Parent Directory
| 
Revision Log
Revision 1790 - (view) (download)
| 1 : | edgomez | 1586 | ;/**************************************************************************** |
| 2 : | ; * | ||
| 3 : | ; * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | ; * - 8<->16 bit transfer functions - | ||
| 5 : | ; * | ||
| 6 : | ; * Copyright (C) 2001 Peter Ross <pross@xvid.org> | ||
| 7 : | ; * 2001 Michael Militzer <isibaar@xvid.org> | ||
| 8 : | ; * 2002 Pascal Massimino <skal@planet-d.net> | ||
| 9 : | ; * 2004 Andre Werthmann <wertmann@aei.mpg.de> | ||
| 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 24 : | ; * | ||
| 25 : | Isibaar | 1790 | ; * $Id: mem_transfer_mmx.asm,v 1.2 2008-08-19 09:06:48 Isibaar Exp $ |
| 26 : | edgomez | 1586 | ; * |
| 27 : | ; ***************************************************************************/ | ||
| 28 : | |||
| 29 : | BITS 64 | ||
| 30 : | |||
| 31 : | %macro cglobal 1 | ||
| 32 : | %ifdef PREFIX | ||
| 33 : | %ifdef MARK_FUNCS | ||
| 34 : | global _%1:function %1.endfunc-%1 | ||
| 35 : | %define %1 _%1:function %1.endfunc-%1 | ||
| 36 : | %else | ||
| 37 : | global _%1 | ||
| 38 : | %define %1 _%1 | ||
| 39 : | %endif | ||
| 40 : | %else | ||
| 41 : | %ifdef MARK_FUNCS | ||
| 42 : | global %1:function %1.endfunc-%1 | ||
| 43 : | %else | ||
| 44 : | global %1 | ||
| 45 : | %endif | ||
| 46 : | %endif | ||
| 47 : | %endmacro | ||
| 48 : | |||
| 49 : | ;============================================================================= | ||
| 50 : | ; Read only data | ||
| 51 : | ;============================================================================= | ||
| 52 : | |||
| 53 : | %ifdef FORMAT_COFF | ||
| 54 : | SECTION .rodata | ||
| 55 : | %else | ||
| 56 : | SECTION .rodata align=16 | ||
| 57 : | %endif | ||
| 58 : | |||
| 59 : | ;============================================================================= | ||
| 60 : | ; Code | ||
| 61 : | ;============================================================================= | ||
| 62 : | |||
| 63 : | SECTION .text align=16 | ||
| 64 : | |||
| 65 : | cglobal transfer_8to16copy_x86_64 | ||
| 66 : | cglobal transfer_16to8copy_x86_64 | ||
| 67 : | cglobal transfer_8to16sub_x86_64 | ||
| 68 : | cglobal transfer_8to16subro_x86_64 | ||
| 69 : | cglobal transfer_8to16sub2_x86_64 | ||
| 70 : | cglobal transfer_8to16sub2ro_x86_64 | ||
| 71 : | cglobal transfer_16to8add_x86_64 | ||
| 72 : | cglobal transfer8x8_copy_x86_64 | ||
| 73 : | |||
| 74 : | ;----------------------------------------------------------------------------- | ||
| 75 : | ; | ||
| 76 : | ; void transfer_8to16copy_x86_64(int16_t * const dst, | ||
| 77 : | ; const uint8_t * const src, | ||
| 78 : | ; uint32_t stride); | ||
| 79 : | ; | ||
| 80 : | ;----------------------------------------------------------------------------- | ||
| 81 : | |||
| 82 : | %macro COPY_8_TO_16 1 | ||
| 83 : | movq mm0, [rax] | ||
| 84 : | movq mm1, [rax+rdx] | ||
| 85 : | movq mm2, mm0 | ||
| 86 : | movq mm3, mm1 | ||
| 87 : | punpcklbw mm0, mm7 | ||
| 88 : | movq [rcx+%1*32], mm0 | ||
| 89 : | punpcklbw mm1, mm7 | ||
| 90 : | movq [rcx+%1*32+16], mm1 | ||
| 91 : | punpckhbw mm2, mm7 | ||
| 92 : | punpckhbw mm3, mm7 | ||
| 93 : | lea rax, [rax+2*rdx] | ||
| 94 : | movq [rcx+%1*32+8], mm2 | ||
| 95 : | movq [rcx+%1*32+24], mm3 | ||
| 96 : | %endmacro | ||
| 97 : | |||
| 98 : | ALIGN 16 | ||
| 99 : | transfer_8to16copy_x86_64: | ||
| 100 : | ; rdx is Stride | ||
| 101 : | mov rax, rsi ; Src | ||
| 102 : | mov rcx, rdi ; Dst | ||
| 103 : | |||
| 104 : | pxor mm7, mm7 | ||
| 105 : | |||
| 106 : | COPY_8_TO_16 0 | ||
| 107 : | COPY_8_TO_16 1 | ||
| 108 : | COPY_8_TO_16 2 | ||
| 109 : | COPY_8_TO_16 3 | ||
| 110 : | ret | ||
| 111 : | .endfunc | ||
| 112 : | |||
| 113 : | ;----------------------------------------------------------------------------- | ||
| 114 : | ; | ||
| 115 : | ; void transfer_16to8copy_x86_64(uint8_t * const dst, | ||
| 116 : | ; const int16_t * const src, | ||
| 117 : | ; uint32_t stride); | ||
| 118 : | ; | ||
| 119 : | ;----------------------------------------------------------------------------- | ||
| 120 : | |||
| 121 : | %macro COPY_16_TO_8 1 | ||
| 122 : | movq mm0, [rax+%1*32] | ||
| 123 : | movq mm1, [rax+%1*32+8] | ||
| 124 : | packuswb mm0, mm1 | ||
| 125 : | movq [rcx], mm0 | ||
| 126 : | movq mm2, [rax+%1*32+16] | ||
| 127 : | movq mm3, [rax+%1*32+24] | ||
| 128 : | packuswb mm2, mm3 | ||
| 129 : | movq [rcx+rdx], mm2 | ||
| 130 : | %endmacro | ||
| 131 : | |||
| 132 : | ALIGN 16 | ||
| 133 : | transfer_16to8copy_x86_64: | ||
| 134 : | ; rdx is Stride | ||
| 135 : | mov rax, rsi ; Src | ||
| 136 : | mov rcx, rdi ; Dst | ||
| 137 : | |||
| 138 : | COPY_16_TO_8 0 | ||
| 139 : | lea rcx,[rcx+2*rdx] | ||
| 140 : | COPY_16_TO_8 1 | ||
| 141 : | lea rcx,[rcx+2*rdx] | ||
| 142 : | COPY_16_TO_8 2 | ||
| 143 : | lea rcx,[rcx+2*rdx] | ||
| 144 : | COPY_16_TO_8 3 | ||
| 145 : | ret | ||
| 146 : | .endfunc | ||
| 147 : | |||
| 148 : | ;----------------------------------------------------------------------------- | ||
| 149 : | ; | ||
| 150 : | ; void transfer_8to16sub_x86_64(int16_t * const dct, | ||
| 151 : | ; uint8_t * const cur, | ||
| 152 : | ; const uint8_t * const ref, | ||
| 153 : | ; const uint32_t stride); | ||
| 154 : | ; | ||
| 155 : | ;----------------------------------------------------------------------------- | ||
| 156 : | |||
| 157 : | ; when second argument == 1, reference (ebx) block is to current (eax) | ||
| 158 : | %macro COPY_8_TO_16_SUB 2 | ||
| 159 : | movq mm0, [rax] ; cur | ||
| 160 : | movq mm2, [rax+rdx] | ||
| 161 : | movq mm1, mm0 | ||
| 162 : | movq mm3, mm2 | ||
| 163 : | |||
| 164 : | punpcklbw mm0, mm7 | ||
| 165 : | punpcklbw mm2, mm7 | ||
| 166 : | movq mm4, [rbx] ; ref | ||
| 167 : | punpckhbw mm1, mm7 | ||
| 168 : | punpckhbw mm3, mm7 | ||
| 169 : | movq mm5, [rbx+rdx] ; ref | ||
| 170 : | |||
| 171 : | movq mm6, mm4 | ||
| 172 : | %if %2 == 1 | ||
| 173 : | movq [rax], mm4 | ||
| 174 : | movq [rax+rdx], mm5 | ||
| 175 : | %endif | ||
| 176 : | punpcklbw mm4, mm7 | ||
| 177 : | punpckhbw mm6, mm7 | ||
| 178 : | psubsw mm0, mm4 | ||
| 179 : | psubsw mm1, mm6 | ||
| 180 : | movq mm6, mm5 | ||
| 181 : | punpcklbw mm5, mm7 | ||
| 182 : | punpckhbw mm6, mm7 | ||
| 183 : | psubsw mm2, mm5 | ||
| 184 : | lea rax, [rax+2*rdx] | ||
| 185 : | psubsw mm3, mm6 | ||
| 186 : | lea rbx,[rbx+2*rdx] | ||
| 187 : | |||
| 188 : | movq [rcx+%1*32+ 0], mm0 ; dst | ||
| 189 : | movq [rcx+%1*32+ 8], mm1 | ||
| 190 : | movq [rcx+%1*32+16], mm2 | ||
| 191 : | movq [rcx+%1*32+24], mm3 | ||
| 192 : | %endmacro | ||
| 193 : | |||
| 194 : | ALIGN 16 | ||
| 195 : | transfer_8to16sub_x86_64: | ||
| 196 : | push rbx | ||
| 197 : | |||
| 198 : | mov rax, rsi ; Cur | ||
| 199 : | mov rbx, rdx ; Ref | ||
| 200 : | mov rdx, rcx ; Stride | ||
| 201 : | mov rcx, rdi ; Dst | ||
| 202 : | |||
| 203 : | pxor mm7, mm7 | ||
| 204 : | |||
| 205 : | COPY_8_TO_16_SUB 0, 1 | ||
| 206 : | COPY_8_TO_16_SUB 1, 1 | ||
| 207 : | COPY_8_TO_16_SUB 2, 1 | ||
| 208 : | COPY_8_TO_16_SUB 3, 1 | ||
| 209 : | |||
| 210 : | pop rbx | ||
| 211 : | ret | ||
| 212 : | .endfunc | ||
| 213 : | |||
| 214 : | ALIGN 16 | ||
| 215 : | transfer_8to16subro_x86_64: | ||
| 216 : | push rbx | ||
| 217 : | |||
| 218 : | mov rax, rsi ; Cur | ||
| 219 : | mov rbx, rdx ; Ref | ||
| 220 : | mov rdx, rcx ; Stride | ||
| 221 : | mov rcx, rdi ; Dst | ||
| 222 : | |||
| 223 : | pxor mm7, mm7 | ||
| 224 : | |||
| 225 : | COPY_8_TO_16_SUB 0, 0 | ||
| 226 : | COPY_8_TO_16_SUB 1, 0 | ||
| 227 : | COPY_8_TO_16_SUB 2, 0 | ||
| 228 : | COPY_8_TO_16_SUB 3, 0 | ||
| 229 : | |||
| 230 : | pop rbx | ||
| 231 : | ret | ||
| 232 : | .endfunc | ||
| 233 : | |||
| 234 : | ;----------------------------------------------------------------------------- | ||
| 235 : | ; | ||
| 236 : | ; void transfer_8to16sub2_x86_64(int16_t * const dct, | ||
| 237 : | ; uint8_t * const cur, | ||
| 238 : | ; const uint8_t * ref1, | ||
| 239 : | ; const uint8_t * ref2, | ||
| 240 : | ; const uint32_t stride) | ||
| 241 : | ; | ||
| 242 : | ;----------------------------------------------------------------------------- | ||
| 243 : | |||
| 244 : | %macro COPY_8_TO_16_SUB2_SSE 1 | ||
| 245 : | movq mm0, [rax] ; cur | ||
| 246 : | movq mm2, [rax+rdx] | ||
| 247 : | movq mm1, mm0 | ||
| 248 : | movq mm3, mm2 | ||
| 249 : | |||
| 250 : | punpcklbw mm0, mm7 | ||
| 251 : | punpcklbw mm2, mm7 | ||
| 252 : | movq mm4, [rbx] ; ref1 | ||
| 253 : | pavgb mm4, [rsi] ; ref2 | ||
| 254 : | movq [rax], mm4 | ||
| 255 : | punpckhbw mm1, mm7 | ||
| 256 : | punpckhbw mm3, mm7 | ||
| 257 : | movq mm5, [rbx+rdx] ; ref | ||
| 258 : | pavgb mm5, [rsi+rdx] ; ref2 | ||
| 259 : | movq [rax+rdx], mm5 | ||
| 260 : | |||
| 261 : | movq mm6, mm4 | ||
| 262 : | punpcklbw mm4, mm7 | ||
| 263 : | punpckhbw mm6, mm7 | ||
| 264 : | psubsw mm0, mm4 | ||
| 265 : | psubsw mm1, mm6 | ||
| 266 : | lea rsi, [rsi+2*rdx] | ||
| 267 : | movq mm6, mm5 | ||
| 268 : | punpcklbw mm5, mm7 | ||
| 269 : | punpckhbw mm6, mm7 | ||
| 270 : | psubsw mm2, mm5 | ||
| 271 : | lea rax, [rax+2*rdx] | ||
| 272 : | psubsw mm3, mm6 | ||
| 273 : | lea rbx, [rbx+2*rdx] | ||
| 274 : | |||
| 275 : | movq [rcx+%1*32+ 0], mm0 ; dst | ||
| 276 : | movq [rcx+%1*32+ 8], mm1 | ||
| 277 : | movq [rcx+%1*32+16], mm2 | ||
| 278 : | movq [rcx+%1*32+24], mm3 | ||
| 279 : | %endmacro | ||
| 280 : | |||
| 281 : | ALIGN 16 | ||
| 282 : | transfer_8to16sub2_x86_64: | ||
| 283 : | push rbx | ||
| 284 : | |||
| 285 : | mov rax, rsi ; Cur | ||
| 286 : | mov rbx, rdx ; Ref1 | ||
| 287 : | mov rdx, r8 ; Stride | ||
| 288 : | mov rsi, rcx ; Ref2 | ||
| 289 : | mov rcx, rdi ; Dst | ||
| 290 : | |||
| 291 : | pxor mm7, mm7 | ||
| 292 : | |||
| 293 : | COPY_8_TO_16_SUB2_SSE 0 | ||
| 294 : | COPY_8_TO_16_SUB2_SSE 1 | ||
| 295 : | COPY_8_TO_16_SUB2_SSE 2 | ||
| 296 : | COPY_8_TO_16_SUB2_SSE 3 | ||
| 297 : | |||
| 298 : | pop rbx | ||
| 299 : | ret | ||
| 300 : | .endfunc | ||
| 301 : | |||
| 302 : | ;----------------------------------------------------------------------------- | ||
| 303 : | ; | ||
| 304 : | ; void transfer_8to16sub2ro_x86_64(int16_t * const dct, | ||
| 305 : | ; const uint8_t * const cur, | ||
| 306 : | ; const uint8_t * ref1, | ||
| 307 : | ; const uint8_t * ref2, | ||
| 308 : | ; const uint32_t stride) | ||
| 309 : | ; | ||
| 310 : | ;----------------------------------------------------------------------------- | ||
| 311 : | |||
| 312 : | %macro COPY_8_TO_16_SUB2RO_SSE 1 | ||
| 313 : | movq mm0, [rsi] ; cur | ||
| 314 : | movq mm2, [rsi+r8] | ||
| 315 : | movq mm1, mm0 | ||
| 316 : | movq mm3, mm2 | ||
| 317 : | |||
| 318 : | punpcklbw mm0, mm7 | ||
| 319 : | punpcklbw mm2, mm7 | ||
| 320 : | movq mm4, [rdx] ; ref1 | ||
| 321 : | pavgb mm4, [rcx] ; ref2 | ||
| 322 : | punpckhbw mm1, mm7 | ||
| 323 : | punpckhbw mm3, mm7 | ||
| 324 : | movq mm5, [rdx+r8] ; ref | ||
| 325 : | pavgb mm5, [rcx+r8] ; ref2 | ||
| 326 : | |||
| 327 : | movq mm6, mm4 | ||
| 328 : | punpcklbw mm4, mm7 | ||
| 329 : | punpckhbw mm6, mm7 | ||
| 330 : | psubsw mm0, mm4 | ||
| 331 : | psubsw mm1, mm6 | ||
| 332 : | lea rcx, [rcx+2*r8] | ||
| 333 : | movq mm6, mm5 | ||
| 334 : | punpcklbw mm5, mm7 | ||
| 335 : | punpckhbw mm6, mm7 | ||
| 336 : | psubsw mm2, mm5 | ||
| 337 : | lea rsi, [rsi+2*r8] | ||
| 338 : | psubsw mm3, mm6 | ||
| 339 : | lea rdx, [rdx+2*r8] | ||
| 340 : | |||
| 341 : | movq [rdi+%1*32+ 0], mm0 ; dst | ||
| 342 : | movq [rdi+%1*32+ 8], mm1 | ||
| 343 : | movq [rdi+%1*32+16], mm2 | ||
| 344 : | movq [rdi+%1*32+24], mm3 | ||
| 345 : | %endmacro | ||
| 346 : | |||
| 347 : | ALIGN 16 | ||
| 348 : | transfer_8to16sub2ro_x86_64: | ||
| 349 : | pxor mm7, mm7 | ||
| 350 : | |||
| 351 : | COPY_8_TO_16_SUB2RO_SSE 0 | ||
| 352 : | COPY_8_TO_16_SUB2RO_SSE 1 | ||
| 353 : | COPY_8_TO_16_SUB2RO_SSE 2 | ||
| 354 : | COPY_8_TO_16_SUB2RO_SSE 3 | ||
| 355 : | |||
| 356 : | ret | ||
| 357 : | .endfunc | ||
| 358 : | |||
| 359 : | ;----------------------------------------------------------------------------- | ||
| 360 : | ; | ||
| 361 : | ; void transfer_16to8add_x86_64(uint8_t * const dst, | ||
| 362 : | ; const int16_t * const src, | ||
| 363 : | ; uint32_t stride); | ||
| 364 : | ; | ||
| 365 : | ;----------------------------------------------------------------------------- | ||
| 366 : | |||
| 367 : | %macro COPY_16_TO_8_ADD 1 | ||
| 368 : | movq mm0, [rcx] | ||
| 369 : | movq mm2, [rcx+rdx] | ||
| 370 : | movq mm1, mm0 | ||
| 371 : | movq mm3, mm2 | ||
| 372 : | punpcklbw mm0, mm7 | ||
| 373 : | punpcklbw mm2, mm7 | ||
| 374 : | punpckhbw mm1, mm7 | ||
| 375 : | punpckhbw mm3, mm7 | ||
| 376 : | paddsw mm0, [rax+%1*32+ 0] | ||
| 377 : | paddsw mm1, [rax+%1*32+ 8] | ||
| 378 : | paddsw mm2, [rax+%1*32+16] | ||
| 379 : | paddsw mm3, [rax+%1*32+24] | ||
| 380 : | packuswb mm0, mm1 | ||
| 381 : | movq [rcx], mm0 | ||
| 382 : | packuswb mm2, mm3 | ||
| 383 : | movq [rcx+rdx], mm2 | ||
| 384 : | %endmacro | ||
| 385 : | |||
| 386 : | |||
| 387 : | ALIGN 16 | ||
| 388 : | transfer_16to8add_x86_64: | ||
| 389 : | ; rdx is Stride | ||
| 390 : | mov rax, rsi ; Src | ||
| 391 : | mov rcx, rdi ; Dst | ||
| 392 : | |||
| 393 : | pxor mm7, mm7 | ||
| 394 : | |||
| 395 : | COPY_16_TO_8_ADD 0 | ||
| 396 : | lea rcx,[rcx+2*rdx] | ||
| 397 : | COPY_16_TO_8_ADD 1 | ||
| 398 : | lea rcx,[rcx+2*rdx] | ||
| 399 : | COPY_16_TO_8_ADD 2 | ||
| 400 : | lea rcx,[rcx+2*rdx] | ||
| 401 : | COPY_16_TO_8_ADD 3 | ||
| 402 : | ret | ||
| 403 : | .endfunc | ||
| 404 : | |||
| 405 : | ;----------------------------------------------------------------------------- | ||
| 406 : | ; | ||
| 407 : | ; void transfer8x8_copy_x86_64(uint8_t * const dst, | ||
| 408 : | ; const uint8_t * const src, | ||
| 409 : | ; const uint32_t stride); | ||
| 410 : | ; | ||
| 411 : | ; | ||
| 412 : | ;----------------------------------------------------------------------------- | ||
| 413 : | |||
| 414 : | %macro COPY_8_TO_8 0 | ||
| 415 : | movq mm0, [rax] | ||
| 416 : | movq mm1, [rax+rdx] | ||
| 417 : | movq [rcx], mm0 | ||
| 418 : | lea rax, [rax+2*rdx] | ||
| 419 : | movq [rcx+rdx], mm1 | ||
| 420 : | %endmacro | ||
| 421 : | |||
| 422 : | ALIGN 16 | ||
| 423 : | transfer8x8_copy_x86_64: | ||
| 424 : | ; rdx is Stride | ||
| 425 : | mov rax, rsi ; Src | ||
| 426 : | mov rcx, rdi ; Dst | ||
| 427 : | |||
| 428 : | COPY_8_TO_8 | ||
| 429 : | lea rcx,[rcx+2*rdx] | ||
| 430 : | COPY_8_TO_8 | ||
| 431 : | lea rcx,[rcx+2*rdx] | ||
| 432 : | COPY_8_TO_8 | ||
| 433 : | lea rcx,[rcx+2*rdx] | ||
| 434 : | COPY_8_TO_8 | ||
| 435 : | ret | ||
| 436 : | .endfunc | ||
| 437 : | Isibaar | 1790 | |
| 438 : | %ifidn __OUTPUT_FORMAT__,elf | ||
| 439 : | section ".note.GNU-stack" noalloc noexec nowrite progbits | ||
| 440 : | %endif | ||
| 441 : |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |