Parent Directory
| 
Revision Log
Revision 430 - (view) (download)
| 1 : | chl | 430 | ;/***************************************************************************** |
| 2 : | Isibaar | 262 | ; * |
| 3 : | chl | 430 | ; * XVID MPEG-4 VIDEO CODEC |
| 4 : | ; * mmx sum of absolute difference | ||
| 5 : | Isibaar | 262 | ; * |
| 6 : | chl | 430 | ; * Copyright(C) 2002 Peter Ross <pross@xvid.org> |
| 7 : | Isibaar | 262 | ; * |
| 8 : | chl | 430 | ; * This program is an implementation of a part of one or more MPEG-4 |
| 9 : | ; * Video tools as specified in ISO/IEC 14496-2 standard. Those intending | ||
| 10 : | ; * to use this software module in hardware or software products are | ||
| 11 : | ; * advised that its use may infringe existing patents or copyrights, and | ||
| 12 : | ; * any such use would be at such party's own risk. The original | ||
| 13 : | ; * developer of this software module and his/her company, and subsequent | ||
| 14 : | ; * editors and their companies, will have no liability for use of this | ||
| 15 : | ; * software or modifications or derivatives thereof. | ||
| 16 : | Isibaar | 262 | ; * |
| 17 : | chl | 430 | ; * This program is free software; you can redistribute it and/or modify |
| 18 : | ; * it under the terms of the GNU General Public License as published by | ||
| 19 : | ; * the Free Software Foundation; either version 2 of the License, or | ||
| 20 : | ; * (at your option) any later version. | ||
| 21 : | Isibaar | 262 | ; * |
| 22 : | chl | 430 | ; * This program is distributed in the hope that it will be useful, |
| 23 : | ; * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 24 : | ; * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 25 : | ; * GNU General Public License for more details. | ||
| 26 : | Isibaar | 262 | ; * |
| 27 : | chl | 430 | ; * You should have received a copy of the GNU General Public License |
| 28 : | ; * along with this program; if not, write to the Free Software | ||
| 29 : | ; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 30 : | Isibaar | 262 | ; * |
| 31 : | chl | 430 | , ****************************************************************************/ |
| 32 : | Isibaar | 262 | |
| 33 : | bits 32 | ||
| 34 : | |||
| 35 : | %macro cglobal 1 | ||
| 36 : | suxen_drol | 329 | %ifdef PREFIX |
| 37 : | Isibaar | 262 | global _%1 |
| 38 : | %define %1 _%1 | ||
| 39 : | %else | ||
| 40 : | global %1 | ||
| 41 : | %endif | ||
| 42 : | %endmacro | ||
| 43 : | |||
| 44 : | section .data | ||
| 45 : | |||
| 46 : | align 16 | ||
| 47 : | mmx_one times 4 dw 1 | ||
| 48 : | |||
| 49 : | section .text | ||
| 50 : | |||
| 51 : | cglobal sad16_mmx | ||
| 52 : | cglobal sad8_mmx | ||
| 53 : | cglobal sad16bi_mmx | ||
| 54 : | cglobal sad8bi_mmx | ||
| 55 : | cglobal dev16_mmx | ||
| 56 : | |||
| 57 : | ;=========================================================================== | ||
| 58 : | ; | ||
| 59 : | ; uint32_t sad16_mmx(const uint8_t * const cur, | ||
| 60 : | ; const uint8_t * const ref, | ||
| 61 : | ; const uint32_t stride, | ||
| 62 : | suxen_drol | 329 | ; const uint32_t best_sad); |
| 63 : | ; | ||
| 64 : | Isibaar | 262 | ; (early termination ignore; slows this down) |
| 65 : | ; | ||
| 66 : | ;=========================================================================== | ||
| 67 : | |||
| 68 : | %macro SAD_16x16_MMX 0 | ||
| 69 : | movq mm0, [eax] | ||
| 70 : | movq mm1, [edx] | ||
| 71 : | |||
| 72 : | movq mm2, [eax+8] | ||
| 73 : | movq mm3, [edx+8] | ||
| 74 : | |||
| 75 : | movq mm4, mm0 | ||
| 76 : | psubusb mm0, mm1 | ||
| 77 : | lea eax,[eax+ecx] | ||
| 78 : | movq mm5, mm2 | ||
| 79 : | psubusb mm2, mm3 | ||
| 80 : | lea edx,[edx+ecx] | ||
| 81 : | |||
| 82 : | psubusb mm1, mm4 | ||
| 83 : | por mm0, mm1 | ||
| 84 : | psubusb mm3, mm5 | ||
| 85 : | por mm2, mm3 | ||
| 86 : | |||
| 87 : | movq mm1,mm0 | ||
| 88 : | movq mm3,mm2 | ||
| 89 : | |||
| 90 : | punpcklbw mm0,mm7 | ||
| 91 : | punpckhbw mm1,mm7 | ||
| 92 : | punpcklbw mm2,mm7 | ||
| 93 : | punpckhbw mm3,mm7 | ||
| 94 : | |||
| 95 : | paddusw mm0,mm1 | ||
| 96 : | paddusw mm6,mm0 | ||
| 97 : | paddusw mm2,mm3 | ||
| 98 : | paddusw mm6,mm2 | ||
| 99 : | %endmacro | ||
| 100 : | |||
| 101 : | align 16 | ||
| 102 : | sad16_mmx: | ||
| 103 : | |||
| 104 : | mov eax, [esp+ 4] ; Src1 | ||
| 105 : | mov edx, [esp+ 8] ; Src2 | ||
| 106 : | mov ecx, [esp+12] ; Stride | ||
| 107 : | |||
| 108 : | pxor mm6, mm6 ; accum | ||
| 109 : | pxor mm7, mm7 ; zero | ||
| 110 : | |||
| 111 : | SAD_16x16_MMX | ||
| 112 : | SAD_16x16_MMX | ||
| 113 : | SAD_16x16_MMX | ||
| 114 : | SAD_16x16_MMX | ||
| 115 : | SAD_16x16_MMX | ||
| 116 : | SAD_16x16_MMX | ||
| 117 : | SAD_16x16_MMX | ||
| 118 : | SAD_16x16_MMX | ||
| 119 : | |||
| 120 : | SAD_16x16_MMX | ||
| 121 : | SAD_16x16_MMX | ||
| 122 : | SAD_16x16_MMX | ||
| 123 : | SAD_16x16_MMX | ||
| 124 : | SAD_16x16_MMX | ||
| 125 : | SAD_16x16_MMX | ||
| 126 : | SAD_16x16_MMX | ||
| 127 : | SAD_16x16_MMX | ||
| 128 : | |||
| 129 : | pmaddwd mm6, [mmx_one] ; collapse | ||
| 130 : | movq mm7, mm6 | ||
| 131 : | psrlq mm7, 32 | ||
| 132 : | paddd mm6, mm7 | ||
| 133 : | |||
| 134 : | movd eax, mm6 | ||
| 135 : | |||
| 136 : | ret | ||
| 137 : | |||
| 138 : | |||
| 139 : | ;=========================================================================== | ||
| 140 : | ; | ||
| 141 : | ; uint32_t sad8_mmx(const uint8_t * const cur, | ||
| 142 : | ; const uint8_t * const ref, | ||
| 143 : | ; const uint32_t stride); | ||
| 144 : | ; | ||
| 145 : | ;=========================================================================== | ||
| 146 : | |||
| 147 : | %macro SAD_8x8_MMX 0 | ||
| 148 : | movq mm0, [eax] | ||
| 149 : | movq mm1, [edx] | ||
| 150 : | |||
| 151 : | movq mm2, [eax+ecx] | ||
| 152 : | movq mm3, [edx+ecx] | ||
| 153 : | |||
| 154 : | lea eax,[eax+2*ecx] | ||
| 155 : | lea edx,[edx+2*ecx] | ||
| 156 : | |||
| 157 : | movq mm4, mm0 | ||
| 158 : | psubusb mm0, mm1 | ||
| 159 : | movq mm5, mm2 | ||
| 160 : | psubusb mm2, mm3 | ||
| 161 : | |||
| 162 : | psubusb mm1, mm4 | ||
| 163 : | por mm0, mm1 | ||
| 164 : | psubusb mm3, mm5 | ||
| 165 : | por mm2, mm3 | ||
| 166 : | |||
| 167 : | movq mm1,mm0 | ||
| 168 : | movq mm3,mm2 | ||
| 169 : | |||
| 170 : | punpcklbw mm0,mm7 | ||
| 171 : | punpckhbw mm1,mm7 | ||
| 172 : | punpcklbw mm2,mm7 | ||
| 173 : | punpckhbw mm3,mm7 | ||
| 174 : | |||
| 175 : | paddusw mm0,mm1 | ||
| 176 : | paddusw mm6,mm0 | ||
| 177 : | paddusw mm2,mm3 | ||
| 178 : | paddusw mm6,mm2 | ||
| 179 : | %endmacro | ||
| 180 : | |||
| 181 : | align 16 | ||
| 182 : | sad8_mmx: | ||
| 183 : | |||
| 184 : | mov eax, [esp+ 4] ; Src1 | ||
| 185 : | mov edx, [esp+ 8] ; Src2 | ||
| 186 : | mov ecx, [esp+12] ; Stride | ||
| 187 : | |||
| 188 : | pxor mm6, mm6 ; accum | ||
| 189 : | pxor mm7, mm7 ; zero | ||
| 190 : | |||
| 191 : | SAD_8x8_MMX | ||
| 192 : | SAD_8x8_MMX | ||
| 193 : | SAD_8x8_MMX | ||
| 194 : | SAD_8x8_MMX | ||
| 195 : | |||
| 196 : | pmaddwd mm6, [mmx_one] ; collapse | ||
| 197 : | movq mm7, mm6 | ||
| 198 : | psrlq mm7, 32 | ||
| 199 : | paddd mm6, mm7 | ||
| 200 : | |||
| 201 : | movd eax, mm6 | ||
| 202 : | |||
| 203 : | ret | ||
| 204 : | |||
| 205 : | |||
| 206 : | |||
| 207 : | |||
| 208 : | |||
| 209 : | ;=========================================================================== | ||
| 210 : | ; | ||
| 211 : | ; uint32_t sad16bi_mmx(const uint8_t * const cur, | ||
| 212 : | ; const uint8_t * const ref1, | ||
| 213 : | suxen_drol | 329 | ; const uint8_t * const ref2, |
| 214 : | ; const uint32_t stride); | ||
| 215 : | ; | ||
| 216 : | ;=========================================================================== | ||
| 217 : | %macro SADBI_16x16_MMX 2 ; SADBI_16x16_MMX( int_ptr_offset, bool_increment_ptr ); | ||
| 218 : | |||
| 219 : | movq mm0, [edx+%1] | ||
| 220 : | movq mm2, [ebx+%1] | ||
| 221 : | movq mm1, mm0 | ||
| 222 : | movq mm3, mm2 | ||
| 223 : | |||
| 224 : | %if %2 != 0 | ||
| 225 : | add edx, ecx | ||
| 226 : | %endif | ||
| 227 : | |||
| 228 : | punpcklbw mm0, mm7 | ||
| 229 : | punpckhbw mm1, mm7 | ||
| 230 : | punpcklbw mm2, mm7 | ||
| 231 : | punpckhbw mm3, mm7 | ||
| 232 : | |||
| 233 : | %if %2 != 0 | ||
| 234 : | add ebx, ecx | ||
| 235 : | %endif | ||
| 236 : | |||
| 237 : | paddusw mm0, mm2 ; mm01 = ref1 + ref2 | ||
| 238 : | paddusw mm1, mm3 | ||
| 239 : | paddusw mm0, [mmx_one] ; mm01 += 1 | ||
| 240 : | paddusw mm1, [mmx_one] | ||
| 241 : | psrlw mm0, 1 ; mm01 >>= 1 | ||
| 242 : | psrlw mm1, 1 | ||
| 243 : | |||
| 244 : | movq mm2, [eax+%1] | ||
| 245 : | movq mm3, mm2 | ||
| 246 : | punpcklbw mm2, mm7 ; mm23 = src | ||
| 247 : | punpckhbw mm3, mm7 | ||
| 248 : | |||
| 249 : | %if %2 != 0 | ||
| 250 : | add eax, ecx | ||
| 251 : | %endif | ||
| 252 : | |||
| 253 : | movq mm4, mm0 | ||
| 254 : | movq mm5, mm1 | ||
| 255 : | psubusw mm0, mm2 | ||
| 256 : | psubusw mm1, mm3 | ||
| 257 : | psubusw mm2, mm4 | ||
| 258 : | psubusw mm3, mm5 | ||
| 259 : | por mm0, mm2 ; mm01 = ABS(mm01 - mm23) | ||
| 260 : | por mm1, mm3 | ||
| 261 : | |||
| 262 : | paddusw mm6,mm0 ; mm6 += mm01 | ||
| 263 : | paddusw mm6,mm1 | ||
| 264 : | |||
| 265 : | %endmacro | ||
| 266 : | |||
| 267 : | align 16 | ||
| 268 : | sad16bi_mmx: | ||
| 269 : | push ebx | ||
| 270 : | mov eax, [esp+4+ 4] ; Src | ||
| 271 : | mov edx, [esp+4+ 8] ; Ref1 | ||
| 272 : | mov ebx, [esp+4+12] ; Ref2 | ||
| 273 : | mov ecx, [esp+4+16] ; Stride | ||
| 274 : | |||
| 275 : | pxor mm6, mm6 ; accum2 | ||
| 276 : | pxor mm7, mm7 | ||
| 277 : | .Loop | ||
| 278 : | SADBI_16x16_MMX 0, 0 | ||
| 279 : | SADBI_16x16_MMX 8, 1 | ||
| 280 : | SADBI_16x16_MMX 0, 0 | ||
| 281 : | SADBI_16x16_MMX 8, 1 | ||
| 282 : | SADBI_16x16_MMX 0, 0 | ||
| 283 : | SADBI_16x16_MMX 8, 1 | ||
| 284 : | SADBI_16x16_MMX 0, 0 | ||
| 285 : | SADBI_16x16_MMX 8, 1 | ||
| 286 : | SADBI_16x16_MMX 0, 0 | ||
| 287 : | SADBI_16x16_MMX 8, 1 | ||
| 288 : | SADBI_16x16_MMX 0, 0 | ||
| 289 : | SADBI_16x16_MMX 8, 1 | ||
| 290 : | SADBI_16x16_MMX 0, 0 | ||
| 291 : | SADBI_16x16_MMX 8, 1 | ||
| 292 : | SADBI_16x16_MMX 0, 0 | ||
| 293 : | SADBI_16x16_MMX 8, 1 | ||
| 294 : | |||
| 295 : | SADBI_16x16_MMX 0, 0 | ||
| 296 : | SADBI_16x16_MMX 8, 1 | ||
| 297 : | SADBI_16x16_MMX 0, 0 | ||
| 298 : | SADBI_16x16_MMX 8, 1 | ||
| 299 : | SADBI_16x16_MMX 0, 0 | ||
| 300 : | SADBI_16x16_MMX 8, 1 | ||
| 301 : | SADBI_16x16_MMX 0, 0 | ||
| 302 : | SADBI_16x16_MMX 8, 1 | ||
| 303 : | SADBI_16x16_MMX 0, 0 | ||
| 304 : | SADBI_16x16_MMX 8, 1 | ||
| 305 : | SADBI_16x16_MMX 0, 0 | ||
| 306 : | SADBI_16x16_MMX 8, 1 | ||
| 307 : | SADBI_16x16_MMX 0, 0 | ||
| 308 : | SADBI_16x16_MMX 8, 1 | ||
| 309 : | SADBI_16x16_MMX 0, 0 | ||
| 310 : | SADBI_16x16_MMX 8, 1 | ||
| 311 : | |||
| 312 : | pmaddwd mm6, [mmx_one] ; collapse | ||
| 313 : | movq mm7, mm6 | ||
| 314 : | psrlq mm7, 32 | ||
| 315 : | paddd mm6, mm7 | ||
| 316 : | |||
| 317 : | movd eax, mm6 | ||
| 318 : | pop ebx | ||
| 319 : | ret | ||
| 320 : | |||
| 321 : | ;=========================================================================== | ||
| 322 : | ; | ||
| 323 : | ; uint32_t sad8bi_mmx(const uint8_t * const cur, | ||
| 324 : | ; const uint8_t * const ref1, | ||
| 325 : | ; const uint8_t * const ref2, | ||
| 326 : | ; const uint32_t stride); | ||
| 327 : | ; | ||
| 328 : | ;=========================================================================== | ||
| 329 : | align 16 | ||
| 330 : | sad8bi_mmx: | ||
| 331 : | push ebx | ||
| 332 : | mov eax, [esp+4+ 4] ; Src | ||
| 333 : | mov edx, [esp+4+ 8] ; Ref1 | ||
| 334 : | mov ebx, [esp+4+12] ; Ref2 | ||
| 335 : | mov ecx, [esp+4+16] ; Stride | ||
| 336 : | |||
| 337 : | pxor mm6, mm6 ; accum2 | ||
| 338 : | pxor mm7, mm7 | ||
| 339 : | .Loop | ||
| 340 : | SADBI_16x16_MMX 0, 1 | ||
| 341 : | SADBI_16x16_MMX 0, 1 | ||
| 342 : | SADBI_16x16_MMX 0, 1 | ||
| 343 : | SADBI_16x16_MMX 0, 1 | ||
| 344 : | SADBI_16x16_MMX 0, 1 | ||
| 345 : | SADBI_16x16_MMX 0, 1 | ||
| 346 : | SADBI_16x16_MMX 0, 1 | ||
| 347 : | SADBI_16x16_MMX 0, 1 | ||
| 348 : | |||
| 349 : | pmaddwd mm6, [mmx_one] ; collapse | ||
| 350 : | movq mm7, mm6 | ||
| 351 : | psrlq mm7, 32 | ||
| 352 : | paddd mm6, mm7 | ||
| 353 : | |||
| 354 : | movd eax, mm6 | ||
| 355 : | pop ebx | ||
| 356 : | ret | ||
| 357 : | |||
| 358 : | |||
| 359 : | |||
| 360 : | |||
| 361 : | ;=========================================================================== | ||
| 362 : | ; | ||
| 363 : | ; uint32_t dev16_mmx(const uint8_t * const cur, | ||
| 364 : | ; const uint32_t stride); | ||
| 365 : | ; | ||
| 366 : | ;=========================================================================== | ||
| 367 : | |||
| 368 : | %macro MEAN_16x16_MMX 0 | ||
| 369 : | Isibaar | 262 | movq mm0, [eax] |
| 370 : | movq mm2, [eax+8] | ||
| 371 : | lea eax,[eax+ecx] | ||
| 372 : | movq mm1, mm0 | ||
| 373 : | movq mm3, mm2 | ||
| 374 : | punpcklbw mm0,mm7 | ||
| 375 : | punpcklbw mm2,mm7 | ||
| 376 : | punpckhbw mm1,mm7 | ||
| 377 : | punpckhbw mm3,mm7 | ||
| 378 : | paddw mm5, mm0 | ||
| 379 : | paddw mm6, mm1 | ||
| 380 : | paddw mm5, mm2 | ||
| 381 : | paddw mm6, mm3 | ||
| 382 : | %endmacro | ||
| 383 : | |||
| 384 : | %macro ABS_16x16_MMX 0 | ||
| 385 : | movq mm0, [eax] | ||
| 386 : | movq mm2, [eax+8] | ||
| 387 : | lea eax,[eax+ecx] | ||
| 388 : | movq mm1, mm0 | ||
| 389 : | movq mm3, mm2 | ||
| 390 : | punpcklbw mm0, mm7 | ||
| 391 : | punpcklbw mm2, mm7 | ||
| 392 : | punpckhbw mm1, mm7 | ||
| 393 : | punpckhbw mm3, mm7 | ||
| 394 : | movq mm4, mm6 | ||
| 395 : | psubusw mm4, mm0 | ||
| 396 : | |||
| 397 : | psubusw mm0, mm6 | ||
| 398 : | por mm0, mm4 | ||
| 399 : | movq mm4, mm6 | ||
| 400 : | psubusw mm4, mm1 | ||
| 401 : | psubusw mm1, mm6 | ||
| 402 : | por mm1, mm4 | ||
| 403 : | |||
| 404 : | movq mm4, mm6 | ||
| 405 : | psubusw mm4, mm2 | ||
| 406 : | psubusw mm2, mm6 | ||
| 407 : | por mm2, mm4 | ||
| 408 : | movq mm4, mm6 | ||
| 409 : | psubusw mm4, mm3 | ||
| 410 : | psubusw mm3, mm6 | ||
| 411 : | por mm3, mm4 | ||
| 412 : | |||
| 413 : | paddw mm0, mm1 | ||
| 414 : | paddw mm2, mm3 | ||
| 415 : | paddw mm5, mm0 | ||
| 416 : | paddw mm5, mm2 | ||
| 417 : | %endmacro | ||
| 418 : | |||
| 419 : | align 16 | ||
| 420 : | dev16_mmx: | ||
| 421 : | mov eax, [esp+ 4] ; Src | ||
| 422 : | mov ecx, [esp+ 8] ; Stride | ||
| 423 : | |||
| 424 : | pxor mm7, mm7 ; zero | ||
| 425 : | pxor mm5, mm5 ; accum1 | ||
| 426 : | pxor mm6, mm6 ; accum2 | ||
| 427 : | |||
| 428 : | MEAN_16x16_MMX | ||
| 429 : | MEAN_16x16_MMX | ||
| 430 : | MEAN_16x16_MMX | ||
| 431 : | MEAN_16x16_MMX | ||
| 432 : | MEAN_16x16_MMX | ||
| 433 : | MEAN_16x16_MMX | ||
| 434 : | MEAN_16x16_MMX | ||
| 435 : | MEAN_16x16_MMX | ||
| 436 : | |||
| 437 : | MEAN_16x16_MMX | ||
| 438 : | MEAN_16x16_MMX | ||
| 439 : | MEAN_16x16_MMX | ||
| 440 : | MEAN_16x16_MMX | ||
| 441 : | MEAN_16x16_MMX | ||
| 442 : | MEAN_16x16_MMX | ||
| 443 : | MEAN_16x16_MMX | ||
| 444 : | MEAN_16x16_MMX | ||
| 445 : | |||
| 446 : | paddusw mm6, mm5 | ||
| 447 : | pmaddwd mm6, [mmx_one] ; collapse | ||
| 448 : | movq mm5, mm6 | ||
| 449 : | psrlq mm5, 32 | ||
| 450 : | paddd mm6, mm5 | ||
| 451 : | |||
| 452 : | psllq mm6, 32 ; blank upper dword | ||
| 453 : | psrlq mm6, 32 + 8 ; /= (16*16) | ||
| 454 : | |||
| 455 : | punpckldq mm6, mm6 | ||
| 456 : | packssdw mm6, mm6 | ||
| 457 : | |||
| 458 : | ; mm6 contains the mean | ||
| 459 : | ; mm5 is the new accum | ||
| 460 : | |||
| 461 : | pxor mm5, mm5 | ||
| 462 : | mov eax, [esp+ 4] ; Src | ||
| 463 : | |||
| 464 : | ABS_16x16_MMX | ||
| 465 : | ABS_16x16_MMX | ||
| 466 : | ABS_16x16_MMX | ||
| 467 : | ABS_16x16_MMX | ||
| 468 : | ABS_16x16_MMX | ||
| 469 : | ABS_16x16_MMX | ||
| 470 : | ABS_16x16_MMX | ||
| 471 : | ABS_16x16_MMX | ||
| 472 : | |||
| 473 : | ABS_16x16_MMX | ||
| 474 : | ABS_16x16_MMX | ||
| 475 : | ABS_16x16_MMX | ||
| 476 : | ABS_16x16_MMX | ||
| 477 : | ABS_16x16_MMX | ||
| 478 : | ABS_16x16_MMX | ||
| 479 : | ABS_16x16_MMX | ||
| 480 : | ABS_16x16_MMX | ||
| 481 : | |||
| 482 : | pmaddwd mm5, [mmx_one] ; collapse | ||
| 483 : | movq mm6, mm5 | ||
| 484 : | psrlq mm6, 32 | ||
| 485 : | paddd mm6, mm5 | ||
| 486 : | |||
| 487 : | movd eax, mm6 | ||
| 488 : | chl | 430 | ret |
| 489 : |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |