_FUNC_FN_F(fmax); 431 432 /** 433 * Return (x > y ? y : x) 434 * 435 * @param x: may be float, float2, float3, float4 436 * @param y: may be float or vector. If vector must match type of x. 437 */ 438 extern float __attribute__((overloadable)) fmin(float x, float y); 439 FN_FUNC_FN_FN(fmin); 440 FN_FUNC_FN_F(fmin); 441 442 /** 443 * Return the remainder from x / y 444 * 445 * Supports float, float2, float3, float4. 446 */ 447 extern float __attribute__((overloadable)) fmod(float x, float y); 448 FN_FUNC_FN_FN(fmod) 449 450 451 /** 452 * Return fractional part of v 453 * 454 * @param iptr iptr[0] will be set to the floor of the input value. 455 * Supports float, float2, float3, float4. 456 */ 457 _RS_RUNTIME float __attribute__((overloadable)) fract(float v, float *iptr); 458 FN_FUNC_FN_PFN(fract) 459 460 /** 461 * Return the mantissa and place the exponent into iptr[0] 462 * 463 * @param v Supports float, float2, float3, float4. 464 * @param iptr Must have the same vector size as v. 465 */ 466 extern float __attribute__((overloadable)) frexp(float v, int *iptr); 467 FN_FUNC_FN_PIN(frexp) 468 469 /** 470 * Return sqrt(x*x + y*y) 471 * 472 * Supports float, float2, float3, float4. 473 */ 474 extern float __attribute__((overloadable)) hypot(float x, float y); 475 FN_FUNC_FN_FN(hypot) 476 477 /** 478 * Return the integer exponent of a value 479 * 480 * Supports 1,2,3,4 components 481 */ 482 extern int __attribute__((overloadable)) ilogb(float); 483 IN_FUNC_FN(ilogb) 484 485 /** 486 * Return (x * 2^y) 487 * 488 * @param x Supports 1,2,3,4 components 489 * @param y Supports single component or matching vector. 490 */ 491 extern float __attribute__((overloadable)) ldexp(float x, int y); 492 FN_FUNC_FN_IN(ldexp) 493 FN_FUNC_FN_I(ldexp) 494 495 /** 496 * Return the log gamma 497 * 498 * Supports 1,2,3,4 components 499 */ 500 extern float __attribute__((overloadable)) lgamma(float); 501 FN_FUNC_FN(lgamma) 502 503 /** 504 * Return the log gamma and sign 505 * 506 * @param x Supports 1,2,3,4 components 507 * @param y Supports matching vector. 508 */ 509 extern float __attribute__((overloadable)) lgamma(float x, int* y); 510 FN_FUNC_FN_PIN(lgamma) 511 512 /** 513 * Return the natural logarithm 514 * 515 * Supports 1,2,3,4 components 516 */ 517 extern float __attribute__((overloadable)) log(float); 518 FN_FUNC_FN(log) 519 520 /** 521 * Return the base 10 logarithm 522 * 523 * Supports 1,2,3,4 components 524 */ 525 extern float __attribute__((overloadable)) log10(float); 526 FN_FUNC_FN(log10) 527 528 /** 529 * Return the base 2 logarithm 530 * 531 * Supports 1,2,3,4 components 532 */ 533 _RS_RUNTIME float __attribute__((overloadable)) log2(float v); 534 FN_FUNC_FN(log2) 535 536 /** 537 * Return the natural logarithm of (v + 1.0f) 538 * 539 * Supports 1,2,3,4 components 540 */ 541 extern float __attribute__((overloadable)) log1p(float v); 542 FN_FUNC_FN(log1p) 543 544 /** 545 * Compute the exponent of the value. 546 * 547 * Supports 1,2,3,4 components 548 */ 549 extern float __attribute__((overloadable)) logb(float); 550 FN_FUNC_FN(logb) 551 552 /** 553 * Compute (a * b) + c 554 * 555 * Supports 1,2,3,4 components 556 */ 557 extern float __attribute__((overloadable)) mad(float a, float b, float c); 558 FN_FUNC_FN_FN_FN(mad) 559 560 /** 561 * Return the integral and fractional components of a number. 562 * Supports 1,2,3,4 components 563 * 564 * @param x Source value 565 * @param iret iret[0] will be set to the integral portion of the number. 566 * @return The floating point portion of the value. 567 */ 568 extern float __attribute__((overloadable)) modf(float x, float *iret); 569 FN_FUNC_FN_PFN(modf); 570 571 extern float __attribute__((overloadable)) nan(uint); 572 573 /** 574 * Return the next floating point number from x towards y. 575 * 576 * Supports 1,2,3,4 components 577 */ 578 extern float __attribute__((overloadable)) nextafter(float x, float y); 579 FN_FUNC_FN_FN(nextafter) 580 581 /** 582 * Return (v ^ p). 583 * 584 * Supports 1,2,3,4 components 585 */ 586 _RS_RUNTIME float __attribute__((overloadable)) pown(float v, int p); 587 FN_FUNC_FN_IN(pown) 588 589 /** 590 * Return (v ^ p). 591 * @param v must be greater than 0. 592 * 593 * Supports 1,2,3,4 components 594 */ 595 _RS_RUNTIME float __attribute__((overloadable)) powr(float v, float p); 596 FN_FUNC_FN_FN(powr) 597 598 /** 599 * Return round x/y to the nearest integer then compute the remander. 600 * 601 * Supports 1,2,3,4 components 602 */ 603 extern float __attribute__((overloadable)) remainder(float x, float y); 604 FN_FUNC_FN_FN(remainder) 605 606 // document once we know the precision of bionic 607 extern float __attribute__((overloadable)) remquo(float, float, int *); 608 FN_FUNC_FN_FN_PIN(remquo) 609 610 /** 611 * Round to the nearest integral value. 612 * 613 * Supports 1,2,3,4 components 614 */ 615 extern float __attribute__((overloadable)) rint(float); 616 FN_FUNC_FN(rint) 617 618 /** 619 * Compute the Nth root of a value. 620 * 621 * Supports 1,2,3,4 components 622 */ 623 _RS_RUNTIME float __attribute__((overloadable)) rootn(float v, int n); 624 FN_FUNC_FN_IN(rootn) 625 626 /** 627 * Round to the nearest integral value. Half values are rounded away from zero. 628 * 629 * Supports 1,2,3,4 components 630 */ 631 extern float __attribute__((overloadable)) round(float); 632 FN_FUNC_FN(round) 633 634 /** 635 * Return the square root of a value. 636 * 637 * Supports 1,2,3,4 components 638 */ 639 extern float __attribute__((overloadable)) sqrt(float); 640 FN_FUNC_FN(sqrt) 641 642 /** 643 * Return (1 / sqrt(value)). 644 * 645 * Supports 1,2,3,4 components 646 */ 647 _RS_RUNTIME float __attribute__((overloadable)) rsqrt(float v); 648 FN_FUNC_FN(rsqrt) 649 650 /** 651 * Return the sine of a value specified in radians. 652 * 653 * @param v The incoming value in radians 654 * Supports 1,2,3,4 components 655 */ 656 extern float __attribute__((overloadable)) sin(float v); 657 FN_FUNC_FN(sin) 658 659 /** 660 * Return the sine and cosine of a value. 661 * 662 * @return sine 663 * @param v The incoming value in radians 664 * @param *cosptr cosptr[0] will be set to the cosine value. 665 * 666 * Supports 1,2,3,4 components 667 */ 668 _RS_RUNTIME float __attribute__((overloadable)) sincos(float v, float *cosptr); 669 FN_FUNC_FN_PFN(sincos); 670 671 /** 672 * Return the hyperbolic sine of a value specified in radians. 673 * 674 * Supports 1,2,3,4 components 675 */ 676 extern float __attribute__((overloadable)) sinh(float); 677 FN_FUNC_FN(sinh) 678 679 /** 680 * Return the sin(v * PI). 681 * 682 * Supports 1,2,3,4 components 683 */ 684 _RS_RUNTIME float __attribute__((overloadable)) sinpi(float v); 685 FN_FUNC_FN(sinpi) 686 687 /** 688 * Return the tangent of a value. 689 * 690 * Supports 1,2,3,4 components 691 * @param v The incoming value in radians 692 */ 693 extern float __attribute__((overloadable)) tan(float v); 694 FN_FUNC_FN(tan) 695 696 /** 697 * Return the hyperbolic tangent of a value. 698 * 699 * Supports 1,2,3,4 components 700 * @param v The incoming value in radians 701 */ 702 extern float __attribute__((overloadable)) tanh(float); 703 FN_FUNC_FN(tanh) 704 705 /** 706 * Return tan(v * PI) 707 * 708 * Supports 1,2,3,4 components 709 */ 710 _RS_RUNTIME float __attribute__((overloadable)) tanpi(float v); 711 FN_FUNC_FN(tanpi) 712 713 /** 714 * Compute the gamma function of a value. 715 * 716 * Supports 1,2,3,4 components 717 */ 718 extern float __attribute__((overloadable)) tgamma(float); 719 FN_FUNC_FN(tgamma) 720 721 /** 722 * Round to integral using truncation. 723 * 724 * Supports 1,2,3,4 components 725 */ 726 extern float __attribute__((overloadable)) trunc(float); 727 FN_FUNC_FN(trunc) 728 729 730 #define XN_FUNC_YN(typeout, fnc, typein) \ 731 extern typeout __attribute__((overloadable)) fnc(typein); \ 732 _RS_RUNTIME typeout##2 __attribute__((overloadable)) fnc(typein##2 v); \ 733 _RS_RUNTIME typeout##3 __attribute__((overloadable)) fnc(typein##3 v); \ 734 _RS_RUNTIME typeout##4 __attribute__((overloadable)) fnc(typein##4 v); 735 736 #define UIN_FUNC_IN(fnc) \ 737 XN_FUNC_YN(uchar, fnc, char) \ 738 XN_FUNC_YN(ushort, fnc, short) \ 739 XN_FUNC_YN(uint, fnc, int) 740 741 #define IN_FUNC_IN(fnc) \ 742 XN_FUNC_YN(uchar, fnc, uchar) \ 743 XN_FUNC_YN(char, fnc, char) \ 744 XN_FUNC_YN(ushort, fnc, ushort) \ 745 XN_FUNC_YN(short, fnc, short) \ 746 XN_FUNC_YN(uint, fnc, uint) \ 747 XN_FUNC_YN(int, fnc, int) 748 749 750 #define XN_FUNC_XN_XN_BODY(type, fnc, body) \ 751 _RS_RUNTIME type __attribute__((overloadable)) \ 752 fnc(type v1, type v2); \ 753 _RS_RUNTIME type##2 __attribute__((overloadable)) \ 754 fnc(type##2 v1, type##2 v2); \ 755 _RS_RUNTIME type##3 __attribute__((overloadable)) \ 756 fnc(type##3 v1, type##3 v2); \ 757 _RS_RUNTIME type##4 __attribute__((overloadable)) \ 758 fnc(type##4 v1, type##4 v2); 759 760 #define IN_FUNC_IN_IN_BODY(fnc, body) \ 761 XN_FUNC_XN_XN_BODY(uchar, fnc, body) \ 762 XN_FUNC_XN_XN_BODY(char, fnc, body) \ 763 XN_FUNC_XN_XN_BODY(ushort, fnc, body) \ 764 XN_FUNC_XN_XN_BODY(short, fnc, body) \ 765 XN_FUNC_XN_XN_BODY(uint, fnc, body) \ 766 XN_FUNC_XN_XN_BODY(int, fnc, body) \ 767 XN_FUNC_XN_XN_BODY(float, fnc, body) 768 769 /** 770 * Return the absolute value of a value. 771 * 772 * Supports 1,2,3,4 components of char, short, int. 773 */ 774 UIN_FUNC_IN(abs) 775 776 /** 777 * Return the number of leading 0-bits in a value. 778 * 779 * Supports 1,2,3,4 components of uchar, char, ushort, short, uint, int. 780 */ 781 IN_FUNC_IN(clz) 782 783 /** 784 * Return the minimum of two values. 785 * 786 * Supports 1,2,3,4 components of uchar, char, ushort, short, uint, int, float. 787 */ 788 IN_FUNC_IN_IN_BODY(min, (v1 < v2 ? v1 : v2)) 789 FN_FUNC_FN_F(min) 790 791 /** 792 * Return the maximum of two values. 793 * 794 * Supports 1,2,3,4 components of uchar, char, ushort, short, uint, int, float. 795 */ 796 IN_FUNC_IN_IN_BODY(max, (v1 > v2 ? v1 : v2)) 797 FN_FUNC_FN_F(max) 798 799 /** 800 * Clamp a value to a specified high and low bound. 801 * 802 * @param amount value to be clamped. Supports 1,2,3,4 components 803 * @param low Lower bound, must be scalar or matching vector. 804 * @param high High bound, must match type of low 805 */ 806 _RS_RUNTIME float __attribute__((overloadable)) clamp(float amount, float low, float high); 807 FN_FUNC_FN_FN_FN(clamp) 808 FN_FUNC_FN_F_F(clamp) 809 810 /** 811 * Convert from radians to degrees. 812 * 813 * Supports 1,2,3,4 components 814 */ 815 _RS_RUNTIME float __attribute__((overloadable)) degrees(float radians); 816 FN_FUNC_FN(degrees) 817 818 /** 819 * return start + ((stop - start) * amount); 820 * 821 * Supports 1,2,3,4 components 822 */ 823 _RS_RUNTIME float __attribute__((overloadable)) mix(float start, float stop, float amount); 824 FN_FUNC_FN_FN_FN(mix) 825 FN_FUNC_FN_FN_F(mix) 826 827 /** 828 * Convert from degrees to radians. 829 * 830 * Supports 1,2,3,4 components 831 */ 832 _RS_RUNTIME float __attribute__((overloadable)) radians(float degrees); 833 FN_FUNC_FN(radians) 834 835 /** 836 * if (v < edge) 837 * return 0.f; 838 * else 839 * return 1.f; 840 * 841 * Supports 1,2,3,4 components 842 */ 843 _RS_RUNTIME float __attribute__((overloadable)) step(float edge, float v); 844 FN_FUNC_FN_FN(step) 845 FN_FUNC_FN_F(step) 846 847 // not implemented 848 extern float __attribute__((overloadable)) smoothstep(float, float, float); 849 extern float2 __attribute__((overloadable)) smoothstep(float2, float2, float2); 850 extern float3 __attribute__((overloadable)) smoothstep(float3, float3, float3); 851 extern float4 __attribute__((overloadable)) smoothstep(float4, float4, float4); 852 extern float2 __attribute__((overloadable)) smoothstep(float, float, float2); 853 extern float3 __attribute__((overloadable)) smoothstep(float, float, float3); 854 extern float4 __attribute__((overloadable)) smoothstep(float, float, float4); 855 856 /** 857 * Return the sign of a value. 858 * 859 * if (v < 0) return -1.f; 860 * else if (v > 0) return 1.f; 861 * else return 0.f; 862 * 863 * Supports 1,2,3,4 components 864 */ 865 _RS_RUNTIME float __attribute__((overloadable)) sign(float v); 866 FN_FUNC_FN(sign) 867 868 /** 869 * Compute the cross product of two vectors. 870 * 871 * Supports 3,4 components 872 */ 873 _RS_RUNTIME float3 __attribute__((overloadable)) cross(float3 lhs, float3 rhs); 874 _RS_RUNTIME float4 __attribute__((overloadable)) cross(float4 lhs, float4 rhs); 875 876 /** 877 * Compute the dot product of two vectors. 878 * 879 * Supports 1,2,3,4 components 880 */ 881 _RS_RUNTIME float __attribute__((overloadable)) dot(float lhs, float rhs); 882 F_FUNC_FN_FN(dot) 883 884 /** 885 * Compute the length of a vector. 886 * 887 * Supports 1,2,3,4 components 888 */ 889 _RS_RUNTIME float __attribute__((overloadable)) length(float v); 890 F_FUNC_FN(length) 891 892 /** 893 * Compute the distance between two points. 894 * 895 * Supports 1,2,3,4 components 896 */ 897 _RS_RUNTIME float __attribute__((overloadable)) distance(float lhs, float rhs); 898 F_FUNC_FN_FN(distance) 899 900 /** 901 * Normalize a vector. 902 * 903 * Supports 1,2,3,4 components 904 */ 905 _RS_RUNTIME float __attribute__((overloadable)) normalize(float v); 906 FN_FUNC_FN(normalize) 907 908 909 // New approx API functions 910 #if (defined(RS_VERSION) && (RS_VERSION >= 17)) 911 912 /** 913 * Return the approximate reciprocal of a value. 914 * 915 * Supports 1,2,3,4 components 916 */ 917 _RS_RUNTIME float __attribute__((overloadable)) half_recip(float); 918 FN_FUNC_FN(half_recip) 919 920 /** 921 * Return the approximate square root of a value. 922 * 923 * Supports 1,2,3,4 components 924 */ 925 _RS_RUNTIME float __attribute__((overloadable)) half_sqrt(float); 926 FN_FUNC_FN(half_sqrt) 927 928 /** 929 * Return the approximate value of (1 / sqrt(value)). 930 * 931 * Supports 1,2,3,4 components 932 */ 933 _RS_RUNTIME float __attribute__((overloadable)) half_rsqrt(float v); 934 FN_FUNC_FN(half_rsqrt) 935 936 /** 937 * Compute the approximate length of a vector. 938 * 939 * Supports 1,2,3,4 components 940 */ 941 _RS_RUNTIME float __attribute__((overloadable)) fast_length(float v); 942 F_FUNC_FN(fast_length) 943 944 /** 945 * Compute the approximate distance between two points. 946 * 947 * Supports 1,2,3,4 components 948 */ 949 _RS_RUNTIME float __attribute__((overloadable)) fast_distance(float lhs, float rhs); 950 F_FUNC_FN_FN(fast_distance) 951 952 /** 953 * Approximately normalize a vector. 954 * 955 * Supports 1,2,3,4 components 956 */ 957 _RS_RUNTIME float __attribute__((overloadable)) fast_normalize(float v); 958 F_FUNC_FN(fast_normalize) 959 960 #endif // (defined(RS_VERSION) && (RS_VERSION >= 17)) 961 962 963 #undef CVT_FUNC 964 #undef CVT_FUNC_2 965 #undef FN_FUNC_FN 966 #undef F_FUNC_FN 967 #undef IN_FUNC_FN 968 #undef FN_FUNC_FN_FN 969 #undef F_FUNC_FN_FN 970 #undef FN_FUNC_FN_F 971 #undef FN_FUNC_FN_IN 972 #undef FN_FUNC_FN_I 973 #undef FN_FUNC_FN_PFN 974 #undef FN_FUNC_FN_PIN 975 #undef FN_FUNC_FN_FN_FN 976 #undef FN_FUNC_FN_FN_F 977 #undef FN_FUNC_FN_F_F 978 #undef FN_FUNC_FN_FN_PIN 979 #undef XN_FUNC_YN 980 #undef UIN_FUNC_IN 981 #undef IN_FUNC_IN 982 #undef XN_FUNC_XN_XN_BODY 983 #undef IN_FUNC_IN_IN_BODY 984 985 #endif 986