| 1 | ;***************************************************************************** |
| 2 | ;* ssd-a.asm: x86 ssd functions |
| 3 | ;***************************************************************************** |
| 4 | ;* Copyright (C) 2003-2013 x264 project |
| 5 | ;* |
| 6 | ;* Authors: Loren Merritt <lorenm@u.washington.edu> |
| 7 | ;* Fiona Glaser <fiona@x264.com> |
| 8 | ;* Laurent Aimar <fenrir@via.ecp.fr> |
| 9 | ;* Alex Izvorski <aizvorksi@gmail.com> |
| 10 | ;* |
| 11 | ;* This program is free software; you can redistribute it and/or modify |
| 12 | ;* it under the terms of the GNU General Public License as published by |
| 13 | ;* the Free Software Foundation; either version 2 of the License, or |
| 14 | ;* (at your option) any later version. |
| 15 | ;* |
| 16 | ;* This program is distributed in the hope that it will be useful, |
| 17 | ;* but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 18 | ;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 19 | ;* GNU General Public License for more details. |
| 20 | ;* |
| 21 | ;* You should have received a copy of the GNU General Public License |
| 22 | ;* along with this program; if not, write to the Free Software |
| 23 | ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA. |
| 24 | ;* |
| 25 | ;* This program is also available under a commercial proprietary license. |
| 26 | ;* For more information, contact us at license @ x265.com. |
| 27 | ;***************************************************************************** |
| 28 | |
| 29 | %include "x86inc.asm" |
| 30 | %include "x86util.asm" |
| 31 | |
| 32 | SECTION_RODATA 32 |
| 33 | |
| 34 | SECTION .text |
| 35 | |
| 36 | cextern pw_00ff |
| 37 | cextern hsub_mul |
| 38 | |
| 39 | ;============================================================================= |
| 40 | ; SSD |
| 41 | ;============================================================================= |
| 42 | |
| 43 | %if HIGH_BIT_DEPTH |
| 44 | ;----------------------------------------------------------------------------- |
| 45 | ; int pixel_ssd_WxH( uint16_t *, intptr_t, uint16_t *, intptr_t ) |
| 46 | ;----------------------------------------------------------------------------- |
| 47 | %macro SSD_ONE 2 |
| 48 | cglobal pixel_ssd_ss_%1x%2, 4,7,8 |
| 49 | FIX_STRIDES r1, r3 |
| 50 | %if mmsize == %1*2 |
| 51 | %define offset0_1 r1 |
| 52 | %define offset0_2 r1*2 |
| 53 | %define offset0_3 r5 |
| 54 | %define offset1_1 r3 |
| 55 | %define offset1_2 r3*2 |
| 56 | %define offset1_3 r6 |
| 57 | lea r5, [3*r1] |
| 58 | lea r6, [3*r3] |
| 59 | %elif mmsize == %1 |
| 60 | %define offset0_1 mmsize |
| 61 | %define offset0_2 r1 |
| 62 | %define offset0_3 r1+mmsize |
| 63 | %define offset1_1 mmsize |
| 64 | %define offset1_2 r3 |
| 65 | %define offset1_3 r3+mmsize |
| 66 | %elif mmsize == %1/2 |
| 67 | %define offset0_1 mmsize |
| 68 | %define offset0_2 mmsize*2 |
| 69 | %define offset0_3 mmsize*3 |
| 70 | %define offset1_1 mmsize |
| 71 | %define offset1_2 mmsize*2 |
| 72 | %define offset1_3 mmsize*3 |
| 73 | %endif |
| 74 | %assign %%n %2/(2*mmsize/%1) |
| 75 | %if %%n > 1 |
| 76 | mov r4d, %%n |
| 77 | %endif |
| 78 | pxor m0, m0 |
| 79 | .loop: |
| 80 | movu m1, [r0] |
| 81 | movu m2, [r0+offset0_1] |
| 82 | movu m3, [r0+offset0_2] |
| 83 | movu m4, [r0+offset0_3] |
| 84 | movu m6, [r2] |
| 85 | movu m7, [r2+offset1_1] |
| 86 | psubw m1, m6 |
| 87 | psubw m2, m7 |
| 88 | movu m6, [r2+offset1_2] |
| 89 | movu m7, [r2+offset1_3] |
| 90 | psubw m3, m6 |
| 91 | psubw m4, m7 |
| 92 | %if %%n > 1 |
| 93 | lea r0, [r0+r1*(%2/%%n)] |
| 94 | lea r2, [r2+r3*(%2/%%n)] |
| 95 | %endif |
| 96 | pmaddwd m1, m1 |
| 97 | pmaddwd m2, m2 |
| 98 | pmaddwd m3, m3 |
| 99 | pmaddwd m4, m4 |
| 100 | paddd m1, m2 |
| 101 | paddd m3, m4 |
| 102 | paddd m0, m1 |
| 103 | paddd m0, m3 |
| 104 | %if %%n > 1 |
| 105 | dec r4d |
| 106 | jg .loop |
| 107 | %endif |
| 108 | HADDD m0, m5 |
| 109 | movd eax, xm0 |
| 110 | %ifidn movu,movq ; detect MMX |
| 111 | EMMS |
| 112 | %endif |
| 113 | RET |
| 114 | %endmacro |
| 115 | |
| 116 | %macro SSD_TWO 2 |
| 117 | cglobal pixel_ssd_ss_%1x%2, 4,7,8 |
| 118 | FIX_STRIDES r1, r3 |
| 119 | pxor m0, m0 |
| 120 | mov r4d, %2/2 |
| 121 | lea r5, [r1 * 2] |
| 122 | lea r6, [r3 * 2] |
| 123 | .loop: |
| 124 | movu m1, [r0] |
| 125 | movu m2, [r0 + 16] |
| 126 | movu m3, [r0 + 32] |
| 127 | movu m4, [r0 + 48] |
| 128 | movu m6, [r2] |
| 129 | movu m7, [r2 + 16] |
| 130 | psubw m1, m6 |
| 131 | psubw m2, m7 |
| 132 | movu m6, [r2 + 32] |
| 133 | movu m7, [r2 + 48] |
| 134 | psubw m3, m6 |
| 135 | psubw m4, m7 |
| 136 | pmaddwd m1, m1 |
| 137 | pmaddwd m2, m2 |
| 138 | pmaddwd m3, m3 |
| 139 | pmaddwd m4, m4 |
| 140 | paddd m1, m2 |
| 141 | paddd m3, m4 |
| 142 | paddd m0, m1 |
| 143 | paddd m0, m3 |
| 144 | movu m1, [r0 + 64] |
| 145 | movu m2, [r0 + 80] |
| 146 | movu m6, [r2 + 64] |
| 147 | movu m7, [r2 + 80] |
| 148 | psubw m1, m6 |
| 149 | psubw m2, m7 |
| 150 | pmaddwd m1, m1 |
| 151 | pmaddwd m2, m2 |
| 152 | paddd m1, m2 |
| 153 | paddd m0, m1 |
| 154 | %if %1 == 64 |
| 155 | movu m3, [r0 + 96] |
| 156 | movu m4, [r0 + 112] |
| 157 | movu m6, [r2 + 96] |
| 158 | movu m7, [r2 + 112] |
| 159 | psubw m3, m6 |
| 160 | psubw m4, m7 |
| 161 | pmaddwd m3, m3 |
| 162 | pmaddwd m4, m4 |
| 163 | paddd m3, m4 |
| 164 | paddd m0, m3 |
| 165 | %endif |
| 166 | movu m1, [r0 + r1] |
| 167 | movu m2, [r0 + r1 + 16] |
| 168 | movu m3, [r0 + r1 + 32] |
| 169 | movu m4, [r0 + r1 + 48] |
| 170 | movu m6, [r2 + r3] |
| 171 | movu m7, [r2 + r3 + 16] |
| 172 | psubw m1, m6 |
| 173 | psubw m2, m7 |
| 174 | movu m6, [r2 + r3 + 32] |
| 175 | movu m7, [r2 + r3 + 48] |
| 176 | psubw m3, m6 |
| 177 | psubw m4, m7 |
| 178 | pmaddwd m1, m1 |
| 179 | pmaddwd m2, m2 |
| 180 | pmaddwd m3, m3 |
| 181 | pmaddwd m4, m4 |
| 182 | paddd m1, m2 |
| 183 | paddd m3, m4 |
| 184 | paddd m0, m1 |
| 185 | paddd m0, m3 |
| 186 | movu m1, [r0 + r1 + 64] |
| 187 | movu m2, [r0 + r1 + 80] |
| 188 | movu m6, [r2 + r3 + 64] |
| 189 | movu m7, [r2 + r3 + 80] |
| 190 | psubw m1, m6 |
| 191 | psubw m2, m7 |
| 192 | pmaddwd m1, m1 |
| 193 | pmaddwd m2, m2 |
| 194 | paddd m1, m2 |
| 195 | paddd m0, m1 |
| 196 | %if %1 == 64 |
| 197 | movu m3, [r0 + r1 + 96] |
| 198 | movu m4, [r0 + r1 + 112] |
| 199 | movu m6, [r2 + r3 + 96] |
| 200 | movu m7, [r2 + r3 + 112] |
| 201 | psubw m3, m6 |
| 202 | psubw m4, m7 |
| 203 | pmaddwd m3, m3 |
| 204 | pmaddwd m4, m4 |
| 205 | paddd m3, m4 |
| 206 | paddd m0, m3 |
| 207 | %endif |
| 208 | lea r0, [r0 + r5] |
| 209 | lea r2, [r2 + r6] |
| 210 | dec r4d |
| 211 | jnz .loop |
| 212 | HADDD m0, m5 |
| 213 | movd eax, xm0 |
| 214 | RET |
| 215 | %endmacro |
| 216 | %macro SSD_24 2 |
| 217 | cglobal pixel_ssd_ss_%1x%2, 4,7,8 |
| 218 | FIX_STRIDES r1, r3 |
| 219 | pxor m0, m0 |
| 220 | mov r4d, %2/2 |
| 221 | lea r5, [r1 * 2] |
| 222 | lea r6, [r3 * 2] |
| 223 | .loop: |
| 224 | movu m1, [r0] |
| 225 | movu m2, [r0 + 16] |
| 226 | movu m3, [r0 + 32] |
| 227 | movu m5, [r2] |
| 228 | movu m6, [r2 + 16] |
| 229 | movu m7, [r2 + 32] |
| 230 | psubw m1, m5 |
| 231 | psubw m2, m6 |
| 232 | psubw m3, m7 |
| 233 | pmaddwd m1, m1 |
| 234 | pmaddwd m2, m2 |
| 235 | pmaddwd m3, m3 |
| 236 | paddd m1, m2 |
| 237 | paddd m0, m1 |
| 238 | movu m1, [r0 + r1] |
| 239 | movu m2, [r0 + r1 + 16] |
| 240 | movu m4, [r0 + r1 + 32] |
| 241 | movu m5, [r2 + r3] |
| 242 | movu m6, [r2 + r3 + 16] |
| 243 | movu m7, [r2 + r3 + 32] |
| 244 | psubw m1, m5 |
| 245 | psubw m2, m6 |
| 246 | psubw m4, m7 |
| 247 | pmaddwd m1, m1 |
| 248 | pmaddwd m2, m2 |
| 249 | pmaddwd m4, m4 |
| 250 | paddd m1, m2 |
| 251 | paddd m3, m4 |
| 252 | paddd m0, m1 |
| 253 | paddd m0, m3 |
| 254 | lea r0, [r0 + r5] |
| 255 | lea r2, [r2 + r6] |
| 256 | dec r4d |
| 257 | jnz .loop |
| 258 | HADDD m0, m5 |
| 259 | movd eax, xm0 |
| 260 | RET |
| 261 | %endmacro |
| 262 | %macro SSD_12 2 |
| 263 | cglobal pixel_ssd_ss_%1x%2, 4,7,8 |
| 264 | FIX_STRIDES r1, r3 |
| 265 | pxor m0, m0 |
| 266 | mov r4d, %2/4 |
| 267 | lea r5, [r1 * 2] |
| 268 | lea r6, [r3 * 2] |
| 269 | .loop: |
| 270 | movu m1, [r0] |
| 271 | movh m2, [r0 + 16] |
| 272 | movu m3, [r0 + r1] |
| 273 | punpcklqdq m2, [r0 + r1 + 16] |
| 274 | movu m7, [r2] |
| 275 | psubw m1, m7 |
| 276 | movh m4, [r2 + 16] |
| 277 | movu m7, [r2 + r3] |
| 278 | psubw m3, m7 |
| 279 | punpcklqdq m4, [r2 + r3 + 16] |
| 280 | psubw m2, m4 |
| 281 | pmaddwd m1, m1 |
| 282 | pmaddwd m2, m2 |
| 283 | pmaddwd m3, m3 |
| 284 | paddd m1, m2 |
| 285 | paddd m0, m1 |
| 286 | |
| 287 | movu m1, [r0 + r5] |
| 288 | movh m2, [r0 + r5 + 16] |
| 289 | lea r0, [r0 + r5] |
| 290 | movu m6, [r0 + r1] |
| 291 | punpcklqdq m2, [r0 + r1 + 16] |
| 292 | movu m7, [r2 + r6] |
| 293 | psubw m1, m7 |
| 294 | movh m4, [r2 + r6 + 16] |
| 295 | lea r2, [r2 + r6] |
| 296 | movu m7, [r2 + r3] |
| 297 | psubw m6, m7 |
| 298 | punpcklqdq m4, [r2 + r3 + 16] |
| 299 | psubw m2, m4 |
| 300 | pmaddwd m1, m1 |
| 301 | pmaddwd m2, m2 |
| 302 | pmaddwd m6, m6 |
| 303 | paddd m1, m2 |
| 304 | paddd m3, m6 |
| 305 | paddd m0, m1 |
| 306 | paddd m0, m3 |
| 307 | lea r0, [r0 + r5] |
| 308 | lea r2, [r2 + r6] |
| 309 | dec r4d |
| 310 | jnz .loop |
| 311 | HADDD m0, m5 |
| 312 | movd eax, xm0 |
| 313 | RET |
| 314 | %endmacro |
| 315 | INIT_MMX mmx2 |
| 316 | SSD_ONE 4, 4 |
| 317 | SSD_ONE 4, 8 |
| 318 | SSD_ONE 4, 16 |
| 319 | SSD_ONE 8, 4 |
| 320 | SSD_ONE 8, 8 |
| 321 | SSD_ONE 8, 16 |
| 322 | SSD_ONE 16, 8 |
| 323 | SSD_ONE 16, 16 |
| 324 | INIT_XMM sse2 |
| 325 | SSD_ONE 8, 4 |
| 326 | SSD_ONE 8, 8 |
| 327 | SSD_ONE 8, 16 |
| 328 | SSD_ONE 8, 32 |
| 329 | SSD_12 12, 16 |
| 330 | SSD_ONE 16, 4 |
| 331 | SSD_ONE 16, 8 |
| 332 | SSD_ONE 16, 12 |
| 333 | SSD_ONE 16, 16 |
| 334 | SSD_ONE 16, 32 |
| 335 | SSD_ONE 16, 64 |
| 336 | SSD_24 24, 32 |
| 337 | SSD_ONE 32, 8 |
| 338 | SSD_ONE 32, 16 |
| 339 | SSD_ONE 32, 24 |
| 340 | SSD_ONE 32, 32 |
| 341 | SSD_ONE 32, 64 |
| 342 | SSD_TWO 48, 64 |
| 343 | SSD_TWO 64, 16 |
| 344 | SSD_TWO 64, 32 |
| 345 | SSD_TWO 64, 48 |
| 346 | SSD_TWO 64, 64 |
| 347 | INIT_YMM avx2 |
| 348 | SSD_ONE 16, 8 |
| 349 | SSD_ONE 16, 16 |
| 350 | %endif ; HIGH_BIT_DEPTH |
| 351 | |
| 352 | ;----------------------------------------------------------------------------- |
| 353 | ; int pixel_ssd_WxH( uint16_t *, intptr_t, uint16_t *, intptr_t ) |
| 354 | ;----------------------------------------------------------------------------- |
| 355 | %if HIGH_BIT_DEPTH == 0 |
| 356 | %macro SSD_SS 2 |
| 357 | cglobal pixel_ssd_ss_%1x%2, 4,7,6 |
| 358 | FIX_STRIDES r1, r3 |
| 359 | %if mmsize == %1*4 || mmsize == %1*2 |
| 360 | %define offset0_1 r1*2 |
| 361 | %define offset0_2 r1*4 |
| 362 | %define offset0_3 r5 |
| 363 | %define offset1_1 r3*2 |
| 364 | %define offset1_2 r3*4 |
| 365 | %define offset1_3 r6 |
| 366 | lea r5, [4*r1] |
| 367 | lea r6, [4*r3] |
| 368 | lea r5, [r5 + 2*r1] |
| 369 | lea r6, [r6 + 2*r3] |
| 370 | %elif mmsize == %1 |
| 371 | %define offset0_1 16 |
| 372 | %define offset0_2 r1*2 |
| 373 | %define offset0_3 r1*2+16 |
| 374 | %define offset1_1 16 |
| 375 | %define offset1_2 r3*2 |
| 376 | %define offset1_3 r3*2+16 |
| 377 | %endif |
| 378 | %if %1 == 4 |
| 379 | %assign %%n %2/(mmsize/%1) |
| 380 | %else |
| 381 | %assign %%n %2/(2*mmsize/%1) |
| 382 | %endif |
| 383 | %if %%n > 1 |
| 384 | mov r4d, %%n |
| 385 | %endif |
| 386 | pxor m0, m0 |
| 387 | .loop: |
| 388 | %if %1 == 4 |
| 389 | movh m1, [r0] |
| 390 | movh m2, [r2] |
| 391 | psubw m1, m2 |
| 392 | pmaddwd m1, m1 |
| 393 | paddd m0, m1 |
| 394 | movh m1, [r0 + offset0_1] |
| 395 | movh m2, [r2 + offset1_1] |
| 396 | psubw m1, m2 |
| 397 | pmaddwd m1, m1 |
| 398 | paddd m0, m1 |
| 399 | movh m1, [r0 + offset0_2] |
| 400 | movh m2, [r2 + offset1_2] |
| 401 | psubw m1, m2 |
| 402 | pmaddwd m1, m1 |
| 403 | paddd m0, m1 |
| 404 | movh m1, [r0 + offset0_3] |
| 405 | movh m2, [r2 + offset1_3] |
| 406 | psubw m1, m2 |
| 407 | pmaddwd m1, m1 |
| 408 | paddd m0, m1 |
| 409 | %else |
| 410 | movu m1, [r0] |
| 411 | movu m2, [r2] |
| 412 | psubw m1, m2 |
| 413 | pmaddwd m1, m1 |
| 414 | paddd m0, m1 |
| 415 | movu m1, [r0 + offset0_1] |
| 416 | movu m2, [r2 + offset1_1] |
| 417 | psubw m1, m2 |
| 418 | pmaddwd m1, m1 |
| 419 | paddd m0, m1 |
| 420 | movu m1, [r0 + offset0_2] |
| 421 | movu m2, [r2 + offset1_2] |
| 422 | psubw m1, m2 |
| 423 | pmaddwd m1, m1 |
| 424 | paddd m0, m1 |
| 425 | movu m1, [r0 + offset0_3] |
| 426 | movu m2, [r2 + offset1_3] |
| 427 | psubw m1, m2 |
| 428 | pmaddwd m1, m1 |
| 429 | paddd m0, m1 |
| 430 | %endif |
| 431 | lea r0, [r0+r1*(%2/%%n)*2] |
| 432 | lea r2, [r2+r3*(%2/%%n)*2] |
| 433 | %if %%n > 1 |
| 434 | dec r4d |
| 435 | jg .loop |
| 436 | %endif |
| 437 | %if %1 == 4 |
| 438 | %if notcpuflag(ssse3) |
| 439 | pshufd m1, m0, 1 |
| 440 | paddd m0, m1 |
| 441 | %else |
| 442 | phaddd m0, m0 |
| 443 | %endif |
| 444 | %else |
| 445 | HADDD m0, m1 |
| 446 | %endif |
| 447 | movd eax, m0 |
| 448 | RET |
| 449 | %endmacro |
| 450 | %macro SSD_SS_ONE 0 |
| 451 | SSD_SS 4, 4 |
| 452 | SSD_SS 4, 8 |
| 453 | SSD_SS 4, 16 |
| 454 | SSD_SS 8, 4 |
| 455 | SSD_SS 8, 8 |
| 456 | SSD_SS 8, 16 |
| 457 | SSD_SS 8, 32 |
| 458 | SSD_SS 16, 4 |
| 459 | SSD_SS 16, 8 |
| 460 | SSD_SS 16, 12 |
| 461 | SSD_SS 16, 16 |
| 462 | SSD_SS 16, 32 |
| 463 | SSD_SS 16, 64 |
| 464 | %endmacro |
| 465 | |
| 466 | %macro SSD_SS_12x16 0 |
| 467 | cglobal pixel_ssd_ss_12x16, 4,7,6 |
| 468 | FIX_STRIDES r1, r3 |
| 469 | mov r4d, 8 |
| 470 | pxor m0, m0 |
| 471 | .loop: |
| 472 | movu m1, [r0] |
| 473 | movu m2, [r2] |
| 474 | psubw m1, m2 |
| 475 | pmaddwd m1, m1 |
| 476 | paddd m0, m1 |
| 477 | movu m1, [r0 + 16] |
| 478 | movu m2, [r2 + 16] |
| 479 | psubw m1, m2 |
| 480 | pmaddwd m1, m1 |
| 481 | pslldq m1, 8 |
| 482 | psrldq m1, 8 |
| 483 | paddd m0, m1 |
| 484 | lea r0, [r0 + 2*r1] |
| 485 | lea r2, [r2 + 2*r3] |
| 486 | movu m1, [r0] |
| 487 | movu m2, [r2] |
| 488 | psubw m1, m2 |
| 489 | pmaddwd m1, m1 |
| 490 | paddd m0, m1 |
| 491 | movu m1, [r0 + 16] |
| 492 | movu m2, [r2 + 16] |
| 493 | psubw m1, m2 |
| 494 | pmaddwd m1, m1 |
| 495 | pslldq m1, 8 |
| 496 | psrldq m1, 8 |
| 497 | paddd m0, m1 |
| 498 | lea r0, [r0 + 2*r1] |
| 499 | lea r2, [r2 + 2*r3] |
| 500 | dec r4d |
| 501 | jnz .loop |
| 502 | HADDD m0, m1 |
| 503 | movd eax, m0 |
| 504 | RET |
| 505 | %endmacro |
| 506 | |
| 507 | %macro SSD_SS_32 1 |
| 508 | cglobal pixel_ssd_ss_32x%1, 4,7,6 |
| 509 | FIX_STRIDES r1, r3 |
| 510 | mov r4d, %1/2 |
| 511 | pxor m0, m0 |
| 512 | .loop: |
| 513 | movu m1, [r0] |
| 514 | movu m2, [r2] |
| 515 | psubw m1, m2 |
| 516 | pmaddwd m1, m1 |
| 517 | paddd m0, m1 |
| 518 | movu m1, [r0 + 16] |
| 519 | movu m2, [r2 + 16] |
| 520 | psubw m1, m2 |
| 521 | pmaddwd m1, m1 |
| 522 | paddd m0, m1 |
| 523 | movu m1, [r0 + 32] |
| 524 | movu m2, [r2 + 32] |
| 525 | psubw m1, m2 |
| 526 | pmaddwd m1, m1 |
| 527 | paddd m0, m1 |
| 528 | movu m1, [r0 + 48] |
| 529 | movu m2, [r2 + 48] |
| 530 | psubw m1, m2 |
| 531 | pmaddwd m1, m1 |
| 532 | paddd m0, m1 |
| 533 | lea r0, [r0 + 2*r1] |
| 534 | lea r2, [r2 + 2*r3] |
| 535 | movu m1, [r0] |
| 536 | movu m2, [r2] |
| 537 | psubw m1, m2 |
| 538 | pmaddwd m1, m1 |
| 539 | paddd m0, m1 |
| 540 | movu m1, [r0 + 16] |
| 541 | movu m2, [r2 + 16] |
| 542 | psubw m1, m2 |
| 543 | pmaddwd m1, m1 |
| 544 | paddd m0, m1 |
| 545 | movu m1, [r0 + 32] |
| 546 | movu m2, [r2 + 32] |
| 547 | psubw m1, m2 |
| 548 | pmaddwd m1, m1 |
| 549 | paddd m0, m1 |
| 550 | movu m1, [r0 + 48] |
| 551 | movu m2, [r2 + 48] |
| 552 | psubw m1, m2 |
| 553 | pmaddwd m1, m1 |
| 554 | paddd m0, m1 |
| 555 | lea r0, [r0 + 2*r1] |
| 556 | lea r2, [r2 + 2*r3] |
| 557 | dec r4d |
| 558 | jnz .loop |
| 559 | HADDD m0, m1 |
| 560 | movd eax, m0 |
| 561 | RET |
| 562 | %endmacro |
| 563 | |
| 564 | %macro SSD_SS_32xN 0 |
| 565 | SSD_SS_32 8 |
| 566 | SSD_SS_32 16 |
| 567 | SSD_SS_32 24 |
| 568 | SSD_SS_32 32 |
| 569 | SSD_SS_32 64 |
| 570 | %endmacro |
| 571 | |
| 572 | %macro SSD_SS_24 0 |
| 573 | cglobal pixel_ssd_ss_24x32, 4,7,6 |
| 574 | FIX_STRIDES r1, r3 |
| 575 | mov r4d, 16 |
| 576 | pxor m0, m0 |
| 577 | .loop: |
| 578 | movu m1, [r0] |
| 579 | movu m2, [r2] |
| 580 | psubw m1, m2 |
| 581 | pmaddwd m1, m1 |
| 582 | paddd m0, m1 |
| 583 | movu m1, [r0 + 16] |
| 584 | movu m2, [r2 + 16] |
| 585 | psubw m1, m2 |
| 586 | pmaddwd m1, m1 |
| 587 | paddd m0, m1 |
| 588 | movu m1, [r0 + 32] |
| 589 | movu m2, [r2 + 32] |
| 590 | psubw m1, m2 |
| 591 | pmaddwd m1, m1 |
| 592 | paddd m0, m1 |
| 593 | lea r0, [r0 + 2*r1] |
| 594 | lea r2, [r2 + 2*r3] |
| 595 | movu m1, [r0] |
| 596 | movu m2, [r2] |
| 597 | psubw m1, m2 |
| 598 | pmaddwd m1, m1 |
| 599 | paddd m0, m1 |
| 600 | movu m1, [r0 + 16] |
| 601 | movu m2, [r2 + 16] |
| 602 | psubw m1, m2 |
| 603 | pmaddwd m1, m1 |
| 604 | paddd m0, m1 |
| 605 | movu m1, [r0 + 32] |
| 606 | movu m2, [r2 + 32] |
| 607 | psubw m1, m2 |
| 608 | pmaddwd m1, m1 |
| 609 | paddd m0, m1 |
| 610 | lea r0, [r0 + 2*r1] |
| 611 | lea r2, [r2 + 2*r3] |
| 612 | dec r4d |
| 613 | jnz .loop |
| 614 | HADDD m0, m1 |
| 615 | movd eax, m0 |
| 616 | RET |
| 617 | %endmacro |
| 618 | |
| 619 | %macro SSD_SS_48 0 |
| 620 | cglobal pixel_ssd_ss_48x64, 4,7,6 |
| 621 | FIX_STRIDES r1, r3 |
| 622 | mov r4d, 32 |
| 623 | pxor m0, m0 |
| 624 | .loop: |
| 625 | movu m1, [r0] |
| 626 | movu m2, [r2] |
| 627 | psubw m1, m2 |
| 628 | pmaddwd m1, m1 |
| 629 | paddd m0, m1 |
| 630 | movu m1, [r0 + 16] |
| 631 | movu m2, [r2 + 16] |
| 632 | psubw m1, m2 |
| 633 | pmaddwd m1, m1 |
| 634 | paddd m0, m1 |
| 635 | movu m1, [r0 + 32] |
| 636 | movu m2, [r2 + 32] |
| 637 | psubw m1, m2 |
| 638 | pmaddwd m1, m1 |
| 639 | paddd m0, m1 |
| 640 | movu m1, [r0 + 48] |
| 641 | movu m2, [r2 + 48] |
| 642 | psubw m1, m2 |
| 643 | pmaddwd m1, m1 |
| 644 | paddd m0, m1 |
| 645 | movu m1, [r0 + 64] |
| 646 | movu m2, [r2 + 64] |
| 647 | psubw m1, m2 |
| 648 | pmaddwd m1, m1 |
| 649 | paddd m0, m1 |
| 650 | movu m1, [r0 + 80] |
| 651 | movu m2, [r2 + 80] |
| 652 | psubw m1, m2 |
| 653 | pmaddwd m1, m1 |
| 654 | paddd m0, m1 |
| 655 | lea r0, [r0 + 2*r1] |
| 656 | lea r2, [r2 + 2*r3] |
| 657 | movu m1, [r0] |
| 658 | movu m2, [r2] |
| 659 | psubw m1, m2 |
| 660 | pmaddwd m1, m1 |
| 661 | paddd m0, m1 |
| 662 | movu m1, [r0 + 16] |
| 663 | movu m2, [r2 + 16] |
| 664 | psubw m1, m2 |
| 665 | pmaddwd m1, m1 |
| 666 | paddd m0, m1 |
| 667 | movu m1, [r0 + 32] |
| 668 | movu m2, [r2 + 32] |
| 669 | psubw m1, m2 |
| 670 | pmaddwd m1, m1 |
| 671 | paddd m0, m1 |
| 672 | movu m1, [r0 + 48] |
| 673 | movu m2, [r2 + 48] |
| 674 | psubw m1, m2 |
| 675 | pmaddwd m1, m1 |
| 676 | paddd m0, m1 |
| 677 | movu m1, [r0 + 64] |
| 678 | movu m2, [r2 + 64] |
| 679 | psubw m1, m2 |
| 680 | pmaddwd m1, m1 |
| 681 | paddd m0, m1 |
| 682 | movu m1, [r0 + 80] |
| 683 | movu m2, [r2 + 80] |
| 684 | psubw m1, m2 |
| 685 | pmaddwd m1, m1 |
| 686 | paddd m0, m1 |
| 687 | lea r0, [r0 + 2*r1] |
| 688 | lea r2, [r2 + 2*r3] |
| 689 | dec r4d |
| 690 | jnz .loop |
| 691 | HADDD m0, m1 |
| 692 | movd eax, m0 |
| 693 | RET |
| 694 | %endmacro |
| 695 | |
| 696 | %macro SSD_SS_64 1 |
| 697 | cglobal pixel_ssd_ss_64x%1, 4,7,6 |
| 698 | FIX_STRIDES r1, r3 |
| 699 | mov r4d, %1/2 |
| 700 | pxor m0, m0 |
| 701 | .loop: |
| 702 | movu m1, [r0] |
| 703 | movu m2, [r2] |
| 704 | psubw m1, m2 |
| 705 | pmaddwd m1, m1 |
| 706 | paddd m0, m1 |
| 707 | movu m1, [r0 + 16] |
| 708 | movu m2, [r2 + 16] |
| 709 | psubw m1, m2 |
| 710 | pmaddwd m1, m1 |
| 711 | paddd m0, m1 |
| 712 | movu m1, [r0 + 32] |
| 713 | movu m2, [r2 + 32] |
| 714 | psubw m1, m2 |
| 715 | pmaddwd m1, m1 |
| 716 | paddd m0, m1 |
| 717 | movu m1, [r0 + 48] |
| 718 | movu m2, [r2 + 48] |
| 719 | psubw m1, m2 |
| 720 | pmaddwd m1, m1 |
| 721 | paddd m0, m1 |
| 722 | movu m1, [r0 + 64] |
| 723 | movu m2, [r2 + 64] |
| 724 | psubw m1, m2 |
| 725 | pmaddwd m1, m1 |
| 726 | paddd m0, m1 |
| 727 | movu m1, [r0 + 80] |
| 728 | movu m2, [r2 + 80] |
| 729 | psubw m1, m2 |
| 730 | pmaddwd m1, m1 |
| 731 | paddd m0, m1 |
| 732 | movu m1, [r0 + 96] |
| 733 | movu m2, [r2 + 96] |
| 734 | psubw m1, m2 |
| 735 | pmaddwd m1, m1 |
| 736 | paddd m0, m1 |
| 737 | movu m1, [r0 + 112] |
| 738 | movu m2, [r2 + 112] |
| 739 | psubw m1, m2 |
| 740 | pmaddwd m1, m1 |
| 741 | paddd m0, m1 |
| 742 | lea r0, [r0 + 2*r1] |
| 743 | lea r2, [r2 + 2*r3] |
| 744 | movu m1, [r0] |
| 745 | movu m2, [r2] |
| 746 | psubw m1, m2 |
| 747 | pmaddwd m1, m1 |
| 748 | paddd m0, m1 |
| 749 | movu m1, [r0 + 16] |
| 750 | movu m2, [r2 + 16] |
| 751 | psubw m1, m2 |
| 752 | pmaddwd m1, m1 |
| 753 | paddd m0, m1 |
| 754 | movu m1, [r0 + 32] |
| 755 | movu m2, [r2 + 32] |
| 756 | psubw m1, m2 |
| 757 | pmaddwd m1, m1 |
| 758 | paddd m0, m1 |
| 759 | movu m1, [r0 + 48] |
| 760 | movu m2, [r2 + 48] |
| 761 | psubw m1, m2 |
| 762 | pmaddwd m1, m1 |
| 763 | paddd m0, m1 |
| 764 | movu m1, [r0 + 64] |
| 765 | movu m2, [r2 + 64] |
| 766 | psubw m1, m2 |
| 767 | pmaddwd m1, m1 |
| 768 | paddd m0, m1 |
| 769 | movu m1, [r0 + 80] |
| 770 | movu m2, [r2 + 80] |
| 771 | psubw m1, m2 |
| 772 | pmaddwd m1, m1 |
| 773 | paddd m0, m1 |
| 774 | movu m1, [r0 + 96] |
| 775 | movu m2, [r2 + 96] |
| 776 | psubw m1, m2 |
| 777 | pmaddwd m1, m1 |
| 778 | paddd m0, m1 |
| 779 | movu m1, [r0 + 112] |
| 780 | movu m2, [r2 + 112] |
| 781 | psubw m1, m2 |
| 782 | pmaddwd m1, m1 |
| 783 | paddd m0, m1 |
| 784 | lea r0, [r0 + 2*r1] |
| 785 | lea r2, [r2 + 2*r3] |
| 786 | dec r4d |
| 787 | jnz .loop |
| 788 | HADDD m0, m1 |
| 789 | movd eax, m0 |
| 790 | RET |
| 791 | %endmacro |
| 792 | |
| 793 | %macro SSD_SS_64xN 0 |
| 794 | SSD_SS_64 16 |
| 795 | SSD_SS_64 32 |
| 796 | SSD_SS_64 48 |
| 797 | SSD_SS_64 64 |
| 798 | %endmacro |
| 799 | |
| 800 | INIT_XMM sse2 |
| 801 | SSD_SS_ONE |
| 802 | SSD_SS_12x16 |
| 803 | SSD_SS_24 |
| 804 | SSD_SS_32xN |
| 805 | SSD_SS_48 |
| 806 | SSD_SS_64xN |
| 807 | INIT_XMM sse4 |
| 808 | SSD_SS_ONE |
| 809 | SSD_SS_12x16 |
| 810 | SSD_SS_24 |
| 811 | SSD_SS_32xN |
| 812 | SSD_SS_48 |
| 813 | SSD_SS_64xN |
| 814 | INIT_XMM avx |
| 815 | SSD_SS_ONE |
| 816 | SSD_SS_12x16 |
| 817 | SSD_SS_24 |
| 818 | SSD_SS_32xN |
| 819 | SSD_SS_48 |
| 820 | SSD_SS_64xN |
| 821 | %endif ; !HIGH_BIT_DEPTH |
| 822 | |
| 823 | %if HIGH_BIT_DEPTH == 0 |
| 824 | %macro SSD_LOAD_FULL 5 |
| 825 | mova m1, [t0+%1] |
| 826 | mova m2, [t2+%2] |
| 827 | mova m3, [t0+%3] |
| 828 | mova m4, [t2+%4] |
| 829 | %if %5==1 |
| 830 | add t0, t1 |
| 831 | add t2, t3 |
| 832 | %elif %5==2 |
| 833 | lea t0, [t0+2*t1] |
| 834 | lea t2, [t2+2*t3] |
| 835 | %endif |
| 836 | %endmacro |
| 837 | |
| 838 | %macro LOAD 5 |
| 839 | movh m%1, %3 |
| 840 | movh m%2, %4 |
| 841 | %if %5 |
| 842 | lea t0, [t0+2*t1] |
| 843 | %endif |
| 844 | %endmacro |
| 845 | |
| 846 | %macro JOIN 7 |
| 847 | movh m%3, %5 |
| 848 | movh m%4, %6 |
| 849 | %if %7 |
| 850 | lea t2, [t2+2*t3] |
| 851 | %endif |
| 852 | punpcklbw m%1, m7 |
| 853 | punpcklbw m%3, m7 |
| 854 | psubw m%1, m%3 |
| 855 | punpcklbw m%2, m7 |
| 856 | punpcklbw m%4, m7 |
| 857 | psubw m%2, m%4 |
| 858 | %endmacro |
| 859 | |
| 860 | %macro JOIN_SSE2 7 |
| 861 | movh m%3, %5 |
| 862 | movh m%4, %6 |
| 863 | %if %7 |
| 864 | lea t2, [t2+2*t3] |
| 865 | %endif |
| 866 | punpcklqdq m%1, m%2 |
| 867 | punpcklqdq m%3, m%4 |
| 868 | DEINTB %2, %1, %4, %3, 7 |
| 869 | psubw m%2, m%4 |
| 870 | psubw m%1, m%3 |
| 871 | %endmacro |
| 872 | |
| 873 | %macro JOIN_SSSE3 7 |
| 874 | movh m%3, %5 |
| 875 | movh m%4, %6 |
| 876 | %if %7 |
| 877 | lea t2, [t2+2*t3] |
| 878 | %endif |
| 879 | punpcklbw m%1, m%3 |
| 880 | punpcklbw m%2, m%4 |
| 881 | %endmacro |
| 882 | |
| 883 | %macro LOAD_AVX2 5 |
| 884 | mova xm%1, %3 |
| 885 | vinserti128 m%1, m%1, %4, 1 |
| 886 | %if %5 |
| 887 | lea t0, [t0+2*t1] |
| 888 | %endif |
| 889 | %endmacro |
| 890 | |
| 891 | %macro JOIN_AVX2 7 |
| 892 | mova xm%2, %5 |
| 893 | vinserti128 m%2, m%2, %6, 1 |
| 894 | %if %7 |
| 895 | lea t2, [t2+2*t3] |
| 896 | %endif |
| 897 | SBUTTERFLY bw, %1, %2, %3 |
| 898 | %endmacro |
| 899 | |
| 900 | %macro SSD_LOAD_HALF 5 |
| 901 | LOAD 1, 2, [t0+%1], [t0+%3], 1 |
| 902 | JOIN 1, 2, 3, 4, [t2+%2], [t2+%4], 1 |
| 903 | LOAD 3, 4, [t0+%1], [t0+%3], %5 |
| 904 | JOIN 3, 4, 5, 6, [t2+%2], [t2+%4], %5 |
| 905 | %endmacro |
| 906 | |
| 907 | %macro SSD_CORE 7-8 |
| 908 | %ifidn %8, FULL |
| 909 | mova m%6, m%2 |
| 910 | mova m%7, m%4 |
| 911 | psubusb m%2, m%1 |
| 912 | psubusb m%4, m%3 |
| 913 | psubusb m%1, m%6 |
| 914 | psubusb m%3, m%7 |
| 915 | por m%1, m%2 |
| 916 | por m%3, m%4 |
| 917 | punpcklbw m%2, m%1, m%5 |
| 918 | punpckhbw m%1, m%5 |
| 919 | punpcklbw m%4, m%3, m%5 |
| 920 | punpckhbw m%3, m%5 |
| 921 | %endif |
| 922 | pmaddwd m%1, m%1 |
| 923 | pmaddwd m%2, m%2 |
| 924 | pmaddwd m%3, m%3 |
| 925 | pmaddwd m%4, m%4 |
| 926 | %endmacro |
| 927 | |
| 928 | %macro SSD_CORE_SSE2 7-8 |
| 929 | %ifidn %8, FULL |
| 930 | DEINTB %6, %1, %7, %2, %5 |
| 931 | psubw m%6, m%7 |
| 932 | psubw m%1, m%2 |
| 933 | SWAP %6, %2, %1 |
| 934 | DEINTB %6, %3, %7, %4, %5 |
| 935 | psubw m%6, m%7 |
| 936 | psubw m%3, m%4 |
| 937 | SWAP %6, %4, %3 |
| 938 | %endif |
| 939 | pmaddwd m%1, m%1 |
| 940 | pmaddwd m%2, m%2 |
| 941 | pmaddwd m%3, m%3 |
| 942 | pmaddwd m%4, m%4 |
| 943 | %endmacro |
| 944 | |
| 945 | %macro SSD_CORE_SSSE3 7-8 |
| 946 | %ifidn %8, FULL |
| 947 | punpckhbw m%6, m%1, m%2 |
| 948 | punpckhbw m%7, m%3, m%4 |
| 949 | punpcklbw m%1, m%2 |
| 950 | punpcklbw m%3, m%4 |
| 951 | SWAP %6, %2, %3 |
| 952 | SWAP %7, %4 |
| 953 | %endif |
| 954 | pmaddubsw m%1, m%5 |
| 955 | pmaddubsw m%2, m%5 |
| 956 | pmaddubsw m%3, m%5 |
| 957 | pmaddubsw m%4, m%5 |
| 958 | pmaddwd m%1, m%1 |
| 959 | pmaddwd m%2, m%2 |
| 960 | pmaddwd m%3, m%3 |
| 961 | pmaddwd m%4, m%4 |
| 962 | %endmacro |
| 963 | |
| 964 | %macro SSD_ITER 6 |
| 965 | SSD_LOAD_%1 %2,%3,%4,%5,%6 |
| 966 | SSD_CORE 1, 2, 3, 4, 7, 5, 6, %1 |
| 967 | paddd m1, m2 |
| 968 | paddd m3, m4 |
| 969 | paddd m0, m1 |
| 970 | paddd m0, m3 |
| 971 | %endmacro |
| 972 | |
| 973 | ;----------------------------------------------------------------------------- |
| 974 | ; int pixel_ssd_16x16( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 975 | ;----------------------------------------------------------------------------- |
| 976 | %macro SSD 2 |
| 977 | %if %1 != %2 |
| 978 | %assign function_align 8 |
| 979 | %else |
| 980 | %assign function_align 16 |
| 981 | %endif |
| 982 | cglobal pixel_ssd_%1x%2, 0,0,0 |
| 983 | mov al, %1*%2/mmsize/2 |
| 984 | |
| 985 | %if %1 != %2 |
| 986 | jmp mangle(x265_pixel_ssd_%1x%1 %+ SUFFIX %+ .startloop) |
| 987 | %else |
| 988 | |
| 989 | .startloop: |
| 990 | %if ARCH_X86_64 |
| 991 | DECLARE_REG_TMP 0,1,2,3 |
| 992 | PROLOGUE 0,0,8 |
| 993 | %else |
| 994 | PROLOGUE 0,5 |
| 995 | DECLARE_REG_TMP 1,2,3,4 |
| 996 | mov t0, r0m |
| 997 | mov t1, r1m |
| 998 | mov t2, r2m |
| 999 | mov t3, r3m |
| 1000 | %endif |
| 1001 | |
| 1002 | %if cpuflag(ssse3) |
| 1003 | mova m7, [hsub_mul] |
| 1004 | %elifidn cpuname, sse2 |
| 1005 | mova m7, [pw_00ff] |
| 1006 | %elif %1 >= mmsize |
| 1007 | pxor m7, m7 |
| 1008 | %endif |
| 1009 | pxor m0, m0 |
| 1010 | |
| 1011 | ALIGN 16 |
| 1012 | .loop: |
| 1013 | %if %1 > mmsize |
| 1014 | SSD_ITER FULL, 0, 0, mmsize, mmsize, 1 |
| 1015 | %elif %1 == mmsize |
| 1016 | SSD_ITER FULL, 0, 0, t1, t3, 2 |
| 1017 | %else |
| 1018 | SSD_ITER HALF, 0, 0, t1, t3, 2 |
| 1019 | %endif |
| 1020 | dec al |
| 1021 | jg .loop |
| 1022 | %if mmsize==32 |
| 1023 | vextracti128 xm1, m0, 1 |
| 1024 | paddd xm0, xm1 |
| 1025 | HADDD xm0, xm1 |
| 1026 | movd eax, xm0 |
| 1027 | %else |
| 1028 | HADDD m0, m1 |
| 1029 | movd eax, m0 |
| 1030 | %endif |
| 1031 | %if (mmsize == 8) |
| 1032 | emms |
| 1033 | %endif |
| 1034 | RET |
| 1035 | %endif |
| 1036 | %endmacro |
| 1037 | |
| 1038 | %macro HEVC_SSD 0 |
| 1039 | SSD 32, 64 |
| 1040 | SSD 16, 64 |
| 1041 | SSD 32, 32 |
| 1042 | SSD 32, 16 |
| 1043 | SSD 16, 32 |
| 1044 | SSD 32, 8 |
| 1045 | SSD 8, 32 |
| 1046 | SSD 32, 24 |
| 1047 | SSD 24, 24 ; not used, but resolves x265_pixel_ssd_24x24_sse2.startloop symbol |
| 1048 | SSD 8, 4 |
| 1049 | SSD 8, 8 |
| 1050 | SSD 16, 16 |
| 1051 | SSD 16, 12 |
| 1052 | SSD 16, 8 |
| 1053 | SSD 8, 16 |
| 1054 | SSD 16, 4 |
| 1055 | %endmacro |
| 1056 | |
| 1057 | INIT_MMX mmx |
| 1058 | SSD 16, 16 |
| 1059 | SSD 16, 8 |
| 1060 | SSD 8, 8 |
| 1061 | SSD 8, 16 |
| 1062 | SSD 4, 4 |
| 1063 | SSD 8, 4 |
| 1064 | SSD 4, 8 |
| 1065 | SSD 4, 16 |
| 1066 | INIT_XMM sse2slow |
| 1067 | SSD 16, 16 |
| 1068 | SSD 8, 8 |
| 1069 | SSD 16, 8 |
| 1070 | SSD 8, 16 |
| 1071 | SSD 8, 4 |
| 1072 | INIT_XMM sse2 |
| 1073 | %define SSD_CORE SSD_CORE_SSE2 |
| 1074 | %define JOIN JOIN_SSE2 |
| 1075 | HEVC_SSD |
| 1076 | INIT_XMM ssse3 |
| 1077 | %define SSD_CORE SSD_CORE_SSSE3 |
| 1078 | %define JOIN JOIN_SSSE3 |
| 1079 | HEVC_SSD |
| 1080 | INIT_XMM avx |
| 1081 | HEVC_SSD |
| 1082 | INIT_MMX ssse3 |
| 1083 | SSD 4, 4 |
| 1084 | SSD 4, 8 |
| 1085 | SSD 4, 16 |
| 1086 | INIT_XMM xop |
| 1087 | SSD 16, 16 |
| 1088 | SSD 8, 8 |
| 1089 | SSD 16, 8 |
| 1090 | SSD 8, 16 |
| 1091 | SSD 8, 4 |
| 1092 | %define LOAD LOAD_AVX2 |
| 1093 | %define JOIN JOIN_AVX2 |
| 1094 | INIT_YMM avx2 |
| 1095 | SSD 16, 16 |
| 1096 | SSD 16, 8 |
| 1097 | %assign function_align 16 |
| 1098 | %endif ; !HIGH_BIT_DEPTH |
| 1099 | |
| 1100 | ;----------------------------------------------------------------------------- |
| 1101 | ; int pixel_ssd_12x16( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1102 | ;----------------------------------------------------------------------------- |
| 1103 | INIT_XMM sse4 |
| 1104 | cglobal pixel_ssd_12x16, 4, 5, 7, src1, stride1, src2, stride2 |
| 1105 | |
| 1106 | pxor m6, m6 |
| 1107 | mov r4d, 4 |
| 1108 | |
| 1109 | .loop: |
| 1110 | movu m0, [r0] |
| 1111 | movu m1, [r2] |
| 1112 | movu m2, [r0 + r1] |
| 1113 | movu m3, [r2 + r3] |
| 1114 | |
| 1115 | punpckhdq m4, m0, m2 |
| 1116 | punpckhdq m5, m1, m3 |
| 1117 | |
| 1118 | pmovzxbw m0, m0 |
| 1119 | pmovzxbw m1, m1 |
| 1120 | pmovzxbw m2, m2 |
| 1121 | pmovzxbw m3, m3 |
| 1122 | pmovzxbw m4, m4 |
| 1123 | pmovzxbw m5, m5 |
| 1124 | |
| 1125 | psubw m0, m1 |
| 1126 | psubw m2, m3 |
| 1127 | psubw m4, m5 |
| 1128 | |
| 1129 | pmaddwd m0, m0 |
| 1130 | pmaddwd m2, m2 |
| 1131 | pmaddwd m4, m4 |
| 1132 | |
| 1133 | paddd m0, m2 |
| 1134 | paddd m6, m4 |
| 1135 | paddd m6, m0 |
| 1136 | |
| 1137 | movu m0, [r0 + 2 * r1] |
| 1138 | movu m1, [r2 + 2 * r3] |
| 1139 | lea r0, [r0 + 2 * r1] |
| 1140 | lea r2, [r2 + 2 * r3] |
| 1141 | movu m2, [r0 + r1] |
| 1142 | movu m3, [r2 + r3] |
| 1143 | |
| 1144 | punpckhdq m4, m0, m2 |
| 1145 | punpckhdq m5, m1, m3 |
| 1146 | |
| 1147 | pmovzxbw m0, m0 |
| 1148 | pmovzxbw m1, m1 |
| 1149 | pmovzxbw m2, m2 |
| 1150 | pmovzxbw m3, m3 |
| 1151 | pmovzxbw m4, m4 |
| 1152 | pmovzxbw m5, m5 |
| 1153 | |
| 1154 | psubw m0, m1 |
| 1155 | psubw m2, m3 |
| 1156 | psubw m4, m5 |
| 1157 | |
| 1158 | pmaddwd m0, m0 |
| 1159 | pmaddwd m2, m2 |
| 1160 | pmaddwd m4, m4 |
| 1161 | |
| 1162 | paddd m0, m2 |
| 1163 | paddd m6, m4 |
| 1164 | paddd m6, m0 |
| 1165 | |
| 1166 | dec r4d |
| 1167 | lea r0, [r0 + 2 * r1] |
| 1168 | lea r2, [r2 + 2 * r3] |
| 1169 | jnz .loop |
| 1170 | |
| 1171 | HADDD m6, m1 |
| 1172 | movd eax, m6 |
| 1173 | |
| 1174 | RET |
| 1175 | |
| 1176 | ;----------------------------------------------------------------------------- |
| 1177 | ; int pixel_ssd_24x32( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1178 | ;----------------------------------------------------------------------------- |
| 1179 | INIT_XMM sse4 |
| 1180 | cglobal pixel_ssd_24x32, 4, 5, 8, src1, stride1, src2, stride2 |
| 1181 | |
| 1182 | pxor m7, m7 |
| 1183 | pxor m6, m6 |
| 1184 | mov r4d, 16 |
| 1185 | |
| 1186 | .loop: |
| 1187 | movu m1, [r0] |
| 1188 | pmovzxbw m0, m1 |
| 1189 | punpckhbw m1, m6 |
| 1190 | pmovzxbw m2, [r0 + 16] |
| 1191 | movu m4, [r2] |
| 1192 | pmovzxbw m3, m4 |
| 1193 | punpckhbw m4, m6 |
| 1194 | pmovzxbw m5, [r2 + 16] |
| 1195 | |
| 1196 | psubw m0, m3 |
| 1197 | psubw m1, m4 |
| 1198 | psubw m2, m5 |
| 1199 | |
| 1200 | pmaddwd m0, m0 |
| 1201 | pmaddwd m1, m1 |
| 1202 | pmaddwd m2, m2 |
| 1203 | |
| 1204 | paddd m0, m1 |
| 1205 | paddd m7, m2 |
| 1206 | paddd m7, m0 |
| 1207 | |
| 1208 | movu m1, [r0 + r1] |
| 1209 | pmovzxbw m0, m1 |
| 1210 | punpckhbw m1, m6 |
| 1211 | pmovzxbw m2, [r0 + r1 + 16] |
| 1212 | movu m4, [r2 + r3] |
| 1213 | pmovzxbw m3, m4 |
| 1214 | punpckhbw m4, m6 |
| 1215 | pmovzxbw m5, [r2 + r3 + 16] |
| 1216 | |
| 1217 | psubw m0, m3 |
| 1218 | psubw m1, m4 |
| 1219 | psubw m2, m5 |
| 1220 | |
| 1221 | pmaddwd m0, m0 |
| 1222 | pmaddwd m1, m1 |
| 1223 | pmaddwd m2, m2 |
| 1224 | |
| 1225 | paddd m0, m1 |
| 1226 | paddd m7, m2 |
| 1227 | paddd m7, m0 |
| 1228 | |
| 1229 | dec r4d |
| 1230 | lea r0, [r0 + 2 * r1] |
| 1231 | lea r2, [r2 + 2 * r3] |
| 1232 | jnz .loop |
| 1233 | |
| 1234 | HADDD m7, m1 |
| 1235 | movd eax, m7 |
| 1236 | |
| 1237 | RET |
| 1238 | |
| 1239 | %macro PIXEL_SSD_16x4 0 |
| 1240 | movu m1, [r0] |
| 1241 | pmovzxbw m0, m1 |
| 1242 | punpckhbw m1, m6 |
| 1243 | movu m3, [r2] |
| 1244 | pmovzxbw m2, m3 |
| 1245 | punpckhbw m3, m6 |
| 1246 | |
| 1247 | psubw m0, m2 |
| 1248 | psubw m1, m3 |
| 1249 | |
| 1250 | movu m5, [r0 + r1] |
| 1251 | pmovzxbw m4, m5 |
| 1252 | punpckhbw m5, m6 |
| 1253 | movu m3, [r2 + r3] |
| 1254 | pmovzxbw m2, m3 |
| 1255 | punpckhbw m3, m6 |
| 1256 | |
| 1257 | psubw m4, m2 |
| 1258 | psubw m5, m3 |
| 1259 | |
| 1260 | pmaddwd m0, m0 |
| 1261 | pmaddwd m1, m1 |
| 1262 | pmaddwd m4, m4 |
| 1263 | pmaddwd m5, m5 |
| 1264 | |
| 1265 | paddd m0, m1 |
| 1266 | paddd m4, m5 |
| 1267 | paddd m4, m0 |
| 1268 | paddd m7, m4 |
| 1269 | |
| 1270 | movu m1, [r0 + r6] |
| 1271 | pmovzxbw m0, m1 |
| 1272 | punpckhbw m1, m6 |
| 1273 | movu m3, [r2 + 2 * r3] |
| 1274 | pmovzxbw m2, m3 |
| 1275 | punpckhbw m3, m6 |
| 1276 | |
| 1277 | psubw m0, m2 |
| 1278 | psubw m1, m3 |
| 1279 | |
| 1280 | lea r0, [r0 + r6] |
| 1281 | lea r2, [r2 + 2 * r3] |
| 1282 | movu m5, [r0 + r1] |
| 1283 | pmovzxbw m4, m5 |
| 1284 | punpckhbw m5, m6 |
| 1285 | movu m3, [r2 + r3] |
| 1286 | pmovzxbw m2, m3 |
| 1287 | punpckhbw m3, m6 |
| 1288 | |
| 1289 | psubw m4, m2 |
| 1290 | psubw m5, m3 |
| 1291 | |
| 1292 | pmaddwd m0, m0 |
| 1293 | pmaddwd m1, m1 |
| 1294 | pmaddwd m4, m4 |
| 1295 | pmaddwd m5, m5 |
| 1296 | |
| 1297 | paddd m0, m1 |
| 1298 | paddd m4, m5 |
| 1299 | paddd m4, m0 |
| 1300 | paddd m7, m4 |
| 1301 | %endmacro |
| 1302 | |
| 1303 | cglobal pixel_ssd_16x16_internal |
| 1304 | PIXEL_SSD_16x4 |
| 1305 | lea r0, [r0 + r6] |
| 1306 | lea r2, [r2 + 2 * r3] |
| 1307 | PIXEL_SSD_16x4 |
| 1308 | lea r0, [r0 + r6] |
| 1309 | lea r2, [r2 + 2 * r3] |
| 1310 | PIXEL_SSD_16x4 |
| 1311 | lea r0, [r0 + r6] |
| 1312 | lea r2, [r2 + 2 * r3] |
| 1313 | PIXEL_SSD_16x4 |
| 1314 | ret |
| 1315 | |
| 1316 | ;----------------------------------------------------------------------------- |
| 1317 | ; int pixel_ssd_48x64( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1318 | ;----------------------------------------------------------------------------- |
| 1319 | INIT_XMM sse4 |
| 1320 | cglobal pixel_ssd_48x64, 4, 7, 8, src1, stride1, src2, stride2 |
| 1321 | |
| 1322 | pxor m7, m7 |
| 1323 | pxor m6, m6 |
| 1324 | mov r4, r0 |
| 1325 | mov r5, r2 |
| 1326 | lea r6, [r1 * 2] |
| 1327 | |
| 1328 | call pixel_ssd_16x16_internal |
| 1329 | lea r0, [r0 + r6] |
| 1330 | lea r2, [r2 + 2 * r3] |
| 1331 | call pixel_ssd_16x16_internal |
| 1332 | lea r0, [r0 + r6] |
| 1333 | lea r2, [r2 + 2 * r3] |
| 1334 | call pixel_ssd_16x16_internal |
| 1335 | lea r0, [r0 + r6] |
| 1336 | lea r2, [r2 + 2 * r3] |
| 1337 | call pixel_ssd_16x16_internal |
| 1338 | lea r0, [r4 + 16] |
| 1339 | lea r2, [r5 + 16] |
| 1340 | call pixel_ssd_16x16_internal |
| 1341 | lea r0, [r0 + r6] |
| 1342 | lea r2, [r2 + 2 * r3] |
| 1343 | call pixel_ssd_16x16_internal |
| 1344 | lea r0, [r0 + r6] |
| 1345 | lea r2, [r2 + 2 * r3] |
| 1346 | call pixel_ssd_16x16_internal |
| 1347 | lea r0, [r0 + r6] |
| 1348 | lea r2, [r2 + 2 * r3] |
| 1349 | call pixel_ssd_16x16_internal |
| 1350 | lea r0, [r4 + 32] |
| 1351 | lea r2, [r5 + 32] |
| 1352 | call pixel_ssd_16x16_internal |
| 1353 | lea r0, [r0 + r6] |
| 1354 | lea r2, [r2 + 2 * r3] |
| 1355 | call pixel_ssd_16x16_internal |
| 1356 | lea r0, [r0 + r6] |
| 1357 | lea r2, [r2 + 2 * r3] |
| 1358 | call pixel_ssd_16x16_internal |
| 1359 | lea r0, [r0 + r6] |
| 1360 | lea r2, [r2 + 2 * r3] |
| 1361 | call pixel_ssd_16x16_internal |
| 1362 | |
| 1363 | HADDD m7, m1 |
| 1364 | movd eax, m7 |
| 1365 | |
| 1366 | RET |
| 1367 | |
| 1368 | ;----------------------------------------------------------------------------- |
| 1369 | ; int pixel_ssd_64x16( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1370 | ;----------------------------------------------------------------------------- |
| 1371 | INIT_XMM sse4 |
| 1372 | cglobal pixel_ssd_64x16, 4, 7, 8, src1, stride1, src2, stride2 |
| 1373 | |
| 1374 | pxor m7, m7 |
| 1375 | pxor m6, m6 |
| 1376 | mov r4, r0 |
| 1377 | mov r5, r2 |
| 1378 | lea r6, [r1 * 2] |
| 1379 | |
| 1380 | call pixel_ssd_16x16_internal |
| 1381 | lea r0, [r4 + 16] |
| 1382 | lea r2, [r5 + 16] |
| 1383 | call pixel_ssd_16x16_internal |
| 1384 | lea r0, [r4 + 32] |
| 1385 | lea r2, [r5 + 32] |
| 1386 | call pixel_ssd_16x16_internal |
| 1387 | lea r0, [r4 + 48] |
| 1388 | lea r2, [r5 + 48] |
| 1389 | call pixel_ssd_16x16_internal |
| 1390 | |
| 1391 | HADDD m7, m1 |
| 1392 | movd eax, m7 |
| 1393 | |
| 1394 | RET |
| 1395 | |
| 1396 | ;----------------------------------------------------------------------------- |
| 1397 | ; int pixel_ssd_64x32( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1398 | ;----------------------------------------------------------------------------- |
| 1399 | INIT_XMM sse4 |
| 1400 | cglobal pixel_ssd_64x32, 4, 7, 8, src1, stride1, src2, stride2 |
| 1401 | |
| 1402 | pxor m7, m7 |
| 1403 | pxor m6, m6 |
| 1404 | mov r4, r0 |
| 1405 | mov r5, r2 |
| 1406 | lea r6, [r1 * 2] |
| 1407 | |
| 1408 | call pixel_ssd_16x16_internal |
| 1409 | lea r0, [r0 + r6] |
| 1410 | lea r2, [r2 + 2 * r3] |
| 1411 | call pixel_ssd_16x16_internal |
| 1412 | lea r0, [r4 + 16] |
| 1413 | lea r2, [r5 + 16] |
| 1414 | call pixel_ssd_16x16_internal |
| 1415 | lea r0, [r0 + r6] |
| 1416 | lea r2, [r2 + 2 * r3] |
| 1417 | call pixel_ssd_16x16_internal |
| 1418 | lea r0, [r4 + 32] |
| 1419 | lea r2, [r5 + 32] |
| 1420 | call pixel_ssd_16x16_internal |
| 1421 | lea r0, [r0 + r6] |
| 1422 | lea r2, [r2 + 2 * r3] |
| 1423 | call pixel_ssd_16x16_internal |
| 1424 | lea r0, [r4 + 48] |
| 1425 | lea r2, [r5 + 48] |
| 1426 | call pixel_ssd_16x16_internal |
| 1427 | lea r0, [r0 + r6] |
| 1428 | lea r2, [r2 + 2 * r3] |
| 1429 | call pixel_ssd_16x16_internal |
| 1430 | |
| 1431 | HADDD m7, m1 |
| 1432 | movd eax, m7 |
| 1433 | |
| 1434 | RET |
| 1435 | |
| 1436 | ;----------------------------------------------------------------------------- |
| 1437 | ; int pixel_ssd_64x48( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1438 | ;----------------------------------------------------------------------------- |
| 1439 | INIT_XMM sse4 |
| 1440 | cglobal pixel_ssd_64x48, 4, 7, 8, src1, stride1, src2, stride2 |
| 1441 | |
| 1442 | pxor m7, m7 |
| 1443 | pxor m6, m6 |
| 1444 | mov r4, r0 |
| 1445 | mov r5, r2 |
| 1446 | lea r6, [r1 * 2] |
| 1447 | |
| 1448 | call pixel_ssd_16x16_internal |
| 1449 | lea r0, [r0 + r6] |
| 1450 | lea r2, [r2 + 2 * r3] |
| 1451 | call pixel_ssd_16x16_internal |
| 1452 | lea r0, [r0 + r6] |
| 1453 | lea r2, [r2 + 2 * r3] |
| 1454 | call pixel_ssd_16x16_internal |
| 1455 | lea r0, [r4 + 16] |
| 1456 | lea r2, [r5 + 16] |
| 1457 | call pixel_ssd_16x16_internal |
| 1458 | lea r0, [r0 + r6] |
| 1459 | lea r2, [r2 + 2 * r3] |
| 1460 | call pixel_ssd_16x16_internal |
| 1461 | lea r0, [r0 + r6] |
| 1462 | lea r2, [r2 + 2 * r3] |
| 1463 | call pixel_ssd_16x16_internal |
| 1464 | lea r0, [r4 + 32] |
| 1465 | lea r2, [r5 + 32] |
| 1466 | call pixel_ssd_16x16_internal |
| 1467 | lea r0, [r0 + r6] |
| 1468 | lea r2, [r2 + 2 * r3] |
| 1469 | call pixel_ssd_16x16_internal |
| 1470 | lea r0, [r0 + r6] |
| 1471 | lea r2, [r2 + 2 * r3] |
| 1472 | call pixel_ssd_16x16_internal |
| 1473 | lea r0, [r4 + 48] |
| 1474 | lea r2, [r5 + 48] |
| 1475 | call pixel_ssd_16x16_internal |
| 1476 | lea r0, [r0 + r6] |
| 1477 | lea r2, [r2 + 2 * r3] |
| 1478 | call pixel_ssd_16x16_internal |
| 1479 | lea r0, [r0 + r6] |
| 1480 | lea r2, [r2 + 2 * r3] |
| 1481 | call pixel_ssd_16x16_internal |
| 1482 | |
| 1483 | HADDD m7, m1 |
| 1484 | movd eax, m7 |
| 1485 | |
| 1486 | RET |
| 1487 | |
| 1488 | ;----------------------------------------------------------------------------- |
| 1489 | ; int pixel_ssd_64x64( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 1490 | ;----------------------------------------------------------------------------- |
| 1491 | INIT_XMM sse4 |
| 1492 | cglobal pixel_ssd_64x64, 4, 7, 8, src1, stride1, src2, stride2 |
| 1493 | |
| 1494 | pxor m7, m7 |
| 1495 | pxor m6, m6 |
| 1496 | mov r4, r0 |
| 1497 | mov r5, r2 |
| 1498 | lea r6, [r1 * 2] |
| 1499 | |
| 1500 | call pixel_ssd_16x16_internal |
| 1501 | lea r0, [r0 + r6] |
| 1502 | lea r2, [r2 + 2 * r3] |
| 1503 | call pixel_ssd_16x16_internal |
| 1504 | lea r0, [r0 + r6] |
| 1505 | lea r2, [r2 + 2 * r3] |
| 1506 | call pixel_ssd_16x16_internal |
| 1507 | lea r0, [r0 + r6] |
| 1508 | lea r2, [r2 + 2 * r3] |
| 1509 | call pixel_ssd_16x16_internal |
| 1510 | lea r0, [r4 + 16] |
| 1511 | lea r2, [r5 + 16] |
| 1512 | call pixel_ssd_16x16_internal |
| 1513 | lea r0, [r0 + r6] |
| 1514 | lea r2, [r2 + 2 * r3] |
| 1515 | call pixel_ssd_16x16_internal |
| 1516 | lea r0, [r0 + r6] |
| 1517 | lea r2, [r2 + 2 * r3] |
| 1518 | call pixel_ssd_16x16_internal |
| 1519 | lea r0, [r0 + r6] |
| 1520 | lea r2, [r2 + 2 * r3] |
| 1521 | call pixel_ssd_16x16_internal |
| 1522 | lea r0, [r4 + 32] |
| 1523 | lea r2, [r5 + 32] |
| 1524 | call pixel_ssd_16x16_internal |
| 1525 | lea r0, [r0 + r6] |
| 1526 | lea r2, [r2 + 2 * r3] |
| 1527 | call pixel_ssd_16x16_internal |
| 1528 | lea r0, [r0 + r6] |
| 1529 | lea r2, [r2 + 2 * r3] |
| 1530 | call pixel_ssd_16x16_internal |
| 1531 | lea r0, [r0 + r6] |
| 1532 | lea r2, [r2 + 2 * r3] |
| 1533 | call pixel_ssd_16x16_internal |
| 1534 | lea r0, [r4 + 48] |
| 1535 | lea r2, [r5 + 48] |
| 1536 | call pixel_ssd_16x16_internal |
| 1537 | lea r0, [r0 + r6] |
| 1538 | lea r2, [r2 + 2 * r3] |
| 1539 | call pixel_ssd_16x16_internal |
| 1540 | lea r0, [r0 + r6] |
| 1541 | lea r2, [r2 + 2 * r3] |
| 1542 | call pixel_ssd_16x16_internal |
| 1543 | lea r0, [r0 + r6] |
| 1544 | lea r2, [r2 + 2 * r3] |
| 1545 | call pixel_ssd_16x16_internal |
| 1546 | |
| 1547 | HADDD m7, m1 |
| 1548 | movd eax, m7 |
| 1549 | |
| 1550 | RET |
| 1551 | |
| 1552 | ;----------------------------------------------------------------------------- |
| 1553 | ; int pixel_ssd_sp ( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1554 | ;----------------------------------------------------------------------------- |
| 1555 | |
| 1556 | cglobal pixel_ssd_sp_4x4_internal |
| 1557 | movh m0, [r0] |
| 1558 | movh m1, [r0 + r1] |
| 1559 | punpcklqdq m0, m1 |
| 1560 | movd m2, [r2] |
| 1561 | movd m3, [r2 + r3] |
| 1562 | punpckldq m2, m3 |
| 1563 | pmovzxbw m2, m2 |
| 1564 | psubw m0, m2 |
| 1565 | movh m4, [r0 + 2 * r1] |
| 1566 | movh m5, [r0 + r4] |
| 1567 | punpcklqdq m4, m5 |
| 1568 | movd m6, [r2 + 2 * r3] |
| 1569 | lea r2, [r2 + 2 * r3] |
| 1570 | movd m1, [r2 + r3] |
| 1571 | punpckldq m6, m1 |
| 1572 | pmovzxbw m6, m6 |
| 1573 | psubw m4, m6 |
| 1574 | pmaddwd m0, m0 |
| 1575 | pmaddwd m4, m4 |
| 1576 | paddd m0, m4 |
| 1577 | paddd m7, m0 |
| 1578 | ret |
| 1579 | |
| 1580 | ;----------------------------------------------------------------------------- |
| 1581 | ; int pixel_ssd_sp_4x4( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1582 | ;----------------------------------------------------------------------------- |
| 1583 | INIT_XMM sse4 |
| 1584 | cglobal pixel_ssd_sp_4x4, 4, 5, 8, src1, stride1, src2, stride2 |
| 1585 | pxor m7, m7 |
| 1586 | add r1, r1 |
| 1587 | lea r4, [r1 * 3] |
| 1588 | call pixel_ssd_sp_4x4_internal |
| 1589 | HADDD m7, m1 |
| 1590 | movd eax, m7 |
| 1591 | RET |
| 1592 | |
| 1593 | ;----------------------------------------------------------------------------- |
| 1594 | ; int pixel_ssd_sp_4x8( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1595 | ;----------------------------------------------------------------------------- |
| 1596 | INIT_XMM sse4 |
| 1597 | cglobal pixel_ssd_sp_4x8, 4, 5, 8, src1, stride1, src2, stride2 |
| 1598 | pxor m7, m7 |
| 1599 | add r1, r1 |
| 1600 | lea r4, [r1 * 3] |
| 1601 | call pixel_ssd_sp_4x4_internal |
| 1602 | lea r0, [r0 + 4 * r1] |
| 1603 | lea r2, [r2 + 2 * r3] |
| 1604 | call pixel_ssd_sp_4x4_internal |
| 1605 | HADDD m7, m1 |
| 1606 | movd eax, m7 |
| 1607 | RET |
| 1608 | |
| 1609 | ;----------------------------------------------------------------------------- |
| 1610 | ; int pixel_ssd_sp_4x16( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1611 | ;----------------------------------------------------------------------------- |
| 1612 | INIT_XMM sse4 |
| 1613 | cglobal pixel_ssd_sp_4x16, 4, 5, 8, src1, stride1, src2, stride2 |
| 1614 | pxor m7, m7 |
| 1615 | add r1, r1 |
| 1616 | lea r4, [r1 * 3] |
| 1617 | call pixel_ssd_sp_4x4_internal |
| 1618 | lea r0, [r0 + 4 * r1] |
| 1619 | lea r2, [r2 + 2 * r3] |
| 1620 | call pixel_ssd_sp_4x4_internal |
| 1621 | lea r0, [r0 + 4 * r1] |
| 1622 | lea r2, [r2 + 2 * r3] |
| 1623 | call pixel_ssd_sp_4x4_internal |
| 1624 | lea r0, [r0 + 4 * r1] |
| 1625 | lea r2, [r2 + 2 * r3] |
| 1626 | call pixel_ssd_sp_4x4_internal |
| 1627 | HADDD m7, m1 |
| 1628 | movd eax, m7 |
| 1629 | RET |
| 1630 | |
| 1631 | cglobal pixel_ssd_sp_8x4_internal |
| 1632 | movu m0, [r0] |
| 1633 | movu m1, [r0 + r1] |
| 1634 | movh m2, [r2] |
| 1635 | movh m3, [r2 + r3] |
| 1636 | pmovzxbw m2, m2 |
| 1637 | pmovzxbw m3, m3 |
| 1638 | |
| 1639 | psubw m0, m2 |
| 1640 | psubw m1, m3 |
| 1641 | |
| 1642 | movu m4, [r0 + 2 * r1] |
| 1643 | movu m5, [r0 + r4] |
| 1644 | movh m2, [r2 + 2 * r3] |
| 1645 | movh m3, [r2 + r5] |
| 1646 | pmovzxbw m2, m2 |
| 1647 | pmovzxbw m3, m3 |
| 1648 | |
| 1649 | psubw m4, m2 |
| 1650 | psubw m5, m3 |
| 1651 | |
| 1652 | pmaddwd m0, m0 |
| 1653 | pmaddwd m1, m1 |
| 1654 | pmaddwd m4, m4 |
| 1655 | pmaddwd m5, m5 |
| 1656 | |
| 1657 | paddd m0, m1 |
| 1658 | paddd m4, m5 |
| 1659 | paddd m4, m0 |
| 1660 | paddd m7, m4 |
| 1661 | ret |
| 1662 | |
| 1663 | ;----------------------------------------------------------------------------- |
| 1664 | ; int pixel_ssd_sp_8x4( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1665 | ;----------------------------------------------------------------------------- |
| 1666 | INIT_XMM sse4 |
| 1667 | cglobal pixel_ssd_sp_8x4, 4, 6, 8, src1, stride1, src2, stride2 |
| 1668 | pxor m7, m7 |
| 1669 | add r1, r1 |
| 1670 | lea r4, [r1 * 3] |
| 1671 | lea r5, [r3 * 3] |
| 1672 | call pixel_ssd_sp_8x4_internal |
| 1673 | HADDD m7, m1 |
| 1674 | movd eax, m7 |
| 1675 | RET |
| 1676 | |
| 1677 | ;----------------------------------------------------------------------------- |
| 1678 | ; int pixel_ssd_sp_8x8( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1679 | ;----------------------------------------------------------------------------- |
| 1680 | INIT_XMM sse4 |
| 1681 | cglobal pixel_ssd_sp_8x8, 4, 6, 8, src1, stride1, src2, stride2 |
| 1682 | pxor m7, m7 |
| 1683 | add r1, r1 |
| 1684 | lea r4, [r1 * 3] |
| 1685 | lea r5, [r3 * 3] |
| 1686 | call pixel_ssd_sp_8x4_internal |
| 1687 | lea r0, [r0 + 4 * r1] |
| 1688 | lea r2, [r2 + 4 * r3] |
| 1689 | call pixel_ssd_sp_8x4_internal |
| 1690 | HADDD m7, m1 |
| 1691 | movd eax, m7 |
| 1692 | RET |
| 1693 | |
| 1694 | ;----------------------------------------------------------------------------- |
| 1695 | ; int pixel_ssd_sp_8x16( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1696 | ;----------------------------------------------------------------------------- |
| 1697 | INIT_XMM sse4 |
| 1698 | cglobal pixel_ssd_sp_8x16, 4, 6, 8, src1, stride1, src2, stride2 |
| 1699 | pxor m7, m7 |
| 1700 | add r1, r1 |
| 1701 | lea r4, [r1 * 3] |
| 1702 | lea r5, [r3 * 3] |
| 1703 | call pixel_ssd_sp_8x4_internal |
| 1704 | lea r0, [r0 + 4 * r1] |
| 1705 | lea r2, [r2 + 4 * r3] |
| 1706 | call pixel_ssd_sp_8x4_internal |
| 1707 | lea r0, [r0 + 4 * r1] |
| 1708 | lea r2, [r2 + 4 * r3] |
| 1709 | call pixel_ssd_sp_8x4_internal |
| 1710 | lea r0, [r0 + 4 * r1] |
| 1711 | lea r2, [r2 + 4 * r3] |
| 1712 | call pixel_ssd_sp_8x4_internal |
| 1713 | HADDD m7, m1 |
| 1714 | movd eax, m7 |
| 1715 | RET |
| 1716 | |
| 1717 | ;----------------------------------------------------------------------------- |
| 1718 | ; int pixel_ssd_sp_8x32( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1719 | ;----------------------------------------------------------------------------- |
| 1720 | INIT_XMM sse4 |
| 1721 | cglobal pixel_ssd_sp_8x32, 4, 6, 8, src1, stride1, src2, stride2 |
| 1722 | pxor m7, m7 |
| 1723 | add r1, r1 |
| 1724 | lea r4, [r1 * 3] |
| 1725 | lea r5, [r3 * 3] |
| 1726 | call pixel_ssd_sp_8x4_internal |
| 1727 | lea r0, [r0 + 4 * r1] |
| 1728 | lea r2, [r2 + 4 * r3] |
| 1729 | call pixel_ssd_sp_8x4_internal |
| 1730 | lea r0, [r0 + 4 * r1] |
| 1731 | lea r2, [r2 + 4 * r3] |
| 1732 | call pixel_ssd_sp_8x4_internal |
| 1733 | lea r0, [r0 + 4 * r1] |
| 1734 | lea r2, [r2 + 4 * r3] |
| 1735 | call pixel_ssd_sp_8x4_internal |
| 1736 | lea r0, [r0 + 4 * r1] |
| 1737 | lea r2, [r2 + 4 * r3] |
| 1738 | call pixel_ssd_sp_8x4_internal |
| 1739 | lea r0, [r0 + 4 * r1] |
| 1740 | lea r2, [r2 + 4 * r3] |
| 1741 | call pixel_ssd_sp_8x4_internal |
| 1742 | lea r0, [r0 + 4 * r1] |
| 1743 | lea r2, [r2 + 4 * r3] |
| 1744 | call pixel_ssd_sp_8x4_internal |
| 1745 | lea r0, [r0 + 4 * r1] |
| 1746 | lea r2, [r2 + 4 * r3] |
| 1747 | call pixel_ssd_sp_8x4_internal |
| 1748 | HADDD m7, m1 |
| 1749 | movd eax, m7 |
| 1750 | RET |
| 1751 | |
| 1752 | ;----------------------------------------------------------------------------- |
| 1753 | ; int pixel_ssd_sp_12x16( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1754 | ;----------------------------------------------------------------------------- |
| 1755 | INIT_XMM sse4 |
| 1756 | cglobal pixel_ssd_sp_12x16, 4, 7, 8, src1, stride1, src2, stride2 |
| 1757 | pxor m7, m7 |
| 1758 | add r1, r1 |
| 1759 | lea r4, [r1 * 3] |
| 1760 | mov r5, r0 |
| 1761 | mov r6, r2 |
| 1762 | call pixel_ssd_sp_4x4_internal |
| 1763 | lea r0, [r0 + 4 * r1] |
| 1764 | lea r2, [r2 + 2 * r3] |
| 1765 | call pixel_ssd_sp_4x4_internal |
| 1766 | lea r0, [r0 + 4 * r1] |
| 1767 | lea r2, [r2 + 2 * r3] |
| 1768 | call pixel_ssd_sp_4x4_internal |
| 1769 | lea r0, [r0 + 4 * r1] |
| 1770 | lea r2, [r2 + 2 * r3] |
| 1771 | call pixel_ssd_sp_4x4_internal |
| 1772 | lea r0, [r5 + 8] |
| 1773 | lea r2, [r6 + 4] |
| 1774 | lea r5, [r3 * 3] |
| 1775 | call pixel_ssd_sp_8x4_internal |
| 1776 | lea r0, [r0 + 4 * r1] |
| 1777 | lea r2, [r2 + 4 * r3] |
| 1778 | call pixel_ssd_sp_8x4_internal |
| 1779 | lea r0, [r0 + 4 * r1] |
| 1780 | lea r2, [r2 + 4 * r3] |
| 1781 | call pixel_ssd_sp_8x4_internal |
| 1782 | lea r0, [r0 + 4 * r1] |
| 1783 | lea r2, [r2 + 4 * r3] |
| 1784 | call pixel_ssd_sp_8x4_internal |
| 1785 | HADDD m7, m1 |
| 1786 | movd eax, m7 |
| 1787 | RET |
| 1788 | |
| 1789 | %macro PIXEL_SSD_SP_16x4 0 |
| 1790 | movu m0, [r0] |
| 1791 | movu m1, [r0 + 16] |
| 1792 | movu m3, [r2] |
| 1793 | pmovzxbw m2, m3 |
| 1794 | punpckhbw m3, m6 |
| 1795 | |
| 1796 | psubw m0, m2 |
| 1797 | psubw m1, m3 |
| 1798 | |
| 1799 | movu m4, [r0 + r1] |
| 1800 | movu m5, [r0 + r1 +16] |
| 1801 | movu m3, [r2 + r3] |
| 1802 | pmovzxbw m2, m3 |
| 1803 | punpckhbw m3, m6 |
| 1804 | |
| 1805 | psubw m4, m2 |
| 1806 | psubw m5, m3 |
| 1807 | |
| 1808 | pmaddwd m0, m0 |
| 1809 | pmaddwd m1, m1 |
| 1810 | pmaddwd m4, m4 |
| 1811 | pmaddwd m5, m5 |
| 1812 | |
| 1813 | paddd m0, m1 |
| 1814 | paddd m4, m5 |
| 1815 | paddd m4, m0 |
| 1816 | paddd m7, m4 |
| 1817 | |
| 1818 | movu m0, [r0 + 2 * r1] |
| 1819 | movu m1, [r0 + 2 * r1 + 16] |
| 1820 | movu m3, [r2 + 2 * r3] |
| 1821 | pmovzxbw m2, m3 |
| 1822 | punpckhbw m3, m6 |
| 1823 | |
| 1824 | psubw m0, m2 |
| 1825 | psubw m1, m3 |
| 1826 | |
| 1827 | lea r0, [r0 + 2 * r1] |
| 1828 | lea r2, [r2 + 2 * r3] |
| 1829 | movu m4, [r0 + r1] |
| 1830 | movu m5, [r0 + r1 + 16] |
| 1831 | movu m3, [r2 + r3] |
| 1832 | pmovzxbw m2, m3 |
| 1833 | punpckhbw m3, m6 |
| 1834 | |
| 1835 | psubw m4, m2 |
| 1836 | psubw m5, m3 |
| 1837 | |
| 1838 | pmaddwd m0, m0 |
| 1839 | pmaddwd m1, m1 |
| 1840 | pmaddwd m4, m4 |
| 1841 | pmaddwd m5, m5 |
| 1842 | |
| 1843 | paddd m0, m1 |
| 1844 | paddd m4, m5 |
| 1845 | paddd m4, m0 |
| 1846 | paddd m7, m4 |
| 1847 | %endmacro |
| 1848 | |
| 1849 | ;----------------------------------------------------------------------------- |
| 1850 | ; int pixel_ssd_sp_16x4( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1851 | ;----------------------------------------------------------------------------- |
| 1852 | INIT_XMM sse4 |
| 1853 | cglobal pixel_ssd_sp_16x4, 4, 6, 8, src1, stride1, src2, stride2 |
| 1854 | |
| 1855 | pxor m6, m6 |
| 1856 | pxor m7, m7 |
| 1857 | add r1, r1 |
| 1858 | PIXEL_SSD_SP_16x4 |
| 1859 | HADDD m7, m1 |
| 1860 | movd eax, m7 |
| 1861 | |
| 1862 | RET |
| 1863 | |
| 1864 | ;----------------------------------------------------------------------------- |
| 1865 | ; int pixel_ssd_sp_16x8( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1866 | ;----------------------------------------------------------------------------- |
| 1867 | INIT_XMM sse4 |
| 1868 | cglobal pixel_ssd_sp_16x8, 4, 4, 8, src1, stride1, src2, stride2 |
| 1869 | |
| 1870 | pxor m6, m6 |
| 1871 | pxor m7, m7 |
| 1872 | add r1, r1 |
| 1873 | PIXEL_SSD_SP_16x4 |
| 1874 | lea r0, [r0 + 2 * r1] |
| 1875 | lea r2, [r2 + 2 * r3] |
| 1876 | PIXEL_SSD_SP_16x4 |
| 1877 | HADDD m7, m1 |
| 1878 | movd eax, m7 |
| 1879 | RET |
| 1880 | |
| 1881 | ;----------------------------------------------------------------------------- |
| 1882 | ; int pixel_ssd_sp_16x12( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1883 | ;----------------------------------------------------------------------------- |
| 1884 | INIT_XMM sse4 |
| 1885 | cglobal pixel_ssd_sp_16x12, 4, 6, 8, src1, stride1, src2, stride2 |
| 1886 | |
| 1887 | pxor m6, m6 |
| 1888 | pxor m7, m7 |
| 1889 | add r1, r1 |
| 1890 | lea r4, [r1 * 2] |
| 1891 | lea r5, [r3 * 2] |
| 1892 | PIXEL_SSD_SP_16x4 |
| 1893 | lea r0, [r0 + r4] |
| 1894 | lea r2, [r2 + r5] |
| 1895 | PIXEL_SSD_SP_16x4 |
| 1896 | lea r0, [r0 + r4] |
| 1897 | lea r2, [r2 + r5] |
| 1898 | PIXEL_SSD_SP_16x4 |
| 1899 | HADDD m7, m1 |
| 1900 | movd eax, m7 |
| 1901 | RET |
| 1902 | |
| 1903 | ;----------------------------------------------------------------------------- |
| 1904 | ; int pixel_ssd_sp_16x16( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1905 | ;----------------------------------------------------------------------------- |
| 1906 | INIT_XMM sse4 |
| 1907 | cglobal pixel_ssd_sp_16x16, 4, 6, 8, src1, stride1, src2, stride2 |
| 1908 | |
| 1909 | pxor m6, m6 |
| 1910 | pxor m7, m7 |
| 1911 | add r1, r1 |
| 1912 | lea r4, [r1 * 2] |
| 1913 | lea r5, [r3 * 2] |
| 1914 | PIXEL_SSD_SP_16x4 |
| 1915 | lea r0, [r0 + r4] |
| 1916 | lea r2, [r2 + r5] |
| 1917 | PIXEL_SSD_SP_16x4 |
| 1918 | lea r0, [r0 + r4] |
| 1919 | lea r2, [r2 + r5] |
| 1920 | PIXEL_SSD_SP_16x4 |
| 1921 | lea r0, [r0 + r4] |
| 1922 | lea r2, [r2 + r5] |
| 1923 | PIXEL_SSD_SP_16x4 |
| 1924 | HADDD m7, m1 |
| 1925 | movd eax, m7 |
| 1926 | RET |
| 1927 | |
| 1928 | cglobal pixel_ssd_sp_16x16_internal |
| 1929 | PIXEL_SSD_SP_16x4 |
| 1930 | lea r0, [r0 + r4] |
| 1931 | lea r2, [r2 + 2 * r3] |
| 1932 | PIXEL_SSD_SP_16x4 |
| 1933 | lea r0, [r0 + r4] |
| 1934 | lea r2, [r2 + 2 * r3] |
| 1935 | PIXEL_SSD_SP_16x4 |
| 1936 | lea r0, [r0 + r4] |
| 1937 | lea r2, [r2 + 2 * r3] |
| 1938 | PIXEL_SSD_SP_16x4 |
| 1939 | ret |
| 1940 | |
| 1941 | ;----------------------------------------------------------------------------- |
| 1942 | ; int pixel_ssd_sp_16x32( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1943 | ;----------------------------------------------------------------------------- |
| 1944 | INIT_XMM sse4 |
| 1945 | cglobal pixel_ssd_sp_16x32, 4, 5, 8, src1, stride1, src2, stride2 |
| 1946 | |
| 1947 | pxor m6, m6 |
| 1948 | pxor m7, m7 |
| 1949 | add r1, r1 |
| 1950 | lea r4, [r1 * 2] |
| 1951 | call pixel_ssd_sp_16x16_internal |
| 1952 | lea r0, [r0 + r4] |
| 1953 | lea r2, [r2 + 2 * r3] |
| 1954 | call pixel_ssd_sp_16x16_internal |
| 1955 | HADDD m7, m1 |
| 1956 | movd eax, m7 |
| 1957 | RET |
| 1958 | |
| 1959 | ;----------------------------------------------------------------------------- |
| 1960 | ; int pixel_ssd_sp_16x64( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1961 | ;----------------------------------------------------------------------------- |
| 1962 | INIT_XMM sse4 |
| 1963 | cglobal pixel_ssd_sp_16x64, 4, 6, 8, src1, stride1, src2, stride2 |
| 1964 | |
| 1965 | pxor m6, m6 |
| 1966 | pxor m7, m7 |
| 1967 | add r1, r1 |
| 1968 | lea r4, [r1 * 2] |
| 1969 | lea r5, [r3 * 2] |
| 1970 | call pixel_ssd_sp_16x16_internal |
| 1971 | lea r0, [r0 + r4] |
| 1972 | lea r2, [r2 + r5] |
| 1973 | call pixel_ssd_sp_16x16_internal |
| 1974 | lea r0, [r0 + r4] |
| 1975 | lea r2, [r2 + r5] |
| 1976 | call pixel_ssd_sp_16x16_internal |
| 1977 | lea r0, [r0 + r4] |
| 1978 | lea r2, [r2 + r5] |
| 1979 | call pixel_ssd_sp_16x16_internal |
| 1980 | |
| 1981 | HADDD m7, m1 |
| 1982 | movd eax, m7 |
| 1983 | RET |
| 1984 | |
| 1985 | ;----------------------------------------------------------------------------- |
| 1986 | ; int pixel_ssd_sp_24x32( int16_t *, intptr_t, uint8_t *, intptr_t ) |
| 1987 | ;----------------------------------------------------------------------------- |
| 1988 | INIT_XMM sse4 |
| 1989 | cglobal pixel_ssd_sp_24x32, 4, 7, 8, src1, stride1, src2, stride2 |
| 1990 | pxor m6, m6 |
| 1991 | pxor m7, m7 |
| 1992 | add r1, r1 |
| 1993 | lea r4, [r1 * 2] |
| 1994 | mov r5, r0 |
| 1995 | mov r6, r2 |
| 1996 | call pixel_ssd_sp_16x16_internal |
| 1997 | lea r0, [r0 + r4] |
| 1998 | lea r2, [r2 + 2 * r3] |
| 1999 | call pixel_ssd_sp_16x16_internal |
| 2000 | lea r0, [r5 + 32] |
| 2001 | lea r2, [r6 + 16] |
| 2002 | lea r4, [r1 * 3] |
| 2003 | lea r5, [r3 * 3] |
| 2004 | call pixel_ssd_sp_8x4_internal |
| 2005 | lea r0, [r0 + 4 * r1] |
| 2006 | lea r2, [r2 + 4 * r3] |
| 2007 | call pixel_ssd_sp_8x4_internal |
| 2008 | lea r0, [r0 + 4 * r1] |
| 2009 | lea r2, [r2 + 4 * r3] |
| 2010 | call pixel_ssd_sp_8x4_internal |
| 2011 | lea r0, [r0 + 4 * r1] |
| 2012 | lea r2, [r2 + 4 * r3] |
| 2013 | call pixel_ssd_sp_8x4_internal |
| 2014 | lea r0, [r0 + 4 * r1] |
| 2015 | lea r2, [r2 + 4 * r3] |
| 2016 | call pixel_ssd_sp_8x4_internal |
| 2017 | lea r0, [r0 + 4 * r1] |
| 2018 | lea r2, [r2 + 4 * r3] |
| 2019 | call pixel_ssd_sp_8x4_internal |
| 2020 | lea r0, [r0 + 4 * r1] |
| 2021 | lea r2, [r2 + 4 * r3] |
| 2022 | call pixel_ssd_sp_8x4_internal |
| 2023 | lea r0, [r0 + 4 * r1] |
| 2024 | lea r2, [r2 + 4 * r3] |
| 2025 | call pixel_ssd_sp_8x4_internal |
| 2026 | HADDD m7, m1 |
| 2027 | movd eax, m7 |
| 2028 | RET |
| 2029 | |
| 2030 | ;----------------------------------------------------------------------------- |
| 2031 | ; int pixel_ssd_32x8( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2032 | ;----------------------------------------------------------------------------- |
| 2033 | INIT_XMM sse4 |
| 2034 | cglobal pixel_ssd_sp_32x8, 4, 7, 8, src1, stride1, src2, stride2 |
| 2035 | |
| 2036 | pxor m7, m7 |
| 2037 | pxor m6, m6 |
| 2038 | mov r5, r0 |
| 2039 | mov r6, r2 |
| 2040 | add r1, r1 |
| 2041 | lea r4, [r1 * 2] |
| 2042 | PIXEL_SSD_SP_16x4 |
| 2043 | lea r0, [r0 + r4] |
| 2044 | lea r2, [r2 + 2 * r3] |
| 2045 | PIXEL_SSD_SP_16x4 |
| 2046 | lea r0, [r5 + 32] |
| 2047 | lea r2, [r6 + 16] |
| 2048 | PIXEL_SSD_SP_16x4 |
| 2049 | lea r0, [r0 + r4] |
| 2050 | lea r2, [r2 + 2 * r3] |
| 2051 | PIXEL_SSD_SP_16x4 |
| 2052 | HADDD m7, m1 |
| 2053 | movd eax, m7 |
| 2054 | RET |
| 2055 | |
| 2056 | ;----------------------------------------------------------------------------- |
| 2057 | ; int pixel_ssd_32x16( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2058 | ;----------------------------------------------------------------------------- |
| 2059 | INIT_XMM sse4 |
| 2060 | cglobal pixel_ssd_sp_32x16, 4, 7, 8, src1, stride1, src2, stride2 |
| 2061 | |
| 2062 | pxor m7, m7 |
| 2063 | pxor m6, m6 |
| 2064 | mov r5, r0 |
| 2065 | mov r6, r2 |
| 2066 | add r1, r1 |
| 2067 | lea r4, [r1 * 2] |
| 2068 | call pixel_ssd_sp_16x16_internal |
| 2069 | lea r0, [r5 + 32] |
| 2070 | lea r2, [r6 + 16] |
| 2071 | call pixel_ssd_sp_16x16_internal |
| 2072 | HADDD m7, m1 |
| 2073 | movd eax, m7 |
| 2074 | RET |
| 2075 | |
| 2076 | ;----------------------------------------------------------------------------- |
| 2077 | ; int pixel_ssd_32x24( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2078 | ;----------------------------------------------------------------------------- |
| 2079 | INIT_XMM sse4 |
| 2080 | cglobal pixel_ssd_sp_32x24, 4, 7, 8, src1, stride1, src2, stride2 |
| 2081 | |
| 2082 | pxor m7, m7 |
| 2083 | pxor m6, m6 |
| 2084 | mov r5, r0 |
| 2085 | mov r6, r2 |
| 2086 | add r1, r1 |
| 2087 | lea r4, [r1 * 2] |
| 2088 | call pixel_ssd_sp_16x16_internal |
| 2089 | lea r0, [r0 + r4] |
| 2090 | lea r2, [r2 + 2 * r3] |
| 2091 | PIXEL_SSD_SP_16x4 |
| 2092 | lea r0, [r0 + r4] |
| 2093 | lea r2, [r2 + 2 * r3] |
| 2094 | PIXEL_SSD_SP_16x4 |
| 2095 | lea r0, [r5 + 32] |
| 2096 | lea r2, [r6 + 16] |
| 2097 | call pixel_ssd_sp_16x16_internal |
| 2098 | lea r0, [r0 + r4] |
| 2099 | lea r2, [r2 + 2 * r3] |
| 2100 | PIXEL_SSD_SP_16x4 |
| 2101 | lea r0, [r0 + r4] |
| 2102 | lea r2, [r2 + 2 * r3] |
| 2103 | PIXEL_SSD_SP_16x4 |
| 2104 | HADDD m7, m1 |
| 2105 | movd eax, m7 |
| 2106 | RET |
| 2107 | |
| 2108 | ;----------------------------------------------------------------------------- |
| 2109 | ; int pixel_ssd_32x32( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2110 | ;----------------------------------------------------------------------------- |
| 2111 | INIT_XMM sse4 |
| 2112 | cglobal pixel_ssd_sp_32x32, 4, 7, 8, src1, stride1, src2, stride2 |
| 2113 | |
| 2114 | pxor m7, m7 |
| 2115 | pxor m6, m6 |
| 2116 | mov r5, r0 |
| 2117 | mov r6, r2 |
| 2118 | add r1, r1 |
| 2119 | lea r4, [r1 * 2] |
| 2120 | call pixel_ssd_sp_16x16_internal |
| 2121 | lea r0, [r0 + r4] |
| 2122 | lea r2, [r2 + 2 * r3] |
| 2123 | call pixel_ssd_sp_16x16_internal |
| 2124 | lea r0, [r5 + 32] |
| 2125 | lea r2, [r6 + 16] |
| 2126 | call pixel_ssd_sp_16x16_internal |
| 2127 | lea r0, [r0 + r4] |
| 2128 | lea r2, [r2 + 2 * r3] |
| 2129 | call pixel_ssd_sp_16x16_internal |
| 2130 | HADDD m7, m1 |
| 2131 | movd eax, m7 |
| 2132 | RET |
| 2133 | |
| 2134 | ;----------------------------------------------------------------------------- |
| 2135 | ; int pixel_ssd_32x64( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2136 | ;----------------------------------------------------------------------------- |
| 2137 | INIT_XMM sse4 |
| 2138 | cglobal pixel_ssd_sp_32x64, 4, 7, 8, src1, stride1, src2, stride2 |
| 2139 | |
| 2140 | pxor m7, m7 |
| 2141 | pxor m6, m6 |
| 2142 | mov r5, r0 |
| 2143 | mov r6, r2 |
| 2144 | add r1, r1 |
| 2145 | lea r4, [r1 * 2] |
| 2146 | call pixel_ssd_sp_16x16_internal |
| 2147 | lea r0, [r0 + r4] |
| 2148 | lea r2, [r2 + 2 * r3] |
| 2149 | call pixel_ssd_sp_16x16_internal |
| 2150 | lea r0, [r0 + r4] |
| 2151 | lea r2, [r2 + 2 * r3] |
| 2152 | call pixel_ssd_sp_16x16_internal |
| 2153 | lea r0, [r0 + r4] |
| 2154 | lea r2, [r2 + 2 * r3] |
| 2155 | call pixel_ssd_sp_16x16_internal |
| 2156 | lea r0, [r5 + 32] |
| 2157 | lea r2, [r6 + 16] |
| 2158 | call pixel_ssd_sp_16x16_internal |
| 2159 | lea r0, [r0 + r4] |
| 2160 | lea r2, [r2 + 2 * r3] |
| 2161 | call pixel_ssd_sp_16x16_internal |
| 2162 | lea r0, [r0 + r4] |
| 2163 | lea r2, [r2 + 2 * r3] |
| 2164 | call pixel_ssd_sp_16x16_internal |
| 2165 | lea r0, [r0 + r4] |
| 2166 | lea r2, [r2 + 2 * r3] |
| 2167 | call pixel_ssd_sp_16x16_internal |
| 2168 | HADDD m7, m1 |
| 2169 | movd eax, m7 |
| 2170 | RET |
| 2171 | |
| 2172 | ;----------------------------------------------------------------------------- |
| 2173 | ; int pixel_ssd_48x64( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2174 | ;----------------------------------------------------------------------------- |
| 2175 | INIT_XMM sse4 |
| 2176 | cglobal pixel_ssd_sp_48x64, 4, 7, 8, src1, stride1, src2, stride2 |
| 2177 | |
| 2178 | pxor m7, m7 |
| 2179 | pxor m6, m6 |
| 2180 | mov r5, r0 |
| 2181 | mov r6, r2 |
| 2182 | add r1, r1 |
| 2183 | lea r4, [r1 * 2] |
| 2184 | call pixel_ssd_sp_16x16_internal |
| 2185 | lea r0, [r0 + r4] |
| 2186 | lea r2, [r2 + 2 * r3] |
| 2187 | call pixel_ssd_sp_16x16_internal |
| 2188 | lea r0, [r0 + r4] |
| 2189 | lea r2, [r2 + 2 * r3] |
| 2190 | call pixel_ssd_sp_16x16_internal |
| 2191 | lea r0, [r0 + r4] |
| 2192 | lea r2, [r2 + 2 * r3] |
| 2193 | call pixel_ssd_sp_16x16_internal |
| 2194 | lea r0, [r5 + 32] |
| 2195 | lea r2, [r6 + 16] |
| 2196 | call pixel_ssd_sp_16x16_internal |
| 2197 | lea r0, [r0 + r4] |
| 2198 | lea r2, [r2 + 2 * r3] |
| 2199 | call pixel_ssd_sp_16x16_internal |
| 2200 | lea r0, [r0 + r4] |
| 2201 | lea r2, [r2 + 2 * r3] |
| 2202 | call pixel_ssd_sp_16x16_internal |
| 2203 | lea r0, [r0 + r4] |
| 2204 | lea r2, [r2 + 2 * r3] |
| 2205 | call pixel_ssd_sp_16x16_internal |
| 2206 | lea r0, [r5 + 64] |
| 2207 | lea r2, [r6 + 32] |
| 2208 | call pixel_ssd_sp_16x16_internal |
| 2209 | lea r0, [r0 + r4] |
| 2210 | lea r2, [r2 + 2 * r3] |
| 2211 | call pixel_ssd_sp_16x16_internal |
| 2212 | lea r0, [r0 + r4] |
| 2213 | lea r2, [r2 + 2 * r3] |
| 2214 | call pixel_ssd_sp_16x16_internal |
| 2215 | lea r0, [r0 + r4] |
| 2216 | lea r2, [r2 + 2 * r3] |
| 2217 | call pixel_ssd_sp_16x16_internal |
| 2218 | HADDD m7, m1 |
| 2219 | movd eax, m7 |
| 2220 | RET |
| 2221 | |
| 2222 | ;----------------------------------------------------------------------------- |
| 2223 | ; int pixel_ssd_64x16( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2224 | ;----------------------------------------------------------------------------- |
| 2225 | INIT_XMM sse4 |
| 2226 | cglobal pixel_ssd_sp_64x16, 4, 7, 8, src1, stride1, src2, stride2 |
| 2227 | |
| 2228 | pxor m7, m7 |
| 2229 | pxor m6, m6 |
| 2230 | mov r5, r0 |
| 2231 | mov r6, r2 |
| 2232 | add r1, r1 |
| 2233 | lea r4, [r1 * 2] |
| 2234 | call pixel_ssd_sp_16x16_internal |
| 2235 | lea r0, [r5 + 32] |
| 2236 | lea r2, [r6 + 16] |
| 2237 | call pixel_ssd_sp_16x16_internal |
| 2238 | lea r0, [r5 + 64] |
| 2239 | lea r2, [r6 + 32] |
| 2240 | call pixel_ssd_sp_16x16_internal |
| 2241 | lea r0, [r5 + 96] |
| 2242 | lea r2, [r6 + 48] |
| 2243 | call pixel_ssd_sp_16x16_internal |
| 2244 | HADDD m7, m1 |
| 2245 | movd eax, m7 |
| 2246 | RET |
| 2247 | |
| 2248 | ;----------------------------------------------------------------------------- |
| 2249 | ; int pixel_ssd_64x32( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2250 | ;----------------------------------------------------------------------------- |
| 2251 | INIT_XMM sse4 |
| 2252 | cglobal pixel_ssd_sp_64x32, 4, 7, 8, src1, stride1, src2, stride2 |
| 2253 | |
| 2254 | pxor m7, m7 |
| 2255 | pxor m6, m6 |
| 2256 | mov r5, r0 |
| 2257 | mov r6, r2 |
| 2258 | add r1, r1 |
| 2259 | lea r4, [r1 * 2] |
| 2260 | call pixel_ssd_sp_16x16_internal |
| 2261 | lea r0, [r0 + r4] |
| 2262 | lea r2, [r2 + 2 * r3] |
| 2263 | call pixel_ssd_sp_16x16_internal |
| 2264 | lea r0, [r5 + 32] |
| 2265 | lea r2, [r6 + 16] |
| 2266 | call pixel_ssd_sp_16x16_internal |
| 2267 | lea r0, [r0 + r4] |
| 2268 | lea r2, [r2 + 2 * r3] |
| 2269 | call pixel_ssd_sp_16x16_internal |
| 2270 | lea r0, [r5 + 64] |
| 2271 | lea r2, [r6 + 32] |
| 2272 | call pixel_ssd_sp_16x16_internal |
| 2273 | lea r0, [r0 + r4] |
| 2274 | lea r2, [r2 + 2 * r3] |
| 2275 | call pixel_ssd_sp_16x16_internal |
| 2276 | lea r0, [r5 + 96] |
| 2277 | lea r2, [r6 + 48] |
| 2278 | call pixel_ssd_sp_16x16_internal |
| 2279 | lea r0, [r0 + r4] |
| 2280 | lea r2, [r2 + 2 * r3] |
| 2281 | call pixel_ssd_sp_16x16_internal |
| 2282 | HADDD m7, m1 |
| 2283 | movd eax, m7 |
| 2284 | RET |
| 2285 | |
| 2286 | ;----------------------------------------------------------------------------- |
| 2287 | ; int pixel_ssd_64x48( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2288 | ;----------------------------------------------------------------------------- |
| 2289 | INIT_XMM sse4 |
| 2290 | cglobal pixel_ssd_sp_64x48, 4, 7, 8, src1, stride1, src2, stride2 |
| 2291 | |
| 2292 | pxor m7, m7 |
| 2293 | pxor m6, m6 |
| 2294 | mov r5, r0 |
| 2295 | mov r6, r2 |
| 2296 | add r1, r1 |
| 2297 | lea r4, [r1 * 2] |
| 2298 | call pixel_ssd_sp_16x16_internal |
| 2299 | lea r0, [r0 + r4] |
| 2300 | lea r2, [r2 + 2 * r3] |
| 2301 | call pixel_ssd_sp_16x16_internal |
| 2302 | lea r0, [r0 + r4] |
| 2303 | lea r2, [r2 + 2 * r3] |
| 2304 | call pixel_ssd_sp_16x16_internal |
| 2305 | lea r0, [r5 + 32] |
| 2306 | lea r2, [r6 + 16] |
| 2307 | call pixel_ssd_sp_16x16_internal |
| 2308 | lea r0, [r0 + r4] |
| 2309 | lea r2, [r2 + 2 * r3] |
| 2310 | call pixel_ssd_sp_16x16_internal |
| 2311 | lea r0, [r0 + r4] |
| 2312 | lea r2, [r2 + 2 * r3] |
| 2313 | call pixel_ssd_sp_16x16_internal |
| 2314 | lea r0, [r5 + 64] |
| 2315 | lea r2, [r6 + 32] |
| 2316 | call pixel_ssd_sp_16x16_internal |
| 2317 | lea r0, [r0 + r4] |
| 2318 | lea r2, [r2 + 2 * r3] |
| 2319 | call pixel_ssd_sp_16x16_internal |
| 2320 | lea r0, [r0 + r4] |
| 2321 | lea r2, [r2 + 2 * r3] |
| 2322 | call pixel_ssd_sp_16x16_internal |
| 2323 | lea r0, [r5 + 96] |
| 2324 | lea r2, [r6 + 48] |
| 2325 | call pixel_ssd_sp_16x16_internal |
| 2326 | lea r0, [r0 + r4] |
| 2327 | lea r2, [r2 + 2 * r3] |
| 2328 | call pixel_ssd_sp_16x16_internal |
| 2329 | lea r0, [r0 + r4] |
| 2330 | lea r2, [r2 + 2 * r3] |
| 2331 | call pixel_ssd_sp_16x16_internal |
| 2332 | HADDD m7, m1 |
| 2333 | movd eax, m7 |
| 2334 | RET |
| 2335 | |
| 2336 | ;----------------------------------------------------------------------------- |
| 2337 | ; int pixel_ssd_64x64( uint8_t *, intptr_t, uint8_t *, intptr_t ) |
| 2338 | ;----------------------------------------------------------------------------- |
| 2339 | INIT_XMM sse4 |
| 2340 | cglobal pixel_ssd_sp_64x64, 4, 7, 8, src1, stride1, src2, stride2 |
| 2341 | |
| 2342 | pxor m7, m7 |
| 2343 | pxor m6, m6 |
| 2344 | mov r5, r0 |
| 2345 | mov r6, r2 |
| 2346 | add r1, r1 |
| 2347 | lea r4, [r1 * 2] |
| 2348 | call pixel_ssd_sp_16x16_internal |
| 2349 | lea r0, [r0 + r4] |
| 2350 | lea r2, [r2 + 2 * r3] |
| 2351 | call pixel_ssd_sp_16x16_internal |
| 2352 | lea r0, [r0 + r4] |
| 2353 | lea r2, [r2 + 2 * r3] |
| 2354 | call pixel_ssd_sp_16x16_internal |
| 2355 | lea r0, [r0 + r4] |
| 2356 | lea r2, [r2 + 2 * r3] |
| 2357 | call pixel_ssd_sp_16x16_internal |
| 2358 | lea r0, [r5 + 32] |
| 2359 | lea r2, [r6 + 16] |
| 2360 | call pixel_ssd_sp_16x16_internal |
| 2361 | lea r0, [r0 + r4] |
| 2362 | lea r2, [r2 + 2 * r3] |
| 2363 | call pixel_ssd_sp_16x16_internal |
| 2364 | lea r0, [r0 + r4] |
| 2365 | lea r2, [r2 + 2 * r3] |
| 2366 | call pixel_ssd_sp_16x16_internal |
| 2367 | lea r0, [r0 + r4] |
| 2368 | lea r2, [r2 + 2 * r3] |
| 2369 | call pixel_ssd_sp_16x16_internal |
| 2370 | lea r0, [r5 + 64] |
| 2371 | lea r2, [r6 + 32] |
| 2372 | call pixel_ssd_sp_16x16_internal |
| 2373 | lea r0, [r0 + r4] |
| 2374 | lea r2, [r2 + 2 * r3] |
| 2375 | call pixel_ssd_sp_16x16_internal |
| 2376 | lea r0, [r0 + r4] |
| 2377 | lea r2, [r2 + 2 * r3] |
| 2378 | call pixel_ssd_sp_16x16_internal |
| 2379 | lea r0, [r0 + r4] |
| 2380 | lea r2, [r2 + 2 * r3] |
| 2381 | call pixel_ssd_sp_16x16_internal |
| 2382 | lea r0, [r5 + 96] |
| 2383 | lea r2, [r6 + 48] |
| 2384 | call pixel_ssd_sp_16x16_internal |
| 2385 | lea r0, [r0 + r4] |
| 2386 | lea r2, [r2 + 2 * r3] |
| 2387 | call pixel_ssd_sp_16x16_internal |
| 2388 | lea r0, [r0 + r4] |
| 2389 | lea r2, [r2 + 2 * r3] |
| 2390 | call pixel_ssd_sp_16x16_internal |
| 2391 | lea r0, [r0 + r4] |
| 2392 | lea r2, [r2 + 2 * r3] |
| 2393 | call pixel_ssd_sp_16x16_internal |
| 2394 | HADDD m7, m1 |
| 2395 | movd eax, m7 |
| 2396 | RET |
| 2397 | |
| 2398 | |
| 2399 | ;----------------------------------------------------------------------------- |
| 2400 | ; int pixel_ssd_s( int16_t *ref, intptr_t i_stride ) |
| 2401 | ;----------------------------------------------------------------------------- |
| 2402 | INIT_XMM sse2 |
| 2403 | cglobal pixel_ssd_s_4, 2,2,2 |
| 2404 | add r1, r1 |
| 2405 | movh m0, [r0] |
| 2406 | movhps m0, [r0 + r1] |
| 2407 | |
| 2408 | lea r0, [r0 + r1 * 2] |
| 2409 | movh m1, [r0] |
| 2410 | movhps m1, [r0 + r1] |
| 2411 | |
| 2412 | pmaddwd m0, m0 |
| 2413 | pmaddwd m1, m1 |
| 2414 | paddd m0, m1 |
| 2415 | |
| 2416 | ; calculate sum and return |
| 2417 | HADDD m0, m1 |
| 2418 | movd eax, m0 |
| 2419 | RET |
| 2420 | |
| 2421 | |
| 2422 | INIT_XMM sse2 |
| 2423 | cglobal pixel_ssd_s_8, 2,3,5 |
| 2424 | add r1, r1 |
| 2425 | lea r2, [r1 * 3] |
| 2426 | movu m0, [r0] |
| 2427 | movu m1, [r0 + r1] |
| 2428 | movu m2, [r0 + r1 * 2] |
| 2429 | movu m3, [r0 + r2] |
| 2430 | |
| 2431 | pmaddwd m0, m0 |
| 2432 | pmaddwd m1, m1 |
| 2433 | pmaddwd m2, m2 |
| 2434 | pmaddwd m3, m3 |
| 2435 | paddd m0, m1 |
| 2436 | paddd m2, m3 |
| 2437 | paddd m0, m2 |
| 2438 | |
| 2439 | lea r0, [r0 + r1 * 4] |
| 2440 | movu m4, [r0] |
| 2441 | movu m1, [r0 + r1] |
| 2442 | movu m2, [r0 + r1 * 2] |
| 2443 | movu m3, [r0 + r2] |
| 2444 | |
| 2445 | pmaddwd m4, m4 |
| 2446 | pmaddwd m1, m1 |
| 2447 | pmaddwd m2, m2 |
| 2448 | pmaddwd m3, m3 |
| 2449 | paddd m4, m1 |
| 2450 | paddd m2, m3 |
| 2451 | paddd m4, m2 |
| 2452 | paddd m0, m4 |
| 2453 | |
| 2454 | ; calculate sum and return |
| 2455 | HADDD m0, m1 |
| 2456 | movd eax, m0 |
| 2457 | RET |
| 2458 | |
| 2459 | |
| 2460 | INIT_XMM sse2 |
| 2461 | cglobal pixel_ssd_s_16, 2,3,5 |
| 2462 | add r1, r1 |
| 2463 | |
| 2464 | mov r2d, 4 |
| 2465 | pxor m0, m0 |
| 2466 | .loop: |
| 2467 | movu m1, [r0] |
| 2468 | movu m2, [r0 + mmsize] |
| 2469 | movu m3, [r0 + r1] |
| 2470 | movu m4, [r0 + r1 + mmsize] |
| 2471 | lea r0, [r0 + r1 * 2] |
| 2472 | |
| 2473 | pmaddwd m1, m1 |
| 2474 | pmaddwd m2, m2 |
| 2475 | pmaddwd m3, m3 |
| 2476 | pmaddwd m4, m4 |
| 2477 | paddd m1, m2 |
| 2478 | paddd m3, m4 |
| 2479 | paddd m1, m3 |
| 2480 | paddd m0, m1 |
| 2481 | |
| 2482 | movu m1, [r0] |
| 2483 | movu m2, [r0 + mmsize] |
| 2484 | movu m3, [r0 + r1] |
| 2485 | movu m4, [r0 + r1 + mmsize] |
| 2486 | lea r0, [r0 + r1 * 2] |
| 2487 | |
| 2488 | pmaddwd m1, m1 |
| 2489 | pmaddwd m2, m2 |
| 2490 | pmaddwd m3, m3 |
| 2491 | pmaddwd m4, m4 |
| 2492 | paddd m1, m2 |
| 2493 | paddd m3, m4 |
| 2494 | paddd m1, m3 |
| 2495 | paddd m0, m1 |
| 2496 | |
| 2497 | dec r2d |
| 2498 | jnz .loop |
| 2499 | |
| 2500 | ; calculate sum and return |
| 2501 | HADDD m0, m1 |
| 2502 | movd eax, m0 |
| 2503 | RET |
| 2504 | |
| 2505 | |
| 2506 | INIT_XMM sse2 |
| 2507 | cglobal pixel_ssd_s_32, 2,3,5 |
| 2508 | add r1, r1 |
| 2509 | |
| 2510 | mov r2d, 16 |
| 2511 | pxor m0, m0 |
| 2512 | .loop: |
| 2513 | movu m1, [r0 + 0 * mmsize] |
| 2514 | movu m2, [r0 + 1 * mmsize] |
| 2515 | movu m3, [r0 + 2 * mmsize] |
| 2516 | movu m4, [r0 + 3 * mmsize] |
| 2517 | add r0, r1 |
| 2518 | |
| 2519 | pmaddwd m1, m1 |
| 2520 | pmaddwd m2, m2 |
| 2521 | pmaddwd m3, m3 |
| 2522 | pmaddwd m4, m4 |
| 2523 | paddd m1, m2 |
| 2524 | paddd m3, m4 |
| 2525 | paddd m1, m3 |
| 2526 | paddd m0, m1 |
| 2527 | |
| 2528 | movu m1, [r0 + 0 * mmsize] |
| 2529 | movu m2, [r0 + 1 * mmsize] |
| 2530 | movu m3, [r0 + 2 * mmsize] |
| 2531 | movu m4, [r0 + 3 * mmsize] |
| 2532 | add r0, r1 |
| 2533 | |
| 2534 | pmaddwd m1, m1 |
| 2535 | pmaddwd m2, m2 |
| 2536 | pmaddwd m3, m3 |
| 2537 | pmaddwd m4, m4 |
| 2538 | paddd m1, m2 |
| 2539 | paddd m3, m4 |
| 2540 | paddd m1, m3 |
| 2541 | paddd m0, m1 |
| 2542 | |
| 2543 | dec r2d |
| 2544 | jnz .loop |
| 2545 | |
| 2546 | ; calculate sum and return |
| 2547 | HADDD m0, m1 |
| 2548 | movd eax, m0 |
| 2549 | RET |
| 2550 | |
| 2551 | |
| 2552 | INIT_YMM avx2 |
| 2553 | cglobal pixel_ssd_s_32, 2,4,5 |
| 2554 | add r1, r1 |
| 2555 | lea r3, [r1 * 3] |
| 2556 | |
| 2557 | mov r2d, 8 |
| 2558 | pxor m0, m0 |
| 2559 | .loop: |
| 2560 | movu m1, [r0 + 0 * mmsize] |
| 2561 | movu m2, [r0 + 1 * mmsize] |
| 2562 | movu m3, [r0 + r1 + 0 * mmsize] |
| 2563 | movu m4, [r0 + r1 + 1 * mmsize] |
| 2564 | |
| 2565 | pmaddwd m1, m1 |
| 2566 | pmaddwd m2, m2 |
| 2567 | pmaddwd m3, m3 |
| 2568 | pmaddwd m4, m4 |
| 2569 | paddd m1, m2 |
| 2570 | paddd m3, m4 |
| 2571 | paddd m1, m3 |
| 2572 | paddd m0, m1 |
| 2573 | |
| 2574 | movu m1, [r0 + r1 * 2 + 0 * mmsize] |
| 2575 | movu m2, [r0 + r1 * 2 + 1 * mmsize] |
| 2576 | movu m3, [r0 + r3 + 0 * mmsize] |
| 2577 | movu m4, [r0 + r3 + 1 * mmsize] |
| 2578 | lea r0, [r0 + 4 * r1] |
| 2579 | |
| 2580 | pmaddwd m1, m1 |
| 2581 | pmaddwd m2, m2 |
| 2582 | pmaddwd m3, m3 |
| 2583 | pmaddwd m4, m4 |
| 2584 | paddd m1, m2 |
| 2585 | paddd m3, m4 |
| 2586 | paddd m1, m3 |
| 2587 | paddd m0, m1 |
| 2588 | |
| 2589 | dec r2d |
| 2590 | jnz .loop |
| 2591 | |
| 2592 | ; calculate sum and return |
| 2593 | HADDD m0, m1 |
| 2594 | movd eax, xm0 |
| 2595 | RET |