dct.c 88.9 KB
Newer Older
Jingning Han's avatar
Jingning Han committed
1
/*
Yaowu Xu's avatar
Yaowu Xu committed
2
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
Jingning Han's avatar
Jingning Han committed
3
 *
Yaowu Xu's avatar
Yaowu Xu committed
4 5 6 7 8 9
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
Jingning Han's avatar
Jingning Han committed
10 11 12 13 14
 */

#include <assert.h>
#include <math.h>

Yaowu Xu's avatar
Yaowu Xu committed
15 16
#include "./aom_config.h"
#include "./aom_dsp_rtcd.h"
Geza Lore's avatar
Geza Lore committed
17
#include "./av1_rtcd.h"
18 19
#include "aom_dsp/fwd_txfm.h"
#include "aom_ports/mem.h"
Geza Lore's avatar
Geza Lore committed
20 21
#include "av1/common/blockd.h"
#include "av1/common/idct.h"
22 23
#if CONFIG_DAALA_TX4 || CONFIG_DAALA_TX8 || CONFIG_DAALA_TX16 || \
    CONFIG_DAALA_TX32 || CONFIG_DAALA_TX64
24 25
#include "av1/common/daala_tx.h"
#endif
26 27
#include "av1/encoder/av1_fwd_txfm1d.h"
#include "av1/encoder/av1_fwd_txfm1d_cfg.h"
Jingning Han's avatar
Jingning Han committed
28

29 30
static INLINE void range_check(const tran_low_t *input, const int size,
                               const int bit) {
31 32 33 34
#if 0  // CONFIG_COEFFICIENT_RANGE_CHECKING
// TODO(angiebird): the range_check is not used because the bit range
// in fdct# is not correct. Since we are going to merge in a new version
// of fdct# from nextgenv2, we won't fix the incorrect bit range now.
35 36 37 38 39 40 41 42 43 44 45
  int i;
  for (i = 0; i < size; ++i) {
    assert(abs(input[i]) < (1 << bit));
  }
#else
  (void)input;
  (void)size;
  (void)bit;
#endif
}

Jingning Han's avatar
Jingning Han committed
46
static void fdct4(const tran_low_t *input, tran_low_t *output) {
47 48 49 50
  tran_high_t temp;
  tran_low_t step[4];

  // stage 0
51
  range_check(input, 4, 14);
52 53 54 55 56 57 58

  // stage 1
  output[0] = input[0] + input[3];
  output[1] = input[1] + input[2];
  output[2] = input[1] - input[2];
  output[3] = input[0] - input[3];

59
  range_check(output, 4, 15);
60 61 62 63 64 65 66 67 68 69 70

  // stage 2
  temp = output[0] * cospi_16_64 + output[1] * cospi_16_64;
  step[0] = (tran_low_t)fdct_round_shift(temp);
  temp = output[1] * -cospi_16_64 + output[0] * cospi_16_64;
  step[1] = (tran_low_t)fdct_round_shift(temp);
  temp = output[2] * cospi_24_64 + output[3] * cospi_8_64;
  step[2] = (tran_low_t)fdct_round_shift(temp);
  temp = output[3] * cospi_24_64 + output[2] * -cospi_8_64;
  step[3] = (tran_low_t)fdct_round_shift(temp);

71
  range_check(step, 4, 16);
72 73 74 75 76 77 78

  // stage 3
  output[0] = step[0];
  output[1] = step[2];
  output[2] = step[1];
  output[3] = step[3];

79
  range_check(output, 4, 16);
Jingning Han's avatar
Jingning Han committed
80 81 82
}

static void fdct8(const tran_low_t *input, tran_low_t *output) {
83 84 85 86
  tran_high_t temp;
  tran_low_t step[8];

  // stage 0
87
  range_check(input, 8, 13);
Jingning Han's avatar
Jingning Han committed
88 89

  // stage 1
90 91 92 93 94 95 96 97 98
  output[0] = input[0] + input[7];
  output[1] = input[1] + input[6];
  output[2] = input[2] + input[5];
  output[3] = input[3] + input[4];
  output[4] = input[3] - input[4];
  output[5] = input[2] - input[5];
  output[6] = input[1] - input[6];
  output[7] = input[0] - input[7];

99
  range_check(output, 8, 14);
100 101 102 103 104 105 106 107 108 109 110 111 112

  // stage 2
  step[0] = output[0] + output[3];
  step[1] = output[1] + output[2];
  step[2] = output[1] - output[2];
  step[3] = output[0] - output[3];
  step[4] = output[4];
  temp = output[5] * -cospi_16_64 + output[6] * cospi_16_64;
  step[5] = (tran_low_t)fdct_round_shift(temp);
  temp = output[6] * cospi_16_64 + output[5] * cospi_16_64;
  step[6] = (tran_low_t)fdct_round_shift(temp);
  step[7] = output[7];

113
  range_check(step, 8, 15);
114 115 116 117 118 119 120 121 122 123 124 125 126 127 128

  // stage 3
  temp = step[0] * cospi_16_64 + step[1] * cospi_16_64;
  output[0] = (tran_low_t)fdct_round_shift(temp);
  temp = step[1] * -cospi_16_64 + step[0] * cospi_16_64;
  output[1] = (tran_low_t)fdct_round_shift(temp);
  temp = step[2] * cospi_24_64 + step[3] * cospi_8_64;
  output[2] = (tran_low_t)fdct_round_shift(temp);
  temp = step[3] * cospi_24_64 + step[2] * -cospi_8_64;
  output[3] = (tran_low_t)fdct_round_shift(temp);
  output[4] = step[4] + step[5];
  output[5] = step[4] - step[5];
  output[6] = step[7] - step[6];
  output[7] = step[7] + step[6];

129
  range_check(output, 8, 16);
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144

  // stage 4
  step[0] = output[0];
  step[1] = output[1];
  step[2] = output[2];
  step[3] = output[3];
  temp = output[4] * cospi_28_64 + output[7] * cospi_4_64;
  step[4] = (tran_low_t)fdct_round_shift(temp);
  temp = output[5] * cospi_12_64 + output[6] * cospi_20_64;
  step[5] = (tran_low_t)fdct_round_shift(temp);
  temp = output[6] * cospi_12_64 + output[5] * -cospi_20_64;
  step[6] = (tran_low_t)fdct_round_shift(temp);
  temp = output[7] * cospi_28_64 + output[4] * -cospi_4_64;
  step[7] = (tran_low_t)fdct_round_shift(temp);

145
  range_check(step, 8, 16);
146 147 148 149 150 151 152 153 154 155 156

  // stage 5
  output[0] = step[0];
  output[1] = step[4];
  output[2] = step[2];
  output[3] = step[6];
  output[4] = step[1];
  output[5] = step[5];
  output[6] = step[3];
  output[7] = step[7];

157
  range_check(output, 8, 16);
Jingning Han's avatar
Jingning Han committed
158 159
}

160 161 162
static void fdct16(const tran_low_t *input, tran_low_t *output) {
  tran_high_t temp;
  tran_low_t step[16];
Jingning Han's avatar
Jingning Han committed
163

164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335
  // stage 0
  range_check(input, 16, 13);

  // stage 1
  output[0] = input[0] + input[15];
  output[1] = input[1] + input[14];
  output[2] = input[2] + input[13];
  output[3] = input[3] + input[12];
  output[4] = input[4] + input[11];
  output[5] = input[5] + input[10];
  output[6] = input[6] + input[9];
  output[7] = input[7] + input[8];
  output[8] = input[7] - input[8];
  output[9] = input[6] - input[9];
  output[10] = input[5] - input[10];
  output[11] = input[4] - input[11];
  output[12] = input[3] - input[12];
  output[13] = input[2] - input[13];
  output[14] = input[1] - input[14];
  output[15] = input[0] - input[15];

  range_check(output, 16, 14);

  // stage 2
  step[0] = output[0] + output[7];
  step[1] = output[1] + output[6];
  step[2] = output[2] + output[5];
  step[3] = output[3] + output[4];
  step[4] = output[3] - output[4];
  step[5] = output[2] - output[5];
  step[6] = output[1] - output[6];
  step[7] = output[0] - output[7];
  step[8] = output[8];
  step[9] = output[9];
  temp = output[10] * -cospi_16_64 + output[13] * cospi_16_64;
  step[10] = (tran_low_t)fdct_round_shift(temp);
  temp = output[11] * -cospi_16_64 + output[12] * cospi_16_64;
  step[11] = (tran_low_t)fdct_round_shift(temp);
  temp = output[12] * cospi_16_64 + output[11] * cospi_16_64;
  step[12] = (tran_low_t)fdct_round_shift(temp);
  temp = output[13] * cospi_16_64 + output[10] * cospi_16_64;
  step[13] = (tran_low_t)fdct_round_shift(temp);
  step[14] = output[14];
  step[15] = output[15];

  range_check(step, 16, 15);

  // stage 3
  output[0] = step[0] + step[3];
  output[1] = step[1] + step[2];
  output[2] = step[1] - step[2];
  output[3] = step[0] - step[3];
  output[4] = step[4];
  temp = step[5] * -cospi_16_64 + step[6] * cospi_16_64;
  output[5] = (tran_low_t)fdct_round_shift(temp);
  temp = step[6] * cospi_16_64 + step[5] * cospi_16_64;
  output[6] = (tran_low_t)fdct_round_shift(temp);
  output[7] = step[7];
  output[8] = step[8] + step[11];
  output[9] = step[9] + step[10];
  output[10] = step[9] - step[10];
  output[11] = step[8] - step[11];
  output[12] = step[15] - step[12];
  output[13] = step[14] - step[13];
  output[14] = step[14] + step[13];
  output[15] = step[15] + step[12];

  range_check(output, 16, 16);

  // stage 4
  temp = output[0] * cospi_16_64 + output[1] * cospi_16_64;
  step[0] = (tran_low_t)fdct_round_shift(temp);
  temp = output[1] * -cospi_16_64 + output[0] * cospi_16_64;
  step[1] = (tran_low_t)fdct_round_shift(temp);
  temp = output[2] * cospi_24_64 + output[3] * cospi_8_64;
  step[2] = (tran_low_t)fdct_round_shift(temp);
  temp = output[3] * cospi_24_64 + output[2] * -cospi_8_64;
  step[3] = (tran_low_t)fdct_round_shift(temp);
  step[4] = output[4] + output[5];
  step[5] = output[4] - output[5];
  step[6] = output[7] - output[6];
  step[7] = output[7] + output[6];
  step[8] = output[8];
  temp = output[9] * -cospi_8_64 + output[14] * cospi_24_64;
  step[9] = (tran_low_t)fdct_round_shift(temp);
  temp = output[10] * -cospi_24_64 + output[13] * -cospi_8_64;
  step[10] = (tran_low_t)fdct_round_shift(temp);
  step[11] = output[11];
  step[12] = output[12];
  temp = output[13] * cospi_24_64 + output[10] * -cospi_8_64;
  step[13] = (tran_low_t)fdct_round_shift(temp);
  temp = output[14] * cospi_8_64 + output[9] * cospi_24_64;
  step[14] = (tran_low_t)fdct_round_shift(temp);
  step[15] = output[15];

  range_check(step, 16, 16);

  // stage 5
  output[0] = step[0];
  output[1] = step[1];
  output[2] = step[2];
  output[3] = step[3];
  temp = step[4] * cospi_28_64 + step[7] * cospi_4_64;
  output[4] = (tran_low_t)fdct_round_shift(temp);
  temp = step[5] * cospi_12_64 + step[6] * cospi_20_64;
  output[5] = (tran_low_t)fdct_round_shift(temp);
  temp = step[6] * cospi_12_64 + step[5] * -cospi_20_64;
  output[6] = (tran_low_t)fdct_round_shift(temp);
  temp = step[7] * cospi_28_64 + step[4] * -cospi_4_64;
  output[7] = (tran_low_t)fdct_round_shift(temp);
  output[8] = step[8] + step[9];
  output[9] = step[8] - step[9];
  output[10] = step[11] - step[10];
  output[11] = step[11] + step[10];
  output[12] = step[12] + step[13];
  output[13] = step[12] - step[13];
  output[14] = step[15] - step[14];
  output[15] = step[15] + step[14];

  range_check(output, 16, 16);

  // stage 6
  step[0] = output[0];
  step[1] = output[1];
  step[2] = output[2];
  step[3] = output[3];
  step[4] = output[4];
  step[5] = output[5];
  step[6] = output[6];
  step[7] = output[7];
  temp = output[8] * cospi_30_64 + output[15] * cospi_2_64;
  step[8] = (tran_low_t)fdct_round_shift(temp);
  temp = output[9] * cospi_14_64 + output[14] * cospi_18_64;
  step[9] = (tran_low_t)fdct_round_shift(temp);
  temp = output[10] * cospi_22_64 + output[13] * cospi_10_64;
  step[10] = (tran_low_t)fdct_round_shift(temp);
  temp = output[11] * cospi_6_64 + output[12] * cospi_26_64;
  step[11] = (tran_low_t)fdct_round_shift(temp);
  temp = output[12] * cospi_6_64 + output[11] * -cospi_26_64;
  step[12] = (tran_low_t)fdct_round_shift(temp);
  temp = output[13] * cospi_22_64 + output[10] * -cospi_10_64;
  step[13] = (tran_low_t)fdct_round_shift(temp);
  temp = output[14] * cospi_14_64 + output[9] * -cospi_18_64;
  step[14] = (tran_low_t)fdct_round_shift(temp);
  temp = output[15] * cospi_30_64 + output[8] * -cospi_2_64;
  step[15] = (tran_low_t)fdct_round_shift(temp);

  range_check(step, 16, 16);

  // stage 7
  output[0] = step[0];
  output[1] = step[8];
  output[2] = step[4];
  output[3] = step[12];
  output[4] = step[2];
  output[5] = step[10];
  output[6] = step[6];
  output[7] = step[14];
  output[8] = step[1];
  output[9] = step[9];
  output[10] = step[5];
  output[11] = step[13];
  output[12] = step[3];
  output[13] = step[11];
  output[14] = step[7];
  output[15] = step[15];

  range_check(output, 16, 16);
}

static void fdct32(const tran_low_t *input, tran_low_t *output) {
  tran_high_t temp;
336
  tran_low_t step[32];
337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375

  // stage 0
  range_check(input, 32, 14);

  // stage 1
  output[0] = input[0] + input[31];
  output[1] = input[1] + input[30];
  output[2] = input[2] + input[29];
  output[3] = input[3] + input[28];
  output[4] = input[4] + input[27];
  output[5] = input[5] + input[26];
  output[6] = input[6] + input[25];
  output[7] = input[7] + input[24];
  output[8] = input[8] + input[23];
  output[9] = input[9] + input[22];
  output[10] = input[10] + input[21];
  output[11] = input[11] + input[20];
  output[12] = input[12] + input[19];
  output[13] = input[13] + input[18];
  output[14] = input[14] + input[17];
  output[15] = input[15] + input[16];
  output[16] = input[15] - input[16];
  output[17] = input[14] - input[17];
  output[18] = input[13] - input[18];
  output[19] = input[12] - input[19];
  output[20] = input[11] - input[20];
  output[21] = input[10] - input[21];
  output[22] = input[9] - input[22];
  output[23] = input[8] - input[23];
  output[24] = input[7] - input[24];
  output[25] = input[6] - input[25];
  output[26] = input[5] - input[26];
  output[27] = input[4] - input[27];
  output[28] = input[3] - input[28];
  output[29] = input[2] - input[29];
  output[30] = input[1] - input[30];
  output[31] = input[0] - input[31];

  range_check(output, 32, 15);
Jingning Han's avatar
Jingning Han committed
376

377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418
  // stage 2
  step[0] = output[0] + output[15];
  step[1] = output[1] + output[14];
  step[2] = output[2] + output[13];
  step[3] = output[3] + output[12];
  step[4] = output[4] + output[11];
  step[5] = output[5] + output[10];
  step[6] = output[6] + output[9];
  step[7] = output[7] + output[8];
  step[8] = output[7] - output[8];
  step[9] = output[6] - output[9];
  step[10] = output[5] - output[10];
  step[11] = output[4] - output[11];
  step[12] = output[3] - output[12];
  step[13] = output[2] - output[13];
  step[14] = output[1] - output[14];
  step[15] = output[0] - output[15];
  step[16] = output[16];
  step[17] = output[17];
  step[18] = output[18];
  step[19] = output[19];
  temp = output[20] * -cospi_16_64 + output[27] * cospi_16_64;
  step[20] = (tran_low_t)fdct_round_shift(temp);
  temp = output[21] * -cospi_16_64 + output[26] * cospi_16_64;
  step[21] = (tran_low_t)fdct_round_shift(temp);
  temp = output[22] * -cospi_16_64 + output[25] * cospi_16_64;
  step[22] = (tran_low_t)fdct_round_shift(temp);
  temp = output[23] * -cospi_16_64 + output[24] * cospi_16_64;
  step[23] = (tran_low_t)fdct_round_shift(temp);
  temp = output[24] * cospi_16_64 + output[23] * cospi_16_64;
  step[24] = (tran_low_t)fdct_round_shift(temp);
  temp = output[25] * cospi_16_64 + output[22] * cospi_16_64;
  step[25] = (tran_low_t)fdct_round_shift(temp);
  temp = output[26] * cospi_16_64 + output[21] * cospi_16_64;
  step[26] = (tran_low_t)fdct_round_shift(temp);
  temp = output[27] * cospi_16_64 + output[20] * cospi_16_64;
  step[27] = (tran_low_t)fdct_round_shift(temp);
  step[28] = output[28];
  step[29] = output[29];
  step[30] = output[30];
  step[31] = output[31];

419
  range_check(step, 32, 16);
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504

  // stage 3
  output[0] = step[0] + step[7];
  output[1] = step[1] + step[6];
  output[2] = step[2] + step[5];
  output[3] = step[3] + step[4];
  output[4] = step[3] - step[4];
  output[5] = step[2] - step[5];
  output[6] = step[1] - step[6];
  output[7] = step[0] - step[7];
  output[8] = step[8];
  output[9] = step[9];
  temp = step[10] * -cospi_16_64 + step[13] * cospi_16_64;
  output[10] = (tran_low_t)fdct_round_shift(temp);
  temp = step[11] * -cospi_16_64 + step[12] * cospi_16_64;
  output[11] = (tran_low_t)fdct_round_shift(temp);
  temp = step[12] * cospi_16_64 + step[11] * cospi_16_64;
  output[12] = (tran_low_t)fdct_round_shift(temp);
  temp = step[13] * cospi_16_64 + step[10] * cospi_16_64;
  output[13] = (tran_low_t)fdct_round_shift(temp);
  output[14] = step[14];
  output[15] = step[15];
  output[16] = step[16] + step[23];
  output[17] = step[17] + step[22];
  output[18] = step[18] + step[21];
  output[19] = step[19] + step[20];
  output[20] = step[19] - step[20];
  output[21] = step[18] - step[21];
  output[22] = step[17] - step[22];
  output[23] = step[16] - step[23];
  output[24] = step[31] - step[24];
  output[25] = step[30] - step[25];
  output[26] = step[29] - step[26];
  output[27] = step[28] - step[27];
  output[28] = step[28] + step[27];
  output[29] = step[29] + step[26];
  output[30] = step[30] + step[25];
  output[31] = step[31] + step[24];

  range_check(output, 32, 17);

  // stage 4
  step[0] = output[0] + output[3];
  step[1] = output[1] + output[2];
  step[2] = output[1] - output[2];
  step[3] = output[0] - output[3];
  step[4] = output[4];
  temp = output[5] * -cospi_16_64 + output[6] * cospi_16_64;
  step[5] = (tran_low_t)fdct_round_shift(temp);
  temp = output[6] * cospi_16_64 + output[5] * cospi_16_64;
  step[6] = (tran_low_t)fdct_round_shift(temp);
  step[7] = output[7];
  step[8] = output[8] + output[11];
  step[9] = output[9] + output[10];
  step[10] = output[9] - output[10];
  step[11] = output[8] - output[11];
  step[12] = output[15] - output[12];
  step[13] = output[14] - output[13];
  step[14] = output[14] + output[13];
  step[15] = output[15] + output[12];
  step[16] = output[16];
  step[17] = output[17];
  temp = output[18] * -cospi_8_64 + output[29] * cospi_24_64;
  step[18] = (tran_low_t)fdct_round_shift(temp);
  temp = output[19] * -cospi_8_64 + output[28] * cospi_24_64;
  step[19] = (tran_low_t)fdct_round_shift(temp);
  temp = output[20] * -cospi_24_64 + output[27] * -cospi_8_64;
  step[20] = (tran_low_t)fdct_round_shift(temp);
  temp = output[21] * -cospi_24_64 + output[26] * -cospi_8_64;
  step[21] = (tran_low_t)fdct_round_shift(temp);
  step[22] = output[22];
  step[23] = output[23];
  step[24] = output[24];
  step[25] = output[25];
  temp = output[26] * cospi_24_64 + output[21] * -cospi_8_64;
  step[26] = (tran_low_t)fdct_round_shift(temp);
  temp = output[27] * cospi_24_64 + output[20] * -cospi_8_64;
  step[27] = (tran_low_t)fdct_round_shift(temp);
  temp = output[28] * cospi_8_64 + output[19] * cospi_24_64;
  step[28] = (tran_low_t)fdct_round_shift(temp);
  temp = output[29] * cospi_8_64 + output[18] * cospi_24_64;
  step[29] = (tran_low_t)fdct_round_shift(temp);
  step[30] = output[30];
  step[31] = output[31];

505
  range_check(step, 32, 18);
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596

  // stage 5
  temp = step[0] * cospi_16_64 + step[1] * cospi_16_64;
  output[0] = (tran_low_t)fdct_round_shift(temp);
  temp = step[1] * -cospi_16_64 + step[0] * cospi_16_64;
  output[1] = (tran_low_t)fdct_round_shift(temp);
  temp = step[2] * cospi_24_64 + step[3] * cospi_8_64;
  output[2] = (tran_low_t)fdct_round_shift(temp);
  temp = step[3] * cospi_24_64 + step[2] * -cospi_8_64;
  output[3] = (tran_low_t)fdct_round_shift(temp);
  output[4] = step[4] + step[5];
  output[5] = step[4] - step[5];
  output[6] = step[7] - step[6];
  output[7] = step[7] + step[6];
  output[8] = step[8];
  temp = step[9] * -cospi_8_64 + step[14] * cospi_24_64;
  output[9] = (tran_low_t)fdct_round_shift(temp);
  temp = step[10] * -cospi_24_64 + step[13] * -cospi_8_64;
  output[10] = (tran_low_t)fdct_round_shift(temp);
  output[11] = step[11];
  output[12] = step[12];
  temp = step[13] * cospi_24_64 + step[10] * -cospi_8_64;
  output[13] = (tran_low_t)fdct_round_shift(temp);
  temp = step[14] * cospi_8_64 + step[9] * cospi_24_64;
  output[14] = (tran_low_t)fdct_round_shift(temp);
  output[15] = step[15];
  output[16] = step[16] + step[19];
  output[17] = step[17] + step[18];
  output[18] = step[17] - step[18];
  output[19] = step[16] - step[19];
  output[20] = step[23] - step[20];
  output[21] = step[22] - step[21];
  output[22] = step[22] + step[21];
  output[23] = step[23] + step[20];
  output[24] = step[24] + step[27];
  output[25] = step[25] + step[26];
  output[26] = step[25] - step[26];
  output[27] = step[24] - step[27];
  output[28] = step[31] - step[28];
  output[29] = step[30] - step[29];
  output[30] = step[30] + step[29];
  output[31] = step[31] + step[28];

  range_check(output, 32, 18);

  // stage 6
  step[0] = output[0];
  step[1] = output[1];
  step[2] = output[2];
  step[3] = output[3];
  temp = output[4] * cospi_28_64 + output[7] * cospi_4_64;
  step[4] = (tran_low_t)fdct_round_shift(temp);
  temp = output[5] * cospi_12_64 + output[6] * cospi_20_64;
  step[5] = (tran_low_t)fdct_round_shift(temp);
  temp = output[6] * cospi_12_64 + output[5] * -cospi_20_64;
  step[6] = (tran_low_t)fdct_round_shift(temp);
  temp = output[7] * cospi_28_64 + output[4] * -cospi_4_64;
  step[7] = (tran_low_t)fdct_round_shift(temp);
  step[8] = output[8] + output[9];
  step[9] = output[8] - output[9];
  step[10] = output[11] - output[10];
  step[11] = output[11] + output[10];
  step[12] = output[12] + output[13];
  step[13] = output[12] - output[13];
  step[14] = output[15] - output[14];
  step[15] = output[15] + output[14];
  step[16] = output[16];
  temp = output[17] * -cospi_4_64 + output[30] * cospi_28_64;
  step[17] = (tran_low_t)fdct_round_shift(temp);
  temp = output[18] * -cospi_28_64 + output[29] * -cospi_4_64;
  step[18] = (tran_low_t)fdct_round_shift(temp);
  step[19] = output[19];
  step[20] = output[20];
  temp = output[21] * -cospi_20_64 + output[26] * cospi_12_64;
  step[21] = (tran_low_t)fdct_round_shift(temp);
  temp = output[22] * -cospi_12_64 + output[25] * -cospi_20_64;
  step[22] = (tran_low_t)fdct_round_shift(temp);
  step[23] = output[23];
  step[24] = output[24];
  temp = output[25] * cospi_12_64 + output[22] * -cospi_20_64;
  step[25] = (tran_low_t)fdct_round_shift(temp);
  temp = output[26] * cospi_20_64 + output[21] * cospi_12_64;
  step[26] = (tran_low_t)fdct_round_shift(temp);
  step[27] = output[27];
  step[28] = output[28];
  temp = output[29] * cospi_28_64 + output[18] * -cospi_4_64;
  step[29] = (tran_low_t)fdct_round_shift(temp);
  temp = output[30] * cospi_4_64 + output[17] * cospi_28_64;
  step[30] = (tran_low_t)fdct_round_shift(temp);
  step[31] = output[31];

597
  range_check(step, 32, 18);
598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692

  // stage 7
  output[0] = step[0];
  output[1] = step[1];
  output[2] = step[2];
  output[3] = step[3];
  output[4] = step[4];
  output[5] = step[5];
  output[6] = step[6];
  output[7] = step[7];
  temp = step[8] * cospi_30_64 + step[15] * cospi_2_64;
  output[8] = (tran_low_t)fdct_round_shift(temp);
  temp = step[9] * cospi_14_64 + step[14] * cospi_18_64;
  output[9] = (tran_low_t)fdct_round_shift(temp);
  temp = step[10] * cospi_22_64 + step[13] * cospi_10_64;
  output[10] = (tran_low_t)fdct_round_shift(temp);
  temp = step[11] * cospi_6_64 + step[12] * cospi_26_64;
  output[11] = (tran_low_t)fdct_round_shift(temp);
  temp = step[12] * cospi_6_64 + step[11] * -cospi_26_64;
  output[12] = (tran_low_t)fdct_round_shift(temp);
  temp = step[13] * cospi_22_64 + step[10] * -cospi_10_64;
  output[13] = (tran_low_t)fdct_round_shift(temp);
  temp = step[14] * cospi_14_64 + step[9] * -cospi_18_64;
  output[14] = (tran_low_t)fdct_round_shift(temp);
  temp = step[15] * cospi_30_64 + step[8] * -cospi_2_64;
  output[15] = (tran_low_t)fdct_round_shift(temp);
  output[16] = step[16] + step[17];
  output[17] = step[16] - step[17];
  output[18] = step[19] - step[18];
  output[19] = step[19] + step[18];
  output[20] = step[20] + step[21];
  output[21] = step[20] - step[21];
  output[22] = step[23] - step[22];
  output[23] = step[23] + step[22];
  output[24] = step[24] + step[25];
  output[25] = step[24] - step[25];
  output[26] = step[27] - step[26];
  output[27] = step[27] + step[26];
  output[28] = step[28] + step[29];
  output[29] = step[28] - step[29];
  output[30] = step[31] - step[30];
  output[31] = step[31] + step[30];

  range_check(output, 32, 18);

  // stage 8
  step[0] = output[0];
  step[1] = output[1];
  step[2] = output[2];
  step[3] = output[3];
  step[4] = output[4];
  step[5] = output[5];
  step[6] = output[6];
  step[7] = output[7];
  step[8] = output[8];
  step[9] = output[9];
  step[10] = output[10];
  step[11] = output[11];
  step[12] = output[12];
  step[13] = output[13];
  step[14] = output[14];
  step[15] = output[15];
  temp = output[16] * cospi_31_64 + output[31] * cospi_1_64;
  step[16] = (tran_low_t)fdct_round_shift(temp);
  temp = output[17] * cospi_15_64 + output[30] * cospi_17_64;
  step[17] = (tran_low_t)fdct_round_shift(temp);
  temp = output[18] * cospi_23_64 + output[29] * cospi_9_64;
  step[18] = (tran_low_t)fdct_round_shift(temp);
  temp = output[19] * cospi_7_64 + output[28] * cospi_25_64;
  step[19] = (tran_low_t)fdct_round_shift(temp);
  temp = output[20] * cospi_27_64 + output[27] * cospi_5_64;
  step[20] = (tran_low_t)fdct_round_shift(temp);
  temp = output[21] * cospi_11_64 + output[26] * cospi_21_64;
  step[21] = (tran_low_t)fdct_round_shift(temp);
  temp = output[22] * cospi_19_64 + output[25] * cospi_13_64;
  step[22] = (tran_low_t)fdct_round_shift(temp);
  temp = output[23] * cospi_3_64 + output[24] * cospi_29_64;
  step[23] = (tran_low_t)fdct_round_shift(temp);
  temp = output[24] * cospi_3_64 + output[23] * -cospi_29_64;
  step[24] = (tran_low_t)fdct_round_shift(temp);
  temp = output[25] * cospi_19_64 + output[22] * -cospi_13_64;
  step[25] = (tran_low_t)fdct_round_shift(temp);
  temp = output[26] * cospi_11_64 + output[21] * -cospi_21_64;
  step[26] = (tran_low_t)fdct_round_shift(temp);
  temp = output[27] * cospi_27_64 + output[20] * -cospi_5_64;
  step[27] = (tran_low_t)fdct_round_shift(temp);
  temp = output[28] * cospi_7_64 + output[19] * -cospi_25_64;
  step[28] = (tran_low_t)fdct_round_shift(temp);
  temp = output[29] * cospi_23_64 + output[18] * -cospi_9_64;
  step[29] = (tran_low_t)fdct_round_shift(temp);
  temp = output[30] * cospi_15_64 + output[17] * -cospi_17_64;
  step[30] = (tran_low_t)fdct_round_shift(temp);
  temp = output[31] * cospi_31_64 + output[16] * -cospi_1_64;
  step[31] = (tran_low_t)fdct_round_shift(temp);

693
  range_check(step, 32, 18);
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729

  // stage 9
  output[0] = step[0];
  output[1] = step[16];
  output[2] = step[8];
  output[3] = step[24];
  output[4] = step[4];
  output[5] = step[20];
  output[6] = step[12];
  output[7] = step[28];
  output[8] = step[2];
  output[9] = step[18];
  output[10] = step[10];
  output[11] = step[26];
  output[12] = step[6];
  output[13] = step[22];
  output[14] = step[14];
  output[15] = step[30];
  output[16] = step[1];
  output[17] = step[17];
  output[18] = step[9];
  output[19] = step[25];
  output[20] = step[5];
  output[21] = step[21];
  output[22] = step[13];
  output[23] = step[29];
  output[24] = step[3];
  output[25] = step[19];
  output[26] = step[11];
  output[27] = step[27];
  output[28] = step[7];
  output[29] = step[23];
  output[30] = step[15];
  output[31] = step[31];

  range_check(output, 32, 18);
Jingning Han's avatar
Jingning Han committed
730 731
}

732
#ifndef AV1_DCT_GTEST
Jingning Han's avatar
Jingning Han committed
733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
static void fadst4(const tran_low_t *input, tran_low_t *output) {
  tran_high_t x0, x1, x2, x3;
  tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;

  x0 = input[0];
  x1 = input[1];
  x2 = input[2];
  x3 = input[3];

  if (!(x0 | x1 | x2 | x3)) {
    output[0] = output[1] = output[2] = output[3] = 0;
    return;
  }

  s0 = sinpi_1_9 * x0;
  s1 = sinpi_4_9 * x0;
  s2 = sinpi_2_9 * x1;
  s3 = sinpi_1_9 * x1;
  s4 = sinpi_3_9 * x2;
  s5 = sinpi_4_9 * x3;
  s6 = sinpi_2_9 * x3;
  s7 = x0 + x1 - x3;

  x0 = s0 + s2 + s5;
  x1 = sinpi_3_9 * s7;
  x2 = s1 - s3 + s6;
  x3 = s4;

  s0 = x0 + x3;
  s1 = x1;
  s2 = x2 - x3;
  s3 = x2 - x0 + x3;

  // 1-D transform scaling factor is sqrt(2).
  output[0] = (tran_low_t)fdct_round_shift(s0);
  output[1] = (tran_low_t)fdct_round_shift(s1);
  output[2] = (tran_low_t)fdct_round_shift(s2);
  output[3] = (tran_low_t)fdct_round_shift(s3);
}

static void fadst8(const tran_low_t *input, tran_low_t *output) {
  tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;

  tran_high_t x0 = input[7];
  tran_high_t x1 = input[0];
  tran_high_t x2 = input[5];
  tran_high_t x3 = input[2];
  tran_high_t x4 = input[3];
  tran_high_t x5 = input[4];
  tran_high_t x6 = input[1];
  tran_high_t x7 = input[6];

  // stage 1
786 787
  s0 = cospi_2_64 * x0 + cospi_30_64 * x1;
  s1 = cospi_30_64 * x0 - cospi_2_64 * x1;
Jingning Han's avatar
Jingning Han committed
788 789 790 791
  s2 = cospi_10_64 * x2 + cospi_22_64 * x3;
  s3 = cospi_22_64 * x2 - cospi_10_64 * x3;
  s4 = cospi_18_64 * x4 + cospi_14_64 * x5;
  s5 = cospi_14_64 * x4 - cospi_18_64 * x5;
792 793
  s6 = cospi_26_64 * x6 + cospi_6_64 * x7;
  s7 = cospi_6_64 * x6 - cospi_26_64 * x7;
Jingning Han's avatar
Jingning Han committed
794

795 796 797 798
  x0 = s0 + s4;
  x1 = s1 + s5;
  x2 = s2 + s6;
  x3 = s3 + s7;
Jingning Han's avatar
Jingning Han committed
799 800 801 802 803 804 805 806 807 808
  x4 = fdct_round_shift(s0 - s4);
  x5 = fdct_round_shift(s1 - s5);
  x6 = fdct_round_shift(s2 - s6);
  x7 = fdct_round_shift(s3 - s7);

  // stage 2
  s0 = x0;
  s1 = x1;
  s2 = x2;
  s3 = x3;
809 810 811 812
  s4 = cospi_8_64 * x4 + cospi_24_64 * x5;
  s5 = cospi_24_64 * x4 - cospi_8_64 * x5;
  s6 = -cospi_24_64 * x6 + cospi_8_64 * x7;
  s7 = cospi_8_64 * x6 + cospi_24_64 * x7;
Jingning Han's avatar
Jingning Han committed
813

814 815 816 817
  x0 = fdct_round_shift(s0 + s2);
  x1 = fdct_round_shift(s1 + s3);
  x2 = fdct_round_shift(s0 - s2);
  x3 = fdct_round_shift(s1 - s3);
Jingning Han's avatar
Jingning Han committed
818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
  x4 = fdct_round_shift(s4 + s6);
  x5 = fdct_round_shift(s5 + s7);
  x6 = fdct_round_shift(s4 - s6);
  x7 = fdct_round_shift(s5 - s7);

  // stage 3
  s2 = cospi_16_64 * (x2 + x3);
  s3 = cospi_16_64 * (x2 - x3);
  s6 = cospi_16_64 * (x6 + x7);
  s7 = cospi_16_64 * (x6 - x7);

  x2 = fdct_round_shift(s2);
  x3 = fdct_round_shift(s3);
  x6 = fdct_round_shift(s6);
  x7 = fdct_round_shift(s7);

  output[0] = (tran_low_t)x0;
  output[1] = (tran_low_t)-x4;
  output[2] = (tran_low_t)x6;
  output[3] = (tran_low_t)-x2;
  output[4] = (tran_low_t)x3;
  output[5] = (tran_low_t)-x7;
  output[6] = (tran_low_t)x5;
  output[7] = (tran_low_t)-x1;
}

static void fadst16(const tran_low_t *input, tran_low_t *output) {
  tran_high_t s0, s1, s2, s3, s4, s5, s6, s7, s8;
  tran_high_t s9, s10, s11, s12, s13, s14, s15;

  tran_high_t x0 = input[15];
  tran_high_t x1 = input[0];
  tran_high_t x2 = input[13];
  tran_high_t x3 = input[2];
  tran_high_t x4 = input[11];
  tran_high_t x5 = input[4];
  tran_high_t x6 = input[9];
  tran_high_t x7 = input[6];
  tran_high_t x8 = input[7];
  tran_high_t x9 = input[8];
  tran_high_t x10 = input[5];
  tran_high_t x11 = input[10];
  tran_high_t x12 = input[3];
  tran_high_t x13 = input[12];
  tran_high_t x14 = input[1];
  tran_high_t x15 = input[14];

  // stage 1
866
  s0 = x0 * cospi_1_64 + x1 * cospi_31_64;
Jingning Han's avatar
Jingning Han committed
867
  s1 = x0 * cospi_31_64 - x1 * cospi_1_64;
868
  s2 = x2 * cospi_5_64 + x3 * cospi_27_64;
Jingning Han's avatar
Jingning Han committed
869
  s3 = x2 * cospi_27_64 - x3 * cospi_5_64;
870
  s4 = x4 * cospi_9_64 + x5 * cospi_23_64;
Jingning Han's avatar
Jingning Han committed
871 872 873 874 875 876 877 878
  s5 = x4 * cospi_23_64 - x5 * cospi_9_64;
  s6 = x6 * cospi_13_64 + x7 * cospi_19_64;
  s7 = x6 * cospi_19_64 - x7 * cospi_13_64;
  s8 = x8 * cospi_17_64 + x9 * cospi_15_64;
  s9 = x8 * cospi_15_64 - x9 * cospi_17_64;
  s10 = x10 * cospi_21_64 + x11 * cospi_11_64;
  s11 = x10 * cospi_11_64 - x11 * cospi_21_64;
  s12 = x12 * cospi_25_64 + x13 * cospi_7_64;
879
  s13 = x12 * cospi_7_64 - x13 * cospi_25_64;
Jingning Han's avatar
Jingning Han committed
880
  s14 = x14 * cospi_29_64 + x15 * cospi_3_64;
881
  s15 = x14 * cospi_3_64 - x15 * cospi_29_64;
Jingning Han's avatar
Jingning Han committed
882

883 884 885 886 887 888 889 890 891
  x0 = s0 + s8;
  x1 = s1 + s9;
  x2 = s2 + s10;
  x3 = s3 + s11;
  x4 = s4 + s12;
  x5 = s5 + s13;
  x6 = s6 + s14;
  x7 = s7 + s15;

892 893
  x8 = fdct_round_shift(s0 - s8);
  x9 = fdct_round_shift(s1 - s9);
Jingning Han's avatar
Jingning Han committed
894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909
  x10 = fdct_round_shift(s2 - s10);
  x11 = fdct_round_shift(s3 - s11);
  x12 = fdct_round_shift(s4 - s12);
  x13 = fdct_round_shift(s5 - s13);
  x14 = fdct_round_shift(s6 - s14);
  x15 = fdct_round_shift(s7 - s15);

  // stage 2
  s0 = x0;
  s1 = x1;
  s2 = x2;
  s3 = x3;
  s4 = x4;
  s5 = x5;
  s6 = x6;
  s7 = x7;
910 911 912 913 914 915 916 917
  s8 = x8 * cospi_4_64 + x9 * cospi_28_64;
  s9 = x8 * cospi_28_64 - x9 * cospi_4_64;
  s10 = x10 * cospi_20_64 + x11 * cospi_12_64;
  s11 = x10 * cospi_12_64 - x11 * cospi_20_64;
  s12 = -x12 * cospi_28_64 + x13 * cospi_4_64;
  s13 = x12 * cospi_4_64 + x13 * cospi_28_64;
  s14 = -x14 * cospi_12_64 + x15 * cospi_20_64;
  s15 = x14 * cospi_20_64 + x15 * cospi_12_64;
Jingning Han's avatar
Jingning Han committed
918 919 920 921 922

  x0 = s0 + s4;
  x1 = s1 + s5;
  x2 = s2 + s6;
  x3 = s3 + s7;
923 924 925 926 927 928 929 930 931
  x4 = fdct_round_shift(s0 - s4);
  x5 = fdct_round_shift(s1 - s5);
  x6 = fdct_round_shift(s2 - s6);
  x7 = fdct_round_shift(s3 - s7);

  x8 = s8 + s12;
  x9 = s9 + s13;
  x10 = s10 + s14;
  x11 = s11 + s15;
Jingning Han's avatar
Jingning Han committed
932 933 934 935 936 937 938 939 940 941
  x12 = fdct_round_shift(s8 - s12);
  x13 = fdct_round_shift(s9 - s13);
  x14 = fdct_round_shift(s10 - s14);
  x15 = fdct_round_shift(s11 - s15);

  // stage 3
  s0 = x0;
  s1 = x1;
  s2 = x2;
  s3 = x3;
942
  s4 = x4 * cospi_8_64 + x5 * cospi_24_64;
Jingning Han's avatar
Jingning Han committed
943
  s5 = x4 * cospi_24_64 - x5 * cospi_8_64;
944 945
  s6 = -x6 * cospi_24_64 + x7 * cospi_8_64;
  s7 = x6 * cospi_8_64 + x7 * cospi_24_64;
Jingning Han's avatar
Jingning Han committed
946 947 948 949
  s8 = x8;
  s9 = x9;
  s10 = x10;
  s11 = x11;
950
  s12 = x12 * cospi_8_64 + x13 * cospi_24_64;
Jingning Han's avatar
Jingning Han committed
951
  s13 = x12 * cospi_24_64 - x13 * cospi_8_64;
952 953
  s14 = -x14 * cospi_24_64 + x15 * cospi_8_64;
  s15 = x14 * cospi_8_64 + x15 * cospi_24_64;
Jingning Han's avatar
Jingning Han committed
954

955 956 957 958 959
  x0 = fdct_round_shift(s0 + s2);
  x1 = fdct_round_shift(s1 + s3);
  x2 = fdct_round_shift(s0 - s2);
  x3 = fdct_round_shift(s1 - s3);

Jingning Han's avatar
Jingning Han committed
960 961 962 963
  x4 = fdct_round_shift(s4 + s6);
  x5 = fdct_round_shift(s5 + s7);
  x6 = fdct_round_shift(s4 - s6);
  x7 = fdct_round_shift(s5 - s7);
964 965 966 967 968 969

  x8 = fdct_round_shift(s8 + s10);
  x9 = fdct_round_shift(s9 + s11);
  x10 = fdct_round_shift(s8 - s10);
  x11 = fdct_round_shift(s9 - s11);

Jingning Han's avatar
Jingning Han committed
970 971 972 973 974 975
  x12 = fdct_round_shift(s12 + s14);
  x13 = fdct_round_shift(s13 + s15);
  x14 = fdct_round_shift(s12 - s14);
  x15 = fdct_round_shift(s13 - s15);

  // stage 4
976
  s2 = (-cospi_16_64) * (x2 + x3);
Jingning Han's avatar
Jingning Han committed
977 978
  s3 = cospi_16_64 * (x2 - x3);
  s6 = cospi_16_64 * (x6 + x7);
979
  s7 = cospi_16_64 * (-x6 + x7);
Jingning Han's avatar
Jingning Han committed
980
  s10 = cospi_16_64 * (x10 + x11);
981 982
  s11 = cospi_16_64 * (-x10 + x11);
  s14 = (-cospi_16_64) * (x14 + x15);
Jingning Han's avatar
Jingning Han committed
983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
  s15 = cospi_16_64 * (x14 - x15);

  x2 = fdct_round_shift(s2);
  x3 = fdct_round_shift(s3);
  x6 = fdct_round_shift(s6);
  x7 = fdct_round_shift(s7);
  x10 = fdct_round_shift(s10);
  x11 = fdct_round_shift(s11);
  x14 = fdct_round_shift(s14);
  x15 = fdct_round_shift(s15);

  output[0] = (tran_low_t)x0;
  output[1] = (tran_low_t)-x8;
  output[2] = (tran_low_t)x12;
  output[3] = (tran_low_t)-x4;
  output[4] = (tran_low_t)x6;
  output[5] = (tran_low_t)x14;
  output[6] = (tran_low_t)x10;
  output[7] = (tran_low_t)x2;
  output[8] = (tran_low_t)x3;
  output[9] = (tran_low_t)x11;
  output[10] = (tran_low_t)x15;
  output[11] = (tran_low_t)x7;
  output[12] = (tran_low_t)x5;
  output[13] = (tran_low_t)-x13;
  output[14] = (tran_low_t)x9;
  output[15] = (tran_low_t)-x1;
}

1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026
// For use in lieu of ADST
static void fhalfright32(const tran_low_t *input, tran_low_t *output) {
  int i;
  tran_low_t inputhalf[16];
  for (i = 0; i < 16; ++i) {
    output[16 + i] = input[i] * 4;
  }
  // Multiply input by sqrt(2)
  for (i = 0; i < 16; ++i) {
    inputhalf[i] = (tran_low_t)fdct_round_shift(input[i + 16] * Sqrt2);
  }
  fdct16(inputhalf, output);
  // Note overall scaling factor is 4 times orthogonal
}

1027 1028 1029 1030
// TODO(sarahparker) these functions will be removed once the highbitdepth
// codepath works properly for rectangular transforms. They have almost
// identical versions in av1_fwd_txfm1d.c, but those are currently only
// being used for square transforms.
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1031 1032
static void fidtx4(const tran_low_t *input, tran_low_t *output) {
  int i;
1033
  for (i = 0; i < 4; ++i) {
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1034
    output[i] = (tran_low_t)fdct_round_shift(input[i] * Sqrt2);
1035
  }
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1036 1037 1038 1039
}

static void fidtx8(const tran_low_t *input, tran_low_t *output) {
  int i;
1040 1041 1042
  for (i = 0; i < 8; ++i) {
    output[i] = input[i] * 2;
  }
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1043 1044 1045 1046
}

static void fidtx16(const tran_low_t *input, tran_low_t *output) {
  int i;
1047
  for (i = 0; i < 16; ++i) {
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1048
    output[i] = (tran_low_t)fdct_round_shift(input[i] * 2 * Sqrt2);
1049
  }
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1050 1051 1052 1053
}

static void fidtx32(const tran_low_t *input, tran_low_t *output) {
  int i;
1054 1055 1056
  for (i = 0; i < 32; ++i) {
    output[i] = input[i] * 4;
  }
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1057 1058
}

1059
static void copy_block(const int16_t *src, int src_stride, int l, int w,
1060 1061 1062
                       int16_t *dest, int dest_stride) {
  int i;
  for (i = 0; i < l; ++i) {
1063
    memcpy(dest + dest_stride * i, src + src_stride * i, w * sizeof(int16_t));
1064 1065 1066
  }
}

1067
static void fliplr(int16_t *dest, int stride, int l, int w) {
1068 1069
  int i, j;
  for (i = 0; i < l; ++i) {
1070
    for (j = 0; j < w / 2; ++j) {
1071
      const int16_t tmp = dest[i * stride + j];
1072 1073
      dest[i * stride + j] = dest[i * stride + w - 1 - j];
      dest[i * stride + w - 1 - j] = tmp;
1074 1075 1076 1077
    }
  }
}

1078
static void flipud(int16_t *dest, int stride, int l, int w) {
1079
  int i, j;
1080
  for (j = 0; j < w; ++j) {
1081 1082 1083 1084 1085 1086 1087 1088
    for (i = 0; i < l / 2; ++i) {
      const int16_t tmp = dest[i * stride + j];
      dest[i * stride + j] = dest[(l - 1 - i) * stride + j];
      dest[(l - 1 - i) * stride + j] = tmp;
    }
  }
}

1089
static void fliplrud(int16_t *dest, int stride, int l, int w) {
1090 1091
  int i, j;
  for (i = 0; i < l / 2; ++i) {
1092
    for (j = 0; j < w; ++j) {
1093
      const int16_t tmp = dest[i * stride + j];
1094 1095
      dest[i * stride + j] = dest[(l - 1 - i) * stride + w - 1 - j];
      dest[(l - 1 - i) * stride + w - 1 - j] = tmp;
1096 1097 1098 1099
    }
  }
}

1100
static void copy_fliplr(const int16_t *src, int src_stride, int l, int w,
1101 1102 1103
                        int16_t *dest, int dest_stride) {
  copy_block(src, src_stride, l, w, dest, dest_stride);
  fliplr(dest, dest_stride, l, w);
1104 1105
}

1106
static void copy_flipud(const int16_t *src, int src_stride, int l, int w,
1107 1108 1109
                        int16_t *dest, int dest_stride) {
  copy_block(src, src_stride, l, w, dest, dest_stride);
  flipud(dest, dest_stride, l, w);
1110 1111
}

1112
static void copy_fliplrud(const int16_t *src, int src_stride, int l, int w,
1113 1114 1115
                          int16_t *dest, int dest_stride) {
  copy_block(src, src_stride, l, w, dest, dest_stride);
  fliplrud(dest, dest_stride, l, w);
1116 1117
}

1118
static void maybe_flip_input(const int16_t **src, int *src_stride, int l, int w,
1119
                             int16_t *buff, TX_TYPE tx_type) {
1120 1121 1122 1123 1124
  switch (tx_type) {
    case DCT_DCT:
    case ADST_DCT:
    case DCT_ADST:
    case ADST_ADST:
Debargha Mukherjee's avatar
Debargha Mukherjee committed
1125
    case IDTX:
Jingning Han's avatar
Jingning Han committed
1126
    case V_DCT:
1127 1128
    case H_DCT:
    case V_ADST:
1129
    case H_ADST: break;
1130 1131
    case FLIPADST_DCT:
    case FLIPADST_ADST:
1132
    case V_FLIPADST:
1133
      copy_flipud(*src, *src_stride, l, w, buff, w);
1134
      *src = buff;
1135
      *src_stride = w;
1136 1137 1138
      break;
    case DCT_FLIPADST:
    case ADST_FLIPADST:
1139
    case H_FLIPADST:
1140
      copy_fliplr(*src, *src_stride, l, w, buff, w);
1141
      *src = buff;
1142
      *src_stride = w;
1143 1144
      break;
    case FLIPADST_FLIPADST:
1145
      copy_fliplrud(*src, *src_stride, l, w, buff, w);
1146
      *src = buff;
1147
      *src_stride = w;
1148
      break;
1149
    default: assert(0); break;
1150 1151 1152
  }
}

Yaowu Xu's avatar
Yaowu Xu committed
1153
void av1_fht4x4_c(const int16_t *input, tran_low_t *output, int stride,
1154
                  TxfmParam *txfm_param) {
1155
  const TX_TYPE tx_type = txfm_param->tx_type;
1156
#if !CONFIG_DAALA_TX4
Jingning Han's avatar
Jingning Han committed
1157
  if (tx_type == DCT_DCT) {
Yaowu Xu's avatar
Yaowu Xu committed
1158
    aom_fdct4x4_c(input, output, stride);
1159 1160 1161 1162
    return;
  }
#endif
  {
Geza Lore's avatar
Geza Lore committed
1163
    static const transform_2d FHT[] = {
1164
#if CONFIG_DAALA_TX4
1165
      { daala_fdct4, daala_fdct4 },  // DCT_DCT
1166 1167 1168 1169 1170 1171 1172 1173
      { daala_fdst4, daala_fdct4 },  // ADST_DCT
      { daala_fdct4, daala_fdst4 },  // DCT_ADST
      { daala_fdst4, daala_fdst4 },  // ADST_ADST
      { daala_fdst4, daala_fdct4 },  // FLIPADST_DCT
      { daala_fdct4, daala_fdst4 },  // DCT_FLIPADST
      { daala_fdst4, daala_fdst4 },  // FLIPADST_FLIPADST
      { daala_fdst4, daala_fdst4 },  // ADST_FLIPADST
      { daala_fdst4, daala_fdst4 },  // FLIPADST_ADST
1174 1175 1176 1177 1178 1179 1180
      { daala_idtx4, daala_idtx4 },  // IDTX
      { daala_fdct4, daala_idtx4 },  // V_DCT
      { daala_idtx4, daala_fdct4 },  // H_DCT
      { daala_fdst4, daala_idtx4 },  // V_ADST
      { daala_idtx4, daala_fdst4 },  // H_ADST
      { daala_fdst4, daala_idtx4 },  // V_FLIPADST
      { daala_idtx4, daala_fdst4 },  // H_FLIPADST
1181
#else
Geza Lore's avatar
Geza Lore committed
1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
      { fdct4, fdct4 },    // DCT_DCT
      { fadst4, fdct4 },   // ADST_DCT
      { fdct4, fadst4 },   // DCT_ADST
      { fadst4, fadst4 },  // ADST_ADST
      { fadst4, fdct4 },   // FLIPADST_DCT
      { fdct4, fadst4 },   // DCT_FLIPADST
      { fadst4, fadst4 },  // FLIPADST_FLIPADST
      { fadst4, fadst4 },  // ADST_FLIPADST
      { fadst4, fadst4 },  // FLIPADST_ADST
      { fidtx4, fidtx4 },  // IDTX
      { fdct4, fidtx4 },   // V_DCT
      { fidtx4, fdct4 },   // H_DCT
      { fadst4, fidtx4 },  // V_ADST
      { fidtx4, fadst4 },  // H_ADST
      { fadst4, fidtx4 },  // V_FLIPADST
      { fidtx4, fadst4 },  // H_FLIPADST
Lester Lu's avatar
Lester Lu committed
1198
#endif
Geza Lore's avatar
Geza Lore committed
1199 1200
    };
    const transform_2d ht = FHT[tx_type];
Jingning Han's avatar
Jingning Han committed
1201 1202 1203 1204
    tran_low_t out[4 * 4];
    int i, j;
    tran_low_t temp_in[4], temp_out[4];

1205
    int16_t flipped_input[4 * 4];
1206
    maybe_flip_input(&input, &stride, 4, 4, flipped_input, tx_type);
1207

Jingning Han's avatar
Jingning Han committed
1208 1209
    // Columns
    for (i = 0; i < 4; ++i) {
1210
      /* A C99-safe upshift by 4 for both Daala and VPx TX. */
1211
      for (j = 0; j < 4; ++j) temp_in[j] = input[j * stride + i] * 16;
1212
#if !CONFIG_DAALA_TX4