| 1 | /***************************************************************************** |
| 2 | * Copyright (C) 2013 x265 project |
| 3 | * |
| 4 | * Authors: Deepthi Devaki <deepthidevaki@multicorewareinc.com>, |
| 5 | * Rajesh Paulraj <rajesh@multicorewareinc.com> |
| 6 | * Praveen Kumar Tiwari <praveen@multicorewareinc.com> |
| 7 | * Min Chen <chenm003@163.com> <min.chen@multicorewareinc.com> |
| 8 | * |
| 9 | * This program is free software; you can redistribute it and/or modify |
| 10 | * it under the terms of the GNU General Public License as published by |
| 11 | * the Free Software Foundation; either version 2 of the License, or |
| 12 | * (at your option) any later version. |
| 13 | * |
| 14 | * This program is distributed in the hope that it will be useful, |
| 15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 17 | * GNU General Public License for more details. |
| 18 | * |
| 19 | * You should have received a copy of the GNU General Public License |
| 20 | * along with this program; if not, write to the Free Software |
| 21 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA. |
| 22 | * |
| 23 | * This program is also available under a commercial proprietary license. |
| 24 | * For more information, contact us at license @ x265.com. |
| 25 | *****************************************************************************/ |
| 26 | |
| 27 | #include "common.h" |
| 28 | #include "ipfilterharness.h" |
| 29 | |
| 30 | using namespace x265; |
| 31 | |
| 32 | IPFilterHarness::IPFilterHarness() |
| 33 | { |
| 34 | /* [0] --- Random values |
| 35 | * [1] --- Minimum |
| 36 | * [2] --- Maximum */ |
| 37 | for (int i = 0; i < TEST_BUF_SIZE; i++) |
| 38 | { |
| 39 | pixel_test_buff[0][i] = rand() & PIXEL_MAX; |
| 40 | short_test_buff[0][i] = (rand() % (2 * SMAX)) - SMAX; |
| 41 | |
| 42 | pixel_test_buff[1][i] = PIXEL_MIN; |
| 43 | short_test_buff[1][i] = SMIN; |
| 44 | |
| 45 | pixel_test_buff[2][i] = PIXEL_MAX; |
| 46 | short_test_buff[2][i] = SMAX; |
| 47 | } |
| 48 | |
| 49 | memset(IPF_C_output_p, 0xCD, TEST_BUF_SIZE * sizeof(pixel)); |
| 50 | memset(IPF_vec_output_p, 0xCD, TEST_BUF_SIZE * sizeof(pixel)); |
| 51 | memset(IPF_C_output_s, 0xCD, TEST_BUF_SIZE * sizeof(int16_t)); |
| 52 | memset(IPF_vec_output_s, 0xCD, TEST_BUF_SIZE * sizeof(int16_t)); |
| 53 | |
| 54 | int pixelMax = (1 << X265_DEPTH) - 1; |
| 55 | int shortMax = (1 << 15) - 1; |
| 56 | for (int i = 0; i < TEST_BUF_SIZE; i++) |
| 57 | { |
| 58 | pixel_buff[i] = (pixel)(rand() & pixelMax); |
| 59 | int isPositive = (rand() & 1) ? 1 : -1; |
| 60 | short_buff[i] = (int16_t)(isPositive * (rand() & shortMax)); |
| 61 | } |
| 62 | } |
| 63 | |
| 64 | bool IPFilterHarness::check_IPFilter_primitive(filter_p2s_t ref, filter_p2s_t opt, int isChroma, int csp) |
| 65 | { |
| 66 | intptr_t rand_srcStride; |
| 67 | int min_size = isChroma ? 2 : 4; |
| 68 | int max_size = isChroma ? (MAX_CU_SIZE >> 1) : MAX_CU_SIZE; |
| 69 | |
| 70 | if (isChroma && (csp == X265_CSP_I444)) |
| 71 | { |
| 72 | min_size = 4; |
| 73 | max_size = MAX_CU_SIZE; |
| 74 | } |
| 75 | |
| 76 | for (int i = 0; i < ITERS; i++) |
| 77 | { |
| 78 | int index = i % TEST_CASES; |
| 79 | int rand_height = (int16_t)rand() % 100; |
| 80 | int rand_width = (int16_t)rand() % 100; |
| 81 | |
| 82 | rand_srcStride = rand_width + rand() % 100; |
| 83 | if (rand_srcStride < rand_width) |
| 84 | rand_srcStride = rand_width; |
| 85 | |
| 86 | rand_width &= ~(min_size - 1); |
| 87 | rand_width = Clip3(min_size, max_size, rand_width); |
| 88 | |
| 89 | rand_height &= ~(min_size - 1); |
| 90 | rand_height = Clip3(min_size, max_size, rand_height); |
| 91 | |
| 92 | ref(pixel_test_buff[index], |
| 93 | rand_srcStride, |
| 94 | IPF_C_output_s, |
| 95 | rand_width, |
| 96 | rand_height); |
| 97 | |
| 98 | checked(opt, pixel_test_buff[index], |
| 99 | rand_srcStride, |
| 100 | IPF_vec_output_s, |
| 101 | rand_width, |
| 102 | rand_height); |
| 103 | |
| 104 | if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 105 | return false; |
| 106 | |
| 107 | reportfail(); |
| 108 | } |
| 109 | |
| 110 | return true; |
| 111 | } |
| 112 | |
| 113 | bool IPFilterHarness::check_IPFilterChroma_primitive(filter_pp_t ref, filter_pp_t opt) |
| 114 | { |
| 115 | intptr_t rand_srcStride, rand_dstStride; |
| 116 | |
| 117 | for (int i = 0; i < ITERS; i++) |
| 118 | { |
| 119 | int index = i % TEST_CASES; |
| 120 | |
| 121 | for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++) |
| 122 | { |
| 123 | rand_srcStride = rand() % 100 + 2; |
| 124 | rand_dstStride = rand() % 100 + 64; |
| 125 | |
| 126 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride, |
| 127 | rand_srcStride, |
| 128 | IPF_vec_output_p, |
| 129 | rand_dstStride, |
| 130 | coeffIdx); |
| 131 | |
| 132 | ref(pixel_test_buff[index] + 3 * rand_srcStride, |
| 133 | rand_srcStride, |
| 134 | IPF_C_output_p, |
| 135 | rand_dstStride, |
| 136 | coeffIdx); |
| 137 | |
| 138 | if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel))) |
| 139 | return false; |
| 140 | |
| 141 | reportfail(); |
| 142 | } |
| 143 | } |
| 144 | |
| 145 | return true; |
| 146 | } |
| 147 | |
| 148 | bool IPFilterHarness::check_IPFilterChroma_ps_primitive(filter_ps_t ref, filter_ps_t opt) |
| 149 | { |
| 150 | intptr_t rand_srcStride, rand_dstStride; |
| 151 | |
| 152 | for (int i = 0; i < ITERS; i++) |
| 153 | { |
| 154 | int index = i % TEST_CASES; |
| 155 | |
| 156 | for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++) |
| 157 | { |
| 158 | rand_srcStride = rand() % 100; |
| 159 | rand_dstStride = rand() % 100 + 64; |
| 160 | |
| 161 | ref(pixel_test_buff[index] + 3 * rand_srcStride, |
| 162 | rand_srcStride, |
| 163 | IPF_C_output_s, |
| 164 | rand_dstStride, |
| 165 | coeffIdx); |
| 166 | |
| 167 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride, |
| 168 | rand_srcStride, |
| 169 | IPF_vec_output_s, |
| 170 | rand_dstStride, |
| 171 | coeffIdx); |
| 172 | |
| 173 | if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 174 | return false; |
| 175 | |
| 176 | reportfail(); |
| 177 | } |
| 178 | } |
| 179 | |
| 180 | return true; |
| 181 | } |
| 182 | |
| 183 | bool IPFilterHarness::check_IPFilterChroma_hps_primitive(filter_hps_t ref, filter_hps_t opt) |
| 184 | { |
| 185 | intptr_t rand_srcStride, rand_dstStride; |
| 186 | |
| 187 | for (int i = 0; i < ITERS; i++) |
| 188 | { |
| 189 | int index = i % TEST_CASES; |
| 190 | |
| 191 | for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++) |
| 192 | { |
| 193 | // 0 : Interpolate W x H, 1 : Interpolate W x (H + 7) |
| 194 | for (int isRowExt = 0; isRowExt < 2; isRowExt++) |
| 195 | { |
| 196 | rand_srcStride = rand() % 100 + 2; |
| 197 | rand_dstStride = rand() % 100 + 64; |
| 198 | |
| 199 | ref(pixel_test_buff[index] + 3 * rand_srcStride, |
| 200 | rand_srcStride, |
| 201 | IPF_C_output_s, |
| 202 | rand_dstStride, |
| 203 | coeffIdx, |
| 204 | isRowExt); |
| 205 | |
| 206 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride, |
| 207 | rand_srcStride, |
| 208 | IPF_vec_output_s, |
| 209 | rand_dstStride, |
| 210 | coeffIdx, |
| 211 | isRowExt); |
| 212 | |
| 213 | if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 214 | return false; |
| 215 | |
| 216 | reportfail(); |
| 217 | } |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | return true; |
| 222 | } |
| 223 | |
| 224 | bool IPFilterHarness::check_IPFilterChroma_sp_primitive(filter_sp_t ref, filter_sp_t opt) |
| 225 | { |
| 226 | intptr_t rand_srcStride, rand_dstStride; |
| 227 | |
| 228 | for (int i = 0; i < ITERS; i++) |
| 229 | { |
| 230 | int index = i % TEST_CASES; |
| 231 | |
| 232 | for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++) |
| 233 | { |
| 234 | rand_srcStride = rand() % 100; |
| 235 | rand_dstStride = rand() % 100 + 64; |
| 236 | |
| 237 | ref(short_test_buff[index] + 3 * rand_srcStride, |
| 238 | rand_srcStride, |
| 239 | IPF_C_output_p, |
| 240 | rand_dstStride, |
| 241 | coeffIdx); |
| 242 | |
| 243 | checked(opt, short_test_buff[index] + 3 * rand_srcStride, |
| 244 | rand_srcStride, |
| 245 | IPF_vec_output_p, |
| 246 | rand_dstStride, |
| 247 | coeffIdx); |
| 248 | |
| 249 | if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel))) |
| 250 | return false; |
| 251 | |
| 252 | reportfail(); |
| 253 | } |
| 254 | } |
| 255 | |
| 256 | return true; |
| 257 | } |
| 258 | |
| 259 | bool IPFilterHarness::check_IPFilterChroma_ss_primitive(filter_ss_t ref, filter_ss_t opt) |
| 260 | { |
| 261 | intptr_t rand_srcStride, rand_dstStride; |
| 262 | |
| 263 | for (int i = 0; i < ITERS; i++) |
| 264 | { |
| 265 | int index = i % TEST_CASES; |
| 266 | |
| 267 | for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++) |
| 268 | { |
| 269 | rand_srcStride = rand() % 100; |
| 270 | rand_dstStride = rand() % 100 + 64; |
| 271 | |
| 272 | ref(short_test_buff[index] + 3 * rand_srcStride, |
| 273 | rand_srcStride, |
| 274 | IPF_C_output_s, |
| 275 | rand_dstStride, |
| 276 | coeffIdx); |
| 277 | |
| 278 | checked(opt, short_test_buff[index] + 3 * rand_srcStride, |
| 279 | rand_srcStride, |
| 280 | IPF_vec_output_s, |
| 281 | rand_dstStride, |
| 282 | coeffIdx); |
| 283 | |
| 284 | if (memcmp(IPF_C_output_s, IPF_vec_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 285 | return false; |
| 286 | |
| 287 | reportfail(); |
| 288 | } |
| 289 | } |
| 290 | |
| 291 | return true; |
| 292 | } |
| 293 | |
| 294 | bool IPFilterHarness::check_IPFilterLuma_primitive(filter_pp_t ref, filter_pp_t opt) |
| 295 | { |
| 296 | intptr_t rand_srcStride, rand_dstStride; |
| 297 | |
| 298 | for (int i = 0; i < ITERS; i++) |
| 299 | { |
| 300 | int index = i % TEST_CASES; |
| 301 | |
| 302 | for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++) |
| 303 | { |
| 304 | rand_srcStride = rand() % 100; |
| 305 | rand_dstStride = rand() % 100 + 64; |
| 306 | |
| 307 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride + 6, |
| 308 | rand_srcStride, |
| 309 | IPF_vec_output_p, |
| 310 | rand_dstStride, |
| 311 | coeffIdx); |
| 312 | |
| 313 | ref(pixel_test_buff[index] + 3 * rand_srcStride + 6, |
| 314 | rand_srcStride, |
| 315 | IPF_C_output_p, |
| 316 | rand_dstStride, |
| 317 | coeffIdx); |
| 318 | |
| 319 | if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE)) |
| 320 | return false; |
| 321 | |
| 322 | reportfail(); |
| 323 | } |
| 324 | } |
| 325 | |
| 326 | return true; |
| 327 | } |
| 328 | |
| 329 | bool IPFilterHarness::check_IPFilterLuma_ps_primitive(filter_ps_t ref, filter_ps_t opt) |
| 330 | { |
| 331 | intptr_t rand_srcStride, rand_dstStride; |
| 332 | |
| 333 | for (int i = 0; i < ITERS; i++) |
| 334 | { |
| 335 | int index = i % TEST_CASES; |
| 336 | |
| 337 | for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++) |
| 338 | { |
| 339 | rand_srcStride = rand() % 100; |
| 340 | rand_dstStride = rand() % 100 + 64; |
| 341 | |
| 342 | ref(pixel_test_buff[index] + 3 * rand_srcStride, |
| 343 | rand_srcStride, |
| 344 | IPF_C_output_s, |
| 345 | rand_dstStride, |
| 346 | coeffIdx); |
| 347 | |
| 348 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride, |
| 349 | rand_srcStride, |
| 350 | IPF_vec_output_s, |
| 351 | rand_dstStride, |
| 352 | coeffIdx); |
| 353 | |
| 354 | if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 355 | return false; |
| 356 | |
| 357 | reportfail(); |
| 358 | } |
| 359 | } |
| 360 | |
| 361 | return true; |
| 362 | } |
| 363 | |
| 364 | bool IPFilterHarness::check_IPFilterLuma_hps_primitive(filter_hps_t ref, filter_hps_t opt) |
| 365 | { |
| 366 | intptr_t rand_srcStride, rand_dstStride; |
| 367 | |
| 368 | for (int i = 0; i < ITERS; i++) |
| 369 | { |
| 370 | int index = i % TEST_CASES; |
| 371 | |
| 372 | for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++) |
| 373 | { |
| 374 | // 0 : Interpolate W x H, 1 : Interpolate W x (H + 7) |
| 375 | for (int isRowExt = 0; isRowExt < 2; isRowExt++) |
| 376 | { |
| 377 | rand_srcStride = rand() % 100; |
| 378 | rand_dstStride = rand() % 100 + 64; |
| 379 | |
| 380 | ref(pixel_test_buff[index] + 3 * rand_srcStride + 6, |
| 381 | rand_srcStride, |
| 382 | IPF_C_output_s, |
| 383 | rand_dstStride, |
| 384 | coeffIdx, |
| 385 | isRowExt); |
| 386 | |
| 387 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride + 6, |
| 388 | rand_srcStride, |
| 389 | IPF_vec_output_s, |
| 390 | rand_dstStride, |
| 391 | coeffIdx, |
| 392 | isRowExt); |
| 393 | |
| 394 | if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 395 | return false; |
| 396 | |
| 397 | reportfail(); |
| 398 | } |
| 399 | } |
| 400 | } |
| 401 | |
| 402 | return true; |
| 403 | } |
| 404 | |
| 405 | bool IPFilterHarness::check_IPFilterLuma_sp_primitive(filter_sp_t ref, filter_sp_t opt) |
| 406 | { |
| 407 | intptr_t rand_srcStride, rand_dstStride; |
| 408 | |
| 409 | for (int i = 0; i < ITERS; i++) |
| 410 | { |
| 411 | int index = i % TEST_CASES; |
| 412 | |
| 413 | for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++) |
| 414 | { |
| 415 | rand_srcStride = rand() % 100; |
| 416 | rand_dstStride = rand() % 100 + 64; |
| 417 | |
| 418 | ref(short_test_buff[index] + 3 * rand_srcStride, |
| 419 | rand_srcStride, |
| 420 | IPF_C_output_p, |
| 421 | rand_dstStride, |
| 422 | coeffIdx); |
| 423 | |
| 424 | checked(opt, short_test_buff[index] + 3 * rand_srcStride, |
| 425 | rand_srcStride, |
| 426 | IPF_vec_output_p, |
| 427 | rand_dstStride, |
| 428 | coeffIdx); |
| 429 | |
| 430 | if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel))) |
| 431 | return false; |
| 432 | |
| 433 | reportfail(); |
| 434 | } |
| 435 | } |
| 436 | |
| 437 | return true; |
| 438 | } |
| 439 | |
| 440 | bool IPFilterHarness::check_IPFilterLuma_ss_primitive(filter_ss_t ref, filter_ss_t opt) |
| 441 | { |
| 442 | intptr_t rand_srcStride, rand_dstStride; |
| 443 | |
| 444 | for (int i = 0; i < ITERS; i++) |
| 445 | { |
| 446 | int index = i % TEST_CASES; |
| 447 | |
| 448 | for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++) |
| 449 | { |
| 450 | rand_srcStride = rand() % 100; |
| 451 | rand_dstStride = rand() % 100 + 64; |
| 452 | |
| 453 | ref(short_test_buff[index] + 3 * rand_srcStride, |
| 454 | rand_srcStride, |
| 455 | IPF_C_output_s, |
| 456 | rand_dstStride, |
| 457 | coeffIdx); |
| 458 | |
| 459 | checked(opt, short_test_buff[index] + 3 * rand_srcStride, |
| 460 | rand_srcStride, |
| 461 | IPF_vec_output_s, |
| 462 | rand_dstStride, |
| 463 | coeffIdx); |
| 464 | |
| 465 | if (memcmp(IPF_C_output_s, IPF_vec_output_s, TEST_BUF_SIZE * sizeof(int16_t))) |
| 466 | return false; |
| 467 | |
| 468 | reportfail(); |
| 469 | } |
| 470 | } |
| 471 | |
| 472 | return true; |
| 473 | } |
| 474 | |
| 475 | bool IPFilterHarness::check_IPFilterLumaHV_primitive(filter_hv_pp_t ref, filter_hv_pp_t opt) |
| 476 | { |
| 477 | intptr_t rand_srcStride, rand_dstStride; |
| 478 | |
| 479 | for (int i = 0; i < ITERS; i++) |
| 480 | { |
| 481 | int index = i % TEST_CASES; |
| 482 | |
| 483 | for (int coeffIdxX = 0; coeffIdxX < 4; coeffIdxX++) |
| 484 | { |
| 485 | for (int coeffIdxY = 0; coeffIdxY < 4; coeffIdxY++) |
| 486 | { |
| 487 | rand_srcStride = rand() % 100; |
| 488 | rand_dstStride = rand() % 100 + 64; |
| 489 | |
| 490 | ref(pixel_test_buff[index] + 3 * rand_srcStride, |
| 491 | rand_srcStride, |
| 492 | IPF_C_output_p, |
| 493 | rand_dstStride, |
| 494 | coeffIdxX, |
| 495 | coeffIdxY); |
| 496 | |
| 497 | checked(opt, pixel_test_buff[index] + 3 * rand_srcStride, |
| 498 | rand_srcStride, |
| 499 | IPF_vec_output_p, |
| 500 | rand_dstStride, |
| 501 | coeffIdxX, |
| 502 | coeffIdxY); |
| 503 | |
| 504 | if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel))) |
| 505 | return false; |
| 506 | |
| 507 | reportfail(); |
| 508 | } |
| 509 | } |
| 510 | } |
| 511 | |
| 512 | return true; |
| 513 | } |
| 514 | |
| 515 | bool IPFilterHarness::testCorrectness(const EncoderPrimitives& ref, const EncoderPrimitives& opt) |
| 516 | { |
| 517 | if (opt.luma_p2s) |
| 518 | { |
| 519 | // last parameter does not matter in case of luma |
| 520 | if (!check_IPFilter_primitive(ref.luma_p2s, opt.luma_p2s, 0, 1)) |
| 521 | { |
| 522 | printf("luma_p2s failed\n"); |
| 523 | return false; |
| 524 | } |
| 525 | } |
| 526 | |
| 527 | for (int value = 0; value < NUM_LUMA_PARTITIONS; value++) |
| 528 | { |
| 529 | if (opt.luma_hpp[value]) |
| 530 | { |
| 531 | if (!check_IPFilterLuma_primitive(ref.luma_hpp[value], opt.luma_hpp[value])) |
| 532 | { |
| 533 | printf("luma_hpp[%s]", lumaPartStr[value]); |
| 534 | return false; |
| 535 | } |
| 536 | } |
| 537 | if (opt.luma_hps[value]) |
| 538 | { |
| 539 | if (!check_IPFilterLuma_hps_primitive(ref.luma_hps[value], opt.luma_hps[value])) |
| 540 | { |
| 541 | printf("luma_hps[%s]", lumaPartStr[value]); |
| 542 | return false; |
| 543 | } |
| 544 | } |
| 545 | if (opt.luma_vpp[value]) |
| 546 | { |
| 547 | if (!check_IPFilterLuma_primitive(ref.luma_vpp[value], opt.luma_vpp[value])) |
| 548 | { |
| 549 | printf("luma_vpp[%s]", lumaPartStr[value]); |
| 550 | return false; |
| 551 | } |
| 552 | } |
| 553 | if (opt.luma_vps[value]) |
| 554 | { |
| 555 | if (!check_IPFilterLuma_ps_primitive(ref.luma_vps[value], opt.luma_vps[value])) |
| 556 | { |
| 557 | printf("luma_vps[%s]", lumaPartStr[value]); |
| 558 | return false; |
| 559 | } |
| 560 | } |
| 561 | if (opt.luma_vsp[value]) |
| 562 | { |
| 563 | if (!check_IPFilterLuma_sp_primitive(ref.luma_vsp[value], opt.luma_vsp[value])) |
| 564 | { |
| 565 | printf("luma_vsp[%s]", lumaPartStr[value]); |
| 566 | return false; |
| 567 | } |
| 568 | } |
| 569 | if (opt.luma_vss[value]) |
| 570 | { |
| 571 | if (!check_IPFilterLuma_ss_primitive(ref.luma_vss[value], opt.luma_vss[value])) |
| 572 | { |
| 573 | printf("luma_vss[%s]", lumaPartStr[value]); |
| 574 | return false; |
| 575 | } |
| 576 | } |
| 577 | if (opt.luma_hvpp[value]) |
| 578 | { |
| 579 | if (!check_IPFilterLumaHV_primitive(ref.luma_hvpp[value], opt.luma_hvpp[value])) |
| 580 | { |
| 581 | printf("luma_hvpp[%s]", lumaPartStr[value]); |
| 582 | return false; |
| 583 | } |
| 584 | } |
| 585 | } |
| 586 | |
| 587 | for (int csp = X265_CSP_I420; csp < X265_CSP_COUNT; csp++) |
| 588 | { |
| 589 | if (opt.chroma[csp].p2s) |
| 590 | { |
| 591 | if (!check_IPFilter_primitive(ref.chroma[csp].p2s, opt.chroma[csp].p2s, 1, csp)) |
| 592 | { |
| 593 | printf("chroma_p2s[%s]", x265_source_csp_names[csp]); |
| 594 | return false; |
| 595 | } |
| 596 | } |
| 597 | for (int value = 0; value < NUM_CHROMA_PARTITIONS; value++) |
| 598 | { |
| 599 | if (opt.chroma[csp].filter_hpp[value]) |
| 600 | { |
| 601 | if (!check_IPFilterChroma_primitive(ref.chroma[csp].filter_hpp[value], opt.chroma[csp].filter_hpp[value])) |
| 602 | { |
| 603 | printf("chroma_hpp[%s]", chromaPartStr[csp][value]); |
| 604 | return false; |
| 605 | } |
| 606 | } |
| 607 | if (opt.chroma[csp].filter_hps[value]) |
| 608 | { |
| 609 | if (!check_IPFilterChroma_hps_primitive(ref.chroma[csp].filter_hps[value], opt.chroma[csp].filter_hps[value])) |
| 610 | { |
| 611 | printf("chroma_hps[%s]", chromaPartStr[csp][value]); |
| 612 | return false; |
| 613 | } |
| 614 | } |
| 615 | if (opt.chroma[csp].filter_vpp[value]) |
| 616 | { |
| 617 | if (!check_IPFilterChroma_primitive(ref.chroma[csp].filter_vpp[value], opt.chroma[csp].filter_vpp[value])) |
| 618 | { |
| 619 | printf("chroma_vpp[%s]", chromaPartStr[csp][value]); |
| 620 | return false; |
| 621 | } |
| 622 | } |
| 623 | if (opt.chroma[csp].filter_vps[value]) |
| 624 | { |
| 625 | if (!check_IPFilterChroma_ps_primitive(ref.chroma[csp].filter_vps[value], opt.chroma[csp].filter_vps[value])) |
| 626 | { |
| 627 | printf("chroma_vps[%s]", chromaPartStr[csp][value]); |
| 628 | return false; |
| 629 | } |
| 630 | } |
| 631 | if (opt.chroma[csp].filter_vsp[value]) |
| 632 | { |
| 633 | if (!check_IPFilterChroma_sp_primitive(ref.chroma[csp].filter_vsp[value], opt.chroma[csp].filter_vsp[value])) |
| 634 | { |
| 635 | printf("chroma_vsp[%s]", chromaPartStr[csp][value]); |
| 636 | return false; |
| 637 | } |
| 638 | } |
| 639 | if (opt.chroma[csp].filter_vss[value]) |
| 640 | { |
| 641 | if (!check_IPFilterChroma_ss_primitive(ref.chroma[csp].filter_vss[value], opt.chroma[csp].filter_vss[value])) |
| 642 | { |
| 643 | printf("chroma_vss[%s]", chromaPartStr[csp][value]); |
| 644 | return false; |
| 645 | } |
| 646 | } |
| 647 | } |
| 648 | } |
| 649 | |
| 650 | return true; |
| 651 | } |
| 652 | |
| 653 | void IPFilterHarness::measureSpeed(const EncoderPrimitives& ref, const EncoderPrimitives& opt) |
| 654 | { |
| 655 | int height = 64; |
| 656 | int width = 64; |
| 657 | int16_t srcStride = 96; |
| 658 | int16_t dstStride = 96; |
| 659 | int maxVerticalfilterHalfDistance = 3; |
| 660 | |
| 661 | if (opt.luma_p2s) |
| 662 | { |
| 663 | printf("luma_p2s\t"); |
| 664 | REPORT_SPEEDUP(opt.luma_p2s, ref.luma_p2s, |
| 665 | pixel_buff, srcStride, IPF_vec_output_s, width, height); |
| 666 | } |
| 667 | |
| 668 | for (int value = 0; value < NUM_LUMA_PARTITIONS; value++) |
| 669 | { |
| 670 | if (opt.luma_hpp[value]) |
| 671 | { |
| 672 | printf("luma_hpp[%s]\t", lumaPartStr[value]); |
| 673 | REPORT_SPEEDUP(opt.luma_hpp[value], ref.luma_hpp[value], |
| 674 | pixel_buff + srcStride, srcStride, IPF_vec_output_p, dstStride, 1); |
| 675 | } |
| 676 | |
| 677 | if (opt.luma_hps[value]) |
| 678 | { |
| 679 | printf("luma_hps[%s]\t", lumaPartStr[value]); |
| 680 | REPORT_SPEEDUP(opt.luma_hps[value], ref.luma_hps[value], |
| 681 | pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 682 | IPF_vec_output_s, dstStride, 1, 1); |
| 683 | } |
| 684 | |
| 685 | if (opt.luma_vpp[value]) |
| 686 | { |
| 687 | printf("luma_vpp[%s]\t", lumaPartStr[value]); |
| 688 | REPORT_SPEEDUP(opt.luma_vpp[value], ref.luma_vpp[value], |
| 689 | pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 690 | IPF_vec_output_p, dstStride, 1); |
| 691 | } |
| 692 | |
| 693 | if (opt.luma_vps[value]) |
| 694 | { |
| 695 | printf("luma_vps[%s]\t", lumaPartStr[value]); |
| 696 | REPORT_SPEEDUP(opt.luma_vps[value], ref.luma_vps[value], |
| 697 | pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 698 | IPF_vec_output_s, dstStride, 1); |
| 699 | } |
| 700 | |
| 701 | if (opt.luma_vsp[value]) |
| 702 | { |
| 703 | printf("luma_vsp[%s]\t", lumaPartStr[value]); |
| 704 | REPORT_SPEEDUP(opt.luma_vsp[value], ref.luma_vsp[value], |
| 705 | short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 706 | IPF_vec_output_p, dstStride, 1); |
| 707 | } |
| 708 | |
| 709 | if (opt.luma_vss[value]) |
| 710 | { |
| 711 | printf("luma_vss[%s]\t", lumaPartStr[value]); |
| 712 | REPORT_SPEEDUP(opt.luma_vss[value], ref.luma_vss[value], |
| 713 | short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 714 | IPF_vec_output_s, dstStride, 1); |
| 715 | } |
| 716 | |
| 717 | if (opt.luma_hvpp[value]) |
| 718 | { |
| 719 | printf("luma_hv [%s]\t", lumaPartStr[value]); |
| 720 | REPORT_SPEEDUP(opt.luma_hvpp[value], ref.luma_hvpp[value], |
| 721 | pixel_buff + 3 * srcStride, srcStride, IPF_vec_output_p, srcStride, 1, 3); |
| 722 | } |
| 723 | } |
| 724 | |
| 725 | for (int csp = X265_CSP_I420; csp < X265_CSP_COUNT; csp++) |
| 726 | { |
| 727 | printf("= Color Space %s =\n", x265_source_csp_names[csp]); |
| 728 | if (opt.chroma[csp].p2s) |
| 729 | { |
| 730 | printf("chroma_p2s\t"); |
| 731 | REPORT_SPEEDUP(opt.chroma[csp].p2s, ref.chroma[csp].p2s, |
| 732 | pixel_buff, srcStride, IPF_vec_output_s, width, height); |
| 733 | } |
| 734 | for (int value = 0; value < NUM_CHROMA_PARTITIONS; value++) |
| 735 | { |
| 736 | if (opt.chroma[csp].filter_hpp[value]) |
| 737 | { |
| 738 | printf("chroma_hpp[%s]", chromaPartStr[csp][value]); |
| 739 | REPORT_SPEEDUP(opt.chroma[csp].filter_hpp[value], ref.chroma[csp].filter_hpp[value], |
| 740 | pixel_buff + srcStride, srcStride, IPF_vec_output_p, dstStride, 1); |
| 741 | } |
| 742 | if (opt.chroma[csp].filter_hps[value]) |
| 743 | { |
| 744 | printf("chroma_hps[%s]", chromaPartStr[csp][value]); |
| 745 | REPORT_SPEEDUP(opt.chroma[csp].filter_hps[value], ref.chroma[csp].filter_hps[value], |
| 746 | pixel_buff + srcStride, srcStride, IPF_vec_output_s, dstStride, 1, 1); |
| 747 | } |
| 748 | if (opt.chroma[csp].filter_vpp[value]) |
| 749 | { |
| 750 | printf("chroma_vpp[%s]", chromaPartStr[csp][value]); |
| 751 | REPORT_SPEEDUP(opt.chroma[csp].filter_vpp[value], ref.chroma[csp].filter_vpp[value], |
| 752 | pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 753 | IPF_vec_output_p, dstStride, 1); |
| 754 | } |
| 755 | if (opt.chroma[csp].filter_vps[value]) |
| 756 | { |
| 757 | printf("chroma_vps[%s]", chromaPartStr[csp][value]); |
| 758 | REPORT_SPEEDUP(opt.chroma[csp].filter_vps[value], ref.chroma[csp].filter_vps[value], |
| 759 | pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 760 | IPF_vec_output_s, dstStride, 1); |
| 761 | } |
| 762 | if (opt.chroma[csp].filter_vsp[value]) |
| 763 | { |
| 764 | printf("chroma_vsp[%s]", chromaPartStr[csp][value]); |
| 765 | REPORT_SPEEDUP(opt.chroma[csp].filter_vsp[value], ref.chroma[csp].filter_vsp[value], |
| 766 | short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 767 | IPF_vec_output_p, dstStride, 1); |
| 768 | } |
| 769 | if (opt.chroma[csp].filter_vss[value]) |
| 770 | { |
| 771 | printf("chroma_vss[%s]", chromaPartStr[csp][value]); |
| 772 | REPORT_SPEEDUP(opt.chroma[csp].filter_vss[value], ref.chroma[csp].filter_vss[value], |
| 773 | short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride, |
| 774 | IPF_vec_output_s, dstStride, 1); |
| 775 | } |
| 776 | } |
| 777 | } |
| 778 | } |