Parent Directory
| 
Revision Log
Revision 1586 - (view) (download)
| 1 : | edgomez | 1586 | ;/***************************************************************************** |
| 2 : | ; * | ||
| 3 : | ; * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | ; * - mmx 8x8 block-based halfpel interpolation - | ||
| 5 : | ; * | ||
| 6 : | ; * Copyright(C) 2001 Peter Ross <pross@xvid.org> | ||
| 7 : | ; * 2002 Michael Militzer <isibaar@xvid.org> | ||
| 8 : | ; * 2004 Andre Werthmann <wertmann@aei.mpg.de> (amd64 conversation) | ||
| 9 : | ; * | ||
| 10 : | ; * This program is free software ; you can redistribute it and/or modify | ||
| 11 : | ; * it under the terms of the GNU General Public License as published by | ||
| 12 : | ; * the Free Software Foundation ; either version 2 of the License, or | ||
| 13 : | ; * (at your option) any later version. | ||
| 14 : | ; * | ||
| 15 : | ; * This program is distributed in the hope that it will be useful, | ||
| 16 : | ; * but WITHOUT ANY WARRANTY ; without even the implied warranty of | ||
| 17 : | ; * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 18 : | ; * GNU General Public License for more details. | ||
| 19 : | ; * | ||
| 20 : | ; * You should have received a copy of the GNU General Public License | ||
| 21 : | ; * along with this program ; if not, write to the Free Software | ||
| 22 : | ; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 23 : | ; * | ||
| 24 : | ; ****************************************************************************/ | ||
| 25 : | |||
| 26 : | BITS 64 | ||
| 27 : | |||
| 28 : | %macro cglobal 1 | ||
| 29 : | %ifdef PREFIX | ||
| 30 : | %ifdef MARK_FUNCS | ||
| 31 : | global _%1:function %1.endfunc-%1 | ||
| 32 : | %define %1 _%1:function %1.endfunc-%1 | ||
| 33 : | %else | ||
| 34 : | global _%1 | ||
| 35 : | %define %1 _%1 | ||
| 36 : | %endif | ||
| 37 : | %else | ||
| 38 : | %ifdef MARK_FUNCS | ||
| 39 : | global %1:function %1.endfunc-%1 | ||
| 40 : | %else | ||
| 41 : | global %1 | ||
| 42 : | %endif | ||
| 43 : | %endif | ||
| 44 : | %endmacro | ||
| 45 : | |||
| 46 : | ;============================================================================= | ||
| 47 : | ; Read only data | ||
| 48 : | ;============================================================================= | ||
| 49 : | |||
| 50 : | %ifdef FORMAT_COFF | ||
| 51 : | SECTION .rodata | ||
| 52 : | %else | ||
| 53 : | SECTION .rodata align=16 | ||
| 54 : | %endif | ||
| 55 : | |||
| 56 : | ;----------------------------------------------------------------------------- | ||
| 57 : | ; (16 - r) rounding table | ||
| 58 : | ;----------------------------------------------------------------------------- | ||
| 59 : | |||
| 60 : | ALIGN 16 | ||
| 61 : | rounding_lowpass_mmx: | ||
| 62 : | times 4 dw 16 | ||
| 63 : | times 4 dw 15 | ||
| 64 : | |||
| 65 : | ;----------------------------------------------------------------------------- | ||
| 66 : | ; (1 - r) rounding table | ||
| 67 : | ;----------------------------------------------------------------------------- | ||
| 68 : | |||
| 69 : | rounding1_mmx: | ||
| 70 : | times 4 dw 1 | ||
| 71 : | times 4 dw 0 | ||
| 72 : | |||
| 73 : | ;----------------------------------------------------------------------------- | ||
| 74 : | ; (2 - r) rounding table | ||
| 75 : | ;----------------------------------------------------------------------------- | ||
| 76 : | |||
| 77 : | rounding2_mmx: | ||
| 78 : | times 4 dw 2 | ||
| 79 : | times 4 dw 1 | ||
| 80 : | |||
| 81 : | mmx_one: | ||
| 82 : | times 8 db 1 | ||
| 83 : | |||
| 84 : | mmx_two: | ||
| 85 : | times 8 db 2 | ||
| 86 : | |||
| 87 : | mmx_three: | ||
| 88 : | times 8 db 3 | ||
| 89 : | |||
| 90 : | mmx_five: | ||
| 91 : | times 4 dw 5 | ||
| 92 : | |||
| 93 : | mmx_mask: | ||
| 94 : | times 8 db 254 | ||
| 95 : | |||
| 96 : | mmx_mask2: | ||
| 97 : | times 8 db 252 | ||
| 98 : | |||
| 99 : | ;============================================================================= | ||
| 100 : | ; Code | ||
| 101 : | ;============================================================================= | ||
| 102 : | |||
| 103 : | SECTION .text align=16 | ||
| 104 : | |||
| 105 : | cglobal interpolate8x8_avg2_x86_64 | ||
| 106 : | cglobal interpolate8x8_avg4_x86_64 | ||
| 107 : | cglobal interpolate8x8_6tap_lowpass_h_x86_64 | ||
| 108 : | cglobal interpolate8x8_6tap_lowpass_v_x86_64 | ||
| 109 : | |||
| 110 : | %macro CALC_AVG 6 | ||
| 111 : | punpcklbw %3, %6 | ||
| 112 : | punpckhbw %4, %6 | ||
| 113 : | |||
| 114 : | paddusw %1, %3 ; mm01 += mm23 | ||
| 115 : | paddusw %2, %4 | ||
| 116 : | paddusw %1, %5 ; mm01 += rounding | ||
| 117 : | paddusw %2, %5 | ||
| 118 : | |||
| 119 : | psrlw %1, 1 ; mm01 >>= 1 | ||
| 120 : | psrlw %2, 1 | ||
| 121 : | %endmacro | ||
| 122 : | |||
| 123 : | ;----------------------------------------------------------------------------- | ||
| 124 : | ; | ||
| 125 : | ; void interpolate8x8_avg2_x86_64(uint8_t const *dst, | ||
| 126 : | ; const uint8_t * const src1, | ||
| 127 : | ; const uint8_t * const src2, | ||
| 128 : | ; const uint32_t stride, | ||
| 129 : | ; const uint32_t rounding, | ||
| 130 : | ; const uint32_t height); | ||
| 131 : | ; | ||
| 132 : | ;----------------------------------------------------------------------------- | ||
| 133 : | |||
| 134 : | %macro AVG2_MMX_RND0 0 | ||
| 135 : | movq mm0, [rax] ; src1 -> mm0 | ||
| 136 : | movq mm1, [rbx] ; src2 -> mm1 | ||
| 137 : | |||
| 138 : | movq mm4, [rax+rdx] | ||
| 139 : | movq mm5, [rbx+rdx] | ||
| 140 : | |||
| 141 : | movq mm2, mm0 ; src1 -> mm2 | ||
| 142 : | movq mm3, mm1 ; src2 -> mm3 | ||
| 143 : | |||
| 144 : | pand mm2, mm7 ; isolate the lsb | ||
| 145 : | pand mm3, mm7 ; isolate the lsb | ||
| 146 : | |||
| 147 : | por mm2, mm3 ; ODD(src1) OR ODD(src2) -> mm2 | ||
| 148 : | |||
| 149 : | movq mm3, mm4 | ||
| 150 : | movq mm6, mm5 | ||
| 151 : | |||
| 152 : | pand mm3, mm7 | ||
| 153 : | pand mm6, mm7 | ||
| 154 : | |||
| 155 : | por mm3, mm6 | ||
| 156 : | |||
| 157 : | pand mm0, [mmx_mask wrt rip] | ||
| 158 : | pand mm1, [mmx_mask wrt rip] | ||
| 159 : | pand mm4, [mmx_mask wrt rip] | ||
| 160 : | pand mm5, [mmx_mask wrt rip] | ||
| 161 : | |||
| 162 : | psrlq mm0, 1 ; src1 / 2 | ||
| 163 : | psrlq mm1, 1 ; src2 / 2 | ||
| 164 : | |||
| 165 : | psrlq mm4, 1 | ||
| 166 : | psrlq mm5, 1 | ||
| 167 : | |||
| 168 : | paddb mm0, mm1 ; src1/2 + src2/2 -> mm0 | ||
| 169 : | paddb mm0, mm2 ; correct rounding error | ||
| 170 : | |||
| 171 : | paddb mm4, mm5 | ||
| 172 : | paddb mm4, mm3 | ||
| 173 : | |||
| 174 : | lea rax, [rax+2*rdx] | ||
| 175 : | lea rbx, [rbx+2*rdx] | ||
| 176 : | |||
| 177 : | movq [rcx], mm0 ; (src1 + src2 + 1) / 2 -> dst | ||
| 178 : | movq [rcx+rdx], mm4 | ||
| 179 : | %endmacro | ||
| 180 : | |||
| 181 : | %macro AVG2_MMX_RND1 0 | ||
| 182 : | movq mm0, [rax] ; src1 -> mm0 | ||
| 183 : | movq mm1, [rbx] ; src2 -> mm1 | ||
| 184 : | |||
| 185 : | movq mm4, [rax+rdx] | ||
| 186 : | movq mm5, [rbx+rdx] | ||
| 187 : | |||
| 188 : | movq mm2, mm0 ; src1 -> mm2 | ||
| 189 : | movq mm3, mm1 ; src2 -> mm3 | ||
| 190 : | |||
| 191 : | pand mm2, mm7 ; isolate the lsb | ||
| 192 : | pand mm3, mm7 ; isolate the lsb | ||
| 193 : | |||
| 194 : | pand mm2, mm3 ; ODD(src1) AND ODD(src2) -> mm2 | ||
| 195 : | |||
| 196 : | movq mm3, mm4 | ||
| 197 : | movq mm6, mm5 | ||
| 198 : | |||
| 199 : | pand mm3, mm7 | ||
| 200 : | pand mm6, mm7 | ||
| 201 : | |||
| 202 : | pand mm3, mm6 | ||
| 203 : | |||
| 204 : | pand mm0, [mmx_mask wrt rip] | ||
| 205 : | pand mm1, [mmx_mask wrt rip] | ||
| 206 : | pand mm4, [mmx_mask wrt rip] | ||
| 207 : | pand mm5, [mmx_mask wrt rip] | ||
| 208 : | |||
| 209 : | psrlq mm0, 1 ; src1 / 2 | ||
| 210 : | psrlq mm1, 1 ; src2 / 2 | ||
| 211 : | |||
| 212 : | psrlq mm4, 1 | ||
| 213 : | psrlq mm5, 1 | ||
| 214 : | |||
| 215 : | paddb mm0, mm1 ; src1/2 + src2/2 -> mm0 | ||
| 216 : | paddb mm0, mm2 ; correct rounding error | ||
| 217 : | |||
| 218 : | paddb mm4, mm5 | ||
| 219 : | paddb mm4, mm3 | ||
| 220 : | |||
| 221 : | lea rax, [rax+2*rdx] | ||
| 222 : | lea rbx, [rbx+2*rdx] | ||
| 223 : | |||
| 224 : | movq [rcx], mm0 ; (src1 + src2 + 1) / 2 -> dst | ||
| 225 : | movq [rcx+rdx], mm4 | ||
| 226 : | %endmacro | ||
| 227 : | |||
| 228 : | ALIGN 16 | ||
| 229 : | interpolate8x8_avg2_x86_64: | ||
| 230 : | |||
| 231 : | push rbx | ||
| 232 : | |||
| 233 : | mov rax, r8 ; 5. par | ||
| 234 : | test rax, rax | ||
| 235 : | |||
| 236 : | jnz near .rounding1 | ||
| 237 : | |||
| 238 : | mov rax, r9 ; 6. par (height) | ||
| 239 : | sub rax, 8 | ||
| 240 : | test rax, rax | ||
| 241 : | |||
| 242 : | mov rbx, rdx ; src2 | ||
| 243 : | mov rdx, rcx ; stride | ||
| 244 : | mov rax, rsi ; src1 | ||
| 245 : | mov rcx, rdi ; dst | ||
| 246 : | |||
| 247 : | movq mm7, [mmx_one wrt rip] | ||
| 248 : | |||
| 249 : | jz near .start0 | ||
| 250 : | |||
| 251 : | AVG2_MMX_RND0 | ||
| 252 : | lea rcx, [rcx+2*rdx] | ||
| 253 : | |||
| 254 : | .start0 | ||
| 255 : | |||
| 256 : | AVG2_MMX_RND0 | ||
| 257 : | lea rcx, [rcx+2*rdx] | ||
| 258 : | AVG2_MMX_RND0 | ||
| 259 : | lea rcx, [rcx+2*rdx] | ||
| 260 : | AVG2_MMX_RND0 | ||
| 261 : | lea rcx, [rcx+2*rdx] | ||
| 262 : | AVG2_MMX_RND0 | ||
| 263 : | |||
| 264 : | pop rbx | ||
| 265 : | ret | ||
| 266 : | |||
| 267 : | .rounding1 | ||
| 268 : | mov rax, r9 ; 6. par (height) | ||
| 269 : | sub rax, 8 | ||
| 270 : | test rax, rax | ||
| 271 : | |||
| 272 : | mov rbx, rdx ; src2 | ||
| 273 : | mov rdx, rcx ; stride | ||
| 274 : | mov rax, rsi ; src1 | ||
| 275 : | mov rcx, rdi ; dst | ||
| 276 : | |||
| 277 : | movq mm7, [mmx_one wrt rip] | ||
| 278 : | |||
| 279 : | jz near .start1 | ||
| 280 : | |||
| 281 : | AVG2_MMX_RND1 | ||
| 282 : | lea rcx, [rcx+2*rdx] | ||
| 283 : | |||
| 284 : | .start1 | ||
| 285 : | |||
| 286 : | AVG2_MMX_RND1 | ||
| 287 : | lea rcx, [rcx+2*rdx] | ||
| 288 : | AVG2_MMX_RND1 | ||
| 289 : | lea rcx, [rcx+2*rdx] | ||
| 290 : | AVG2_MMX_RND1 | ||
| 291 : | lea rcx, [rcx+2*rdx] | ||
| 292 : | AVG2_MMX_RND1 | ||
| 293 : | |||
| 294 : | pop rbx | ||
| 295 : | ret | ||
| 296 : | .endfunc | ||
| 297 : | |||
| 298 : | |||
| 299 : | ;----------------------------------------------------------------------------- | ||
| 300 : | ; | ||
| 301 : | ; void interpolate8x8_avg4_x86_64(uint8_t const *dst, | ||
| 302 : | ; const uint8_t * const src1, | ||
| 303 : | ; const uint8_t * const src2, | ||
| 304 : | ; const uint8_t * const src3, | ||
| 305 : | ; const uint8_t * const src4, | ||
| 306 : | ; const uint32_t stride, | ||
| 307 : | ; const uint32_t rounding); | ||
| 308 : | ; | ||
| 309 : | ;----------------------------------------------------------------------------- | ||
| 310 : | |||
| 311 : | %macro AVG4_MMX_RND0 0 | ||
| 312 : | movq mm0, [rax] ; src1 -> mm0 | ||
| 313 : | movq mm1, [rbx] ; src2 -> mm1 | ||
| 314 : | |||
| 315 : | movq mm2, mm0 | ||
| 316 : | movq mm3, mm1 | ||
| 317 : | |||
| 318 : | pand mm2, [mmx_three wrt rip] | ||
| 319 : | pand mm3, [mmx_three wrt rip] | ||
| 320 : | |||
| 321 : | pand mm0, [mmx_mask2 wrt rip] | ||
| 322 : | pand mm1, [mmx_mask2 wrt rip] | ||
| 323 : | |||
| 324 : | psrlq mm0, 2 | ||
| 325 : | psrlq mm1, 2 | ||
| 326 : | |||
| 327 : | lea rax, [rax+rdx] | ||
| 328 : | lea rbx, [rbx+rdx] | ||
| 329 : | |||
| 330 : | paddb mm0, mm1 | ||
| 331 : | paddb mm2, mm3 | ||
| 332 : | |||
| 333 : | movq mm4, [rsi] ; src3 -> mm0 | ||
| 334 : | movq mm5, [rdi] ; src4 -> mm1 | ||
| 335 : | |||
| 336 : | movq mm1, mm4 | ||
| 337 : | movq mm3, mm5 | ||
| 338 : | |||
| 339 : | pand mm1, [mmx_three wrt rip] | ||
| 340 : | pand mm3, [mmx_three wrt rip] | ||
| 341 : | |||
| 342 : | pand mm4, [mmx_mask2 wrt rip] | ||
| 343 : | pand mm5, [mmx_mask2 wrt rip] | ||
| 344 : | |||
| 345 : | psrlq mm4, 2 | ||
| 346 : | psrlq mm5, 2 | ||
| 347 : | |||
| 348 : | paddb mm4, mm5 | ||
| 349 : | paddb mm0, mm4 | ||
| 350 : | |||
| 351 : | paddb mm1, mm3 | ||
| 352 : | paddb mm2, mm1 | ||
| 353 : | |||
| 354 : | paddb mm2, [mmx_two wrt rip] | ||
| 355 : | pand mm2, [mmx_mask2 wrt rip] | ||
| 356 : | |||
| 357 : | psrlq mm2, 2 | ||
| 358 : | paddb mm0, mm2 | ||
| 359 : | |||
| 360 : | lea rsi, [rsi+rdx] | ||
| 361 : | lea rdi, [rdi+rdx] | ||
| 362 : | |||
| 363 : | movq [rcx], mm0 ; (src1 + src2 + src3 + src4 + 2) / 4 -> dst | ||
| 364 : | %endmacro | ||
| 365 : | |||
| 366 : | %macro AVG4_MMX_RND1 0 | ||
| 367 : | movq mm0, [rax] ; src1 -> mm0 | ||
| 368 : | movq mm1, [rbx] ; src2 -> mm1 | ||
| 369 : | |||
| 370 : | movq mm2, mm0 | ||
| 371 : | movq mm3, mm1 | ||
| 372 : | |||
| 373 : | pand mm2, [mmx_three wrt rip] | ||
| 374 : | pand mm3, [mmx_three wrt rip] | ||
| 375 : | |||
| 376 : | pand mm0, [mmx_mask2 wrt rip] | ||
| 377 : | pand mm1, [mmx_mask2 wrt rip] | ||
| 378 : | |||
| 379 : | psrlq mm0, 2 | ||
| 380 : | psrlq mm1, 2 | ||
| 381 : | |||
| 382 : | lea rax,[rax+rdx] | ||
| 383 : | lea rbx,[rbx+rdx] | ||
| 384 : | |||
| 385 : | paddb mm0, mm1 | ||
| 386 : | paddb mm2, mm3 | ||
| 387 : | |||
| 388 : | movq mm4, [rsi] ; src3 -> mm0 | ||
| 389 : | movq mm5, [rdi] ; src4 -> mm1 | ||
| 390 : | |||
| 391 : | movq mm1, mm4 | ||
| 392 : | movq mm3, mm5 | ||
| 393 : | |||
| 394 : | pand mm1, [mmx_three wrt rip] | ||
| 395 : | pand mm3, [mmx_three wrt rip] | ||
| 396 : | |||
| 397 : | pand mm4, [mmx_mask2 wrt rip] | ||
| 398 : | pand mm5, [mmx_mask2 wrt rip] | ||
| 399 : | |||
| 400 : | psrlq mm4, 2 | ||
| 401 : | psrlq mm5, 2 | ||
| 402 : | |||
| 403 : | paddb mm4, mm5 | ||
| 404 : | paddb mm0, mm4 | ||
| 405 : | |||
| 406 : | paddb mm1, mm3 | ||
| 407 : | paddb mm2, mm1 | ||
| 408 : | |||
| 409 : | paddb mm2, [mmx_one wrt rip] | ||
| 410 : | pand mm2, [mmx_mask2 wrt rip] | ||
| 411 : | |||
| 412 : | psrlq mm2, 2 | ||
| 413 : | paddb mm0, mm2 | ||
| 414 : | |||
| 415 : | lea rsi,[rsi+rdx] | ||
| 416 : | lea rdi,[rdi+rdx] | ||
| 417 : | |||
| 418 : | movq [rcx], mm0 ; (src1 + src2 + src3 + src4 + 2) / 4 -> dst | ||
| 419 : | %endmacro | ||
| 420 : | |||
| 421 : | ALIGN 16 | ||
| 422 : | interpolate8x8_avg4_x86_64: | ||
| 423 : | |||
| 424 : | push rbx | ||
| 425 : | |||
| 426 : | mov rax, [rsp + 8 + 8] ; rounding | ||
| 427 : | |||
| 428 : | test rax, rax | ||
| 429 : | |||
| 430 : | mov rbx, rdx ; src2 | ||
| 431 : | mov rdx, r9 ; stride | ||
| 432 : | mov rax, rsi ; src1 | ||
| 433 : | mov rsi, rcx ; src3 | ||
| 434 : | mov rcx, rdi ; dst | ||
| 435 : | mov rdi, r8 ; src4 | ||
| 436 : | |||
| 437 : | movq mm7, [mmx_one wrt rip] | ||
| 438 : | |||
| 439 : | jnz near .rounding1 | ||
| 440 : | |||
| 441 : | AVG4_MMX_RND0 | ||
| 442 : | lea rcx, [rcx+rdx] | ||
| 443 : | AVG4_MMX_RND0 | ||
| 444 : | lea rcx, [rcx+rdx] | ||
| 445 : | AVG4_MMX_RND0 | ||
| 446 : | lea rcx, [rcx+rdx] | ||
| 447 : | AVG4_MMX_RND0 | ||
| 448 : | lea rcx, [rcx+rdx] | ||
| 449 : | AVG4_MMX_RND0 | ||
| 450 : | lea rcx, [rcx+rdx] | ||
| 451 : | AVG4_MMX_RND0 | ||
| 452 : | lea rcx, [rcx+rdx] | ||
| 453 : | AVG4_MMX_RND0 | ||
| 454 : | lea rcx, [rcx+rdx] | ||
| 455 : | AVG4_MMX_RND0 | ||
| 456 : | |||
| 457 : | pop rbx | ||
| 458 : | ret | ||
| 459 : | |||
| 460 : | .rounding1 | ||
| 461 : | AVG4_MMX_RND1 | ||
| 462 : | lea rcx, [rcx+rdx] | ||
| 463 : | AVG4_MMX_RND1 | ||
| 464 : | lea rcx, [rcx+rdx] | ||
| 465 : | AVG4_MMX_RND1 | ||
| 466 : | lea rcx, [rcx+rdx] | ||
| 467 : | AVG4_MMX_RND1 | ||
| 468 : | lea rcx, [rcx+rdx] | ||
| 469 : | AVG4_MMX_RND1 | ||
| 470 : | lea rcx, [rcx+rdx] | ||
| 471 : | AVG4_MMX_RND1 | ||
| 472 : | lea rcx, [rcx+rdx] | ||
| 473 : | AVG4_MMX_RND1 | ||
| 474 : | lea rcx, [rcx+rdx] | ||
| 475 : | AVG4_MMX_RND1 | ||
| 476 : | |||
| 477 : | pop rbx | ||
| 478 : | ret | ||
| 479 : | .endfunc | ||
| 480 : | |||
| 481 : | ;----------------------------------------------------------------------------- | ||
| 482 : | ; | ||
| 483 : | ; void interpolate8x8_6tap_lowpass_h_x86_64(uint8_t const *dst, | ||
| 484 : | ; const uint8_t * const src, | ||
| 485 : | ; const uint32_t stride, | ||
| 486 : | ; const uint32_t rounding); | ||
| 487 : | ; | ||
| 488 : | ;----------------------------------------------------------------------------- | ||
| 489 : | |||
| 490 : | %macro LOWPASS_6TAP_H_MMX 0 | ||
| 491 : | movq mm0, [rax] | ||
| 492 : | movq mm2, [rax+1] | ||
| 493 : | |||
| 494 : | movq mm1, mm0 | ||
| 495 : | movq mm3, mm2 | ||
| 496 : | |||
| 497 : | punpcklbw mm0, mm7 | ||
| 498 : | punpcklbw mm2, mm7 | ||
| 499 : | |||
| 500 : | punpckhbw mm1, mm7 | ||
| 501 : | punpckhbw mm3, mm7 | ||
| 502 : | |||
| 503 : | paddw mm0, mm2 | ||
| 504 : | paddw mm1, mm3 | ||
| 505 : | |||
| 506 : | psllw mm0, 2 | ||
| 507 : | psllw mm1, 2 | ||
| 508 : | |||
| 509 : | movq mm2, [rax-1] | ||
| 510 : | movq mm4, [rax+2] | ||
| 511 : | |||
| 512 : | movq mm3, mm2 | ||
| 513 : | movq mm5, mm4 | ||
| 514 : | |||
| 515 : | punpcklbw mm2, mm7 | ||
| 516 : | punpcklbw mm4, mm7 | ||
| 517 : | |||
| 518 : | punpckhbw mm3, mm7 | ||
| 519 : | punpckhbw mm5, mm7 | ||
| 520 : | |||
| 521 : | paddw mm2, mm4 | ||
| 522 : | paddw mm3, mm5 | ||
| 523 : | |||
| 524 : | psubsw mm0, mm2 | ||
| 525 : | psubsw mm1, mm3 | ||
| 526 : | |||
| 527 : | pmullw mm0, [mmx_five wrt rip] | ||
| 528 : | pmullw mm1, [mmx_five wrt rip] | ||
| 529 : | |||
| 530 : | movq mm2, [rax-2] | ||
| 531 : | movq mm4, [rax+3] | ||
| 532 : | |||
| 533 : | movq mm3, mm2 | ||
| 534 : | movq mm5, mm4 | ||
| 535 : | |||
| 536 : | punpcklbw mm2, mm7 | ||
| 537 : | punpcklbw mm4, mm7 | ||
| 538 : | |||
| 539 : | punpckhbw mm3, mm7 | ||
| 540 : | punpckhbw mm5, mm7 | ||
| 541 : | |||
| 542 : | paddw mm2, mm4 | ||
| 543 : | paddw mm3, mm5 | ||
| 544 : | |||
| 545 : | paddsw mm0, mm2 | ||
| 546 : | paddsw mm1, mm3 | ||
| 547 : | |||
| 548 : | paddsw mm0, mm6 | ||
| 549 : | paddsw mm1, mm6 | ||
| 550 : | |||
| 551 : | psraw mm0, 5 | ||
| 552 : | psraw mm1, 5 | ||
| 553 : | |||
| 554 : | lea rax, [rax+rdx] | ||
| 555 : | packuswb mm0, mm1 | ||
| 556 : | movq [rcx], mm0 | ||
| 557 : | %endmacro | ||
| 558 : | |||
| 559 : | ALIGN 16 | ||
| 560 : | interpolate8x8_6tap_lowpass_h_x86_64: | ||
| 561 : | |||
| 562 : | mov rax, rcx ; rounding | ||
| 563 : | |||
| 564 : | lea r8, [rounding_lowpass_mmx wrt rip] | ||
| 565 : | movq mm6, [r8 + rax * 8] | ||
| 566 : | |||
| 567 : | ; rdx is stride | ||
| 568 : | mov rax, rsi ; src | ||
| 569 : | mov rcx, rdi ; dst | ||
| 570 : | |||
| 571 : | pxor mm7, mm7 | ||
| 572 : | |||
| 573 : | LOWPASS_6TAP_H_MMX | ||
| 574 : | lea rcx, [rcx+rdx] | ||
| 575 : | LOWPASS_6TAP_H_MMX | ||
| 576 : | lea rcx, [rcx+rdx] | ||
| 577 : | LOWPASS_6TAP_H_MMX | ||
| 578 : | lea rcx, [rcx+rdx] | ||
| 579 : | LOWPASS_6TAP_H_MMX | ||
| 580 : | lea rcx, [rcx+rdx] | ||
| 581 : | LOWPASS_6TAP_H_MMX | ||
| 582 : | lea rcx, [rcx+rdx] | ||
| 583 : | LOWPASS_6TAP_H_MMX | ||
| 584 : | lea rcx, [rcx+rdx] | ||
| 585 : | LOWPASS_6TAP_H_MMX | ||
| 586 : | lea rcx, [rcx+rdx] | ||
| 587 : | LOWPASS_6TAP_H_MMX | ||
| 588 : | |||
| 589 : | ret | ||
| 590 : | .endfunc | ||
| 591 : | |||
| 592 : | ;----------------------------------------------------------------------------- | ||
| 593 : | ; | ||
| 594 : | ; void interpolate8x8_6tap_lowpass_v_x86_64(uint8_t const *dst, | ||
| 595 : | ; const uint8_t * const src, | ||
| 596 : | ; const uint32_t stride, | ||
| 597 : | ; const uint32_t rounding); | ||
| 598 : | ; | ||
| 599 : | ;----------------------------------------------------------------------------- | ||
| 600 : | |||
| 601 : | %macro LOWPASS_6TAP_V_MMX 0 | ||
| 602 : | movq mm0, [rax] | ||
| 603 : | movq mm2, [rax+rdx] | ||
| 604 : | |||
| 605 : | movq mm1, mm0 | ||
| 606 : | movq mm3, mm2 | ||
| 607 : | |||
| 608 : | punpcklbw mm0, mm7 | ||
| 609 : | punpcklbw mm2, mm7 | ||
| 610 : | |||
| 611 : | punpckhbw mm1, mm7 | ||
| 612 : | punpckhbw mm3, mm7 | ||
| 613 : | |||
| 614 : | paddw mm0, mm2 | ||
| 615 : | paddw mm1, mm3 | ||
| 616 : | |||
| 617 : | psllw mm0, 2 | ||
| 618 : | psllw mm1, 2 | ||
| 619 : | |||
| 620 : | movq mm4, [rax+2*rdx] | ||
| 621 : | sub rax, rbx | ||
| 622 : | movq mm2, [rax+2*rdx] | ||
| 623 : | |||
| 624 : | movq mm3, mm2 | ||
| 625 : | movq mm5, mm4 | ||
| 626 : | |||
| 627 : | punpcklbw mm2, mm7 | ||
| 628 : | punpcklbw mm4, mm7 | ||
| 629 : | |||
| 630 : | punpckhbw mm3, mm7 | ||
| 631 : | punpckhbw mm5, mm7 | ||
| 632 : | |||
| 633 : | paddw mm2, mm4 | ||
| 634 : | paddw mm3, mm5 | ||
| 635 : | |||
| 636 : | psubsw mm0, mm2 | ||
| 637 : | psubsw mm1, mm3 | ||
| 638 : | |||
| 639 : | pmullw mm0, [mmx_five wrt rip] | ||
| 640 : | pmullw mm1, [mmx_five wrt rip] | ||
| 641 : | |||
| 642 : | movq mm2, [rax+rdx] | ||
| 643 : | movq mm4, [rax+2*rbx] | ||
| 644 : | |||
| 645 : | movq mm3, mm2 | ||
| 646 : | movq mm5, mm4 | ||
| 647 : | |||
| 648 : | punpcklbw mm2, mm7 | ||
| 649 : | punpcklbw mm4, mm7 | ||
| 650 : | |||
| 651 : | punpckhbw mm3, mm7 | ||
| 652 : | punpckhbw mm5, mm7 | ||
| 653 : | |||
| 654 : | paddw mm2, mm4 | ||
| 655 : | paddw mm3, mm5 | ||
| 656 : | |||
| 657 : | paddsw mm0, mm2 | ||
| 658 : | paddsw mm1, mm3 | ||
| 659 : | |||
| 660 : | paddsw mm0, mm6 | ||
| 661 : | paddsw mm1, mm6 | ||
| 662 : | |||
| 663 : | psraw mm0, 5 | ||
| 664 : | psraw mm1, 5 | ||
| 665 : | |||
| 666 : | lea rax, [rax+4*rdx] | ||
| 667 : | packuswb mm0, mm1 | ||
| 668 : | movq [rcx], mm0 | ||
| 669 : | %endmacro | ||
| 670 : | |||
| 671 : | ALIGN 16 | ||
| 672 : | interpolate8x8_6tap_lowpass_v_x86_64: | ||
| 673 : | |||
| 674 : | push rbx | ||
| 675 : | |||
| 676 : | mov rax, rcx ; rounding | ||
| 677 : | |||
| 678 : | lea r8, [rounding_lowpass_mmx wrt rip] | ||
| 679 : | movq mm6, [r8 + rax * 8] | ||
| 680 : | |||
| 681 : | ; rdx is stride | ||
| 682 : | mov rax, rsi ; src | ||
| 683 : | mov rcx, rdi ; dst | ||
| 684 : | |||
| 685 : | mov rbx, rdx | ||
| 686 : | shl rbx, 1 | ||
| 687 : | add rbx, rdx | ||
| 688 : | |||
| 689 : | pxor mm7, mm7 | ||
| 690 : | |||
| 691 : | LOWPASS_6TAP_V_MMX | ||
| 692 : | lea rcx, [rcx+rdx] | ||
| 693 : | LOWPASS_6TAP_V_MMX | ||
| 694 : | lea rcx, [rcx+rdx] | ||
| 695 : | LOWPASS_6TAP_V_MMX | ||
| 696 : | lea rcx, [rcx+rdx] | ||
| 697 : | LOWPASS_6TAP_V_MMX | ||
| 698 : | lea rcx, [rcx+rdx] | ||
| 699 : | LOWPASS_6TAP_V_MMX | ||
| 700 : | lea rcx, [rcx+rdx] | ||
| 701 : | LOWPASS_6TAP_V_MMX | ||
| 702 : | lea rcx, [rcx+rdx] | ||
| 703 : | LOWPASS_6TAP_V_MMX | ||
| 704 : | lea rcx, [rcx+rdx] | ||
| 705 : | LOWPASS_6TAP_V_MMX | ||
| 706 : | |||
| 707 : | pop rbx | ||
| 708 : | ret | ||
| 709 : | .endfunc |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |