40 #define OP_AVG(S,D) "pavgb " #S ", " #D " \n\t"
43 #define NORMALIZE_MMX(SHIFT) \
44 "paddw %%mm7, %%mm3 \n\t" \
45 "paddw %%mm7, %%mm4 \n\t" \
46 "psraw "SHIFT", %%mm3 \n\t" \
47 "psraw "SHIFT", %%mm4 \n\t"
49 #define TRANSFER_DO_PACK(OP) \
50 "packuswb %%mm4, %%mm3 \n\t" \
52 "movq %%mm3, (%2) \n\t"
54 #define TRANSFER_DONT_PACK(OP) \
57 "movq %%mm3, 0(%2) \n\t" \
58 "movq %%mm4, 8(%2) \n\t"
61 #define DO_UNPACK(reg) "punpcklbw %%mm0, " reg "\n\t"
62 #define DONT_UNPACK(reg)
65 #define LOAD_ROUNDER_MMX(ROUND) \
66 "movd "ROUND", %%mm7 \n\t" \
67 "punpcklwd %%mm7, %%mm7 \n\t" \
68 "punpckldq %%mm7, %%mm7 \n\t"
70 #define SHIFT2_LINE(OFF, R0,R1,R2,R3) \
71 "paddw %%mm"#R2", %%mm"#R1" \n\t" \
72 "movd (%0,%3), %%mm"#R0" \n\t" \
73 "pmullw %%mm6, %%mm"#R1" \n\t" \
74 "punpcklbw %%mm0, %%mm"#R0" \n\t" \
75 "movd (%0,%2), %%mm"#R3" \n\t" \
76 "psubw %%mm"#R0", %%mm"#R1" \n\t" \
77 "punpcklbw %%mm0, %%mm"#R3" \n\t" \
78 "paddw %%mm7, %%mm"#R1" \n\t" \
79 "psubw %%mm"#R3", %%mm"#R1" \n\t" \
80 "psraw %4, %%mm"#R1" \n\t" \
81 "movq %%mm"#R1", "#OFF"(%1) \n\t" \
85 static void vc1_put_ver_16b_shift2_mmx(int16_t *dst,
87 int rnd, int64_t shift)
90 "mov $3, %%"REG_c
" \n\t"
91 LOAD_ROUNDER_MMX(
"%5")
94 "movd (%0), %%mm2 \n\t"
96 "movd (%0), %%mm3 \n\t"
97 "punpcklbw %%mm0, %%mm2 \n\t"
98 "punpcklbw %%mm0, %%mm3 \n\t"
99 SHIFT2_LINE( 0, 1, 2, 3, 4)
100 SHIFT2_LINE( 24, 2, 3, 4, 1)
101 SHIFT2_LINE( 48, 3, 4, 1, 2)
102 SHIFT2_LINE( 72, 4, 1, 2, 3)
103 SHIFT2_LINE( 96, 1, 2, 3, 4)
104 SHIFT2_LINE(120, 2, 3, 4, 1)
105 SHIFT2_LINE(144, 3, 4, 1, 2)
106 SHIFT2_LINE(168, 4, 1, 2, 3)
111 : "+
r"(src), "+
r"(dst)
112 : "
r"(stride), "
r"(-2*stride),
113 "m"(shift), "m"(rnd), "
r"(9*stride-4)
122 #define VC1_HOR_16b_SHIFT2(OP, OPNAME)\
123 static void OPNAME ## vc1_hor_16b_shift2_mmx(uint8_t *dst, x86_reg stride,\
124 const int16_t *src, int rnd)\
129 rnd -= (-1+9+9-1)*1024; \
131 LOAD_ROUNDER_MMX("%4")\
132 "movq "MANGLE(ff_pw_128)", %%mm6\n\t"\
133 "movq "MANGLE(ff_pw_9)", %%mm5 \n\t"\
135 "movq 2*0+0(%1), %%mm1 \n\t"\
136 "movq 2*0+8(%1), %%mm2 \n\t"\
137 "movq 2*1+0(%1), %%mm3 \n\t"\
138 "movq 2*1+8(%1), %%mm4 \n\t"\
139 "paddw 2*3+0(%1), %%mm1 \n\t"\
140 "paddw 2*3+8(%1), %%mm2 \n\t"\
141 "paddw 2*2+0(%1), %%mm3 \n\t"\
142 "paddw 2*2+8(%1), %%mm4 \n\t"\
143 "pmullw %%mm5, %%mm3 \n\t"\
144 "pmullw %%mm5, %%mm4 \n\t"\
145 "psubw %%mm1, %%mm3 \n\t"\
146 "psubw %%mm2, %%mm4 \n\t"\
149 "paddw %%mm6, %%mm3 \n\t"\
150 "paddw %%mm6, %%mm4 \n\t"\
151 TRANSFER_DO_PACK(OP)\
156 : "+r"(h), "+r" (src), "+r" (dst)\
157 : "r"(stride), "m"(rnd)\
162 VC1_HOR_16b_SHIFT2(
OP_PUT, put_)
163 VC1_HOR_16b_SHIFT2(
OP_AVG, avg_)
170 #define VC1_SHIFT2(OP, OPNAME)\
171 static void OPNAME ## vc1_shift2_mmx(uint8_t *dst, const uint8_t *src,\
172 x86_reg stride, int rnd, x86_reg offset)\
176 "mov $8, %%"REG_c" \n\t"\
177 LOAD_ROUNDER_MMX("%5")\
178 "movq "MANGLE(ff_pw_9)", %%mm6\n\t"\
180 "movd 0(%0 ), %%mm3 \n\t"\
181 "movd 4(%0 ), %%mm4 \n\t"\
182 "movd 0(%0,%2), %%mm1 \n\t"\
183 "movd 4(%0,%2), %%mm2 \n\t"\
185 "punpcklbw %%mm0, %%mm3 \n\t"\
186 "punpcklbw %%mm0, %%mm4 \n\t"\
187 "punpcklbw %%mm0, %%mm1 \n\t"\
188 "punpcklbw %%mm0, %%mm2 \n\t"\
189 "paddw %%mm1, %%mm3 \n\t"\
190 "paddw %%mm2, %%mm4 \n\t"\
191 "movd 0(%0,%3), %%mm1 \n\t"\
192 "movd 4(%0,%3), %%mm2 \n\t"\
193 "pmullw %%mm6, %%mm3 \n\t" \
194 "pmullw %%mm6, %%mm4 \n\t" \
195 "punpcklbw %%mm0, %%mm1 \n\t"\
196 "punpcklbw %%mm0, %%mm2 \n\t"\
197 "psubw %%mm1, %%mm3 \n\t" \
198 "psubw %%mm2, %%mm4 \n\t" \
199 "movd 0(%0,%2), %%mm1 \n\t"\
200 "movd 4(%0,%2), %%mm2 \n\t"\
201 "punpcklbw %%mm0, %%mm1 \n\t"\
202 "punpcklbw %%mm0, %%mm2 \n\t"\
203 "psubw %%mm1, %%mm3 \n\t" \
204 "psubw %%mm2, %%mm4 \n\t" \
206 "packuswb %%mm4, %%mm3 \n\t"\
208 "movq %%mm3, (%1) \n\t"\
211 "dec %%"REG_c" \n\t"\
213 : "+r"(src), "+r"(dst)\
214 : "r"(offset), "r"(-2*offset), "g"(stride), "m"(rnd),\
216 : "%"REG_c, "memory"\
221 VC1_SHIFT2(OP_AVG, avg_)
233 #define MSPEL_FILTER13_CORE(UNPACK, MOVQ, A1, A2, A3, A4) \
234 MOVQ "*0+"A1", %%mm1 \n\t" \
235 MOVQ "*4+"A1", %%mm2 \n\t" \
238 "pmullw "MANGLE(ff_pw_3)", %%mm1\n\t" \
239 "pmullw "MANGLE(ff_pw_3)", %%mm2\n\t" \
240 MOVQ "*0+"A2", %%mm3 \n\t" \
241 MOVQ "*4+"A2", %%mm4 \n\t" \
244 "pmullw %%mm6, %%mm3 \n\t" \
245 "pmullw %%mm6, %%mm4 \n\t" \
246 "psubw %%mm1, %%mm3 \n\t" \
247 "psubw %%mm2, %%mm4 \n\t" \
248 MOVQ "*0+"A4", %%mm1 \n\t" \
249 MOVQ "*4+"A4", %%mm2 \n\t" \
252 "psllw $2, %%mm1 \n\t" \
253 "psllw $2, %%mm2 \n\t" \
254 "psubw %%mm1, %%mm3 \n\t" \
255 "psubw %%mm2, %%mm4 \n\t" \
256 MOVQ "*0+"A3", %%mm1 \n\t" \
257 MOVQ "*4+"A3", %%mm2 \n\t" \
260 "pmullw %%mm5, %%mm1 \n\t" \
261 "pmullw %%mm5, %%mm2 \n\t" \
262 "paddw %%mm1, %%mm3 \n\t" \
263 "paddw %%mm2, %%mm4 \n\t"
273 #define MSPEL_FILTER13_VER_16B(NAME, A1, A2, A3, A4) \
275 vc1_put_ver_16b_ ## NAME ## _mmx(int16_t *dst, const uint8_t *src, \
276 x86_reg src_stride, \
277 int rnd, int64_t shift) \
282 LOAD_ROUNDER_MMX("%5") \
283 "movq "MANGLE(ff_pw_53)", %%mm5\n\t" \
284 "movq "MANGLE(ff_pw_18)", %%mm6\n\t" \
287 MSPEL_FILTER13_CORE(DO_UNPACK, "movd 1", A1, A2, A3, A4) \
288 NORMALIZE_MMX("%6") \
289 TRANSFER_DONT_PACK(OP_PUT) \
291 "movd 8+"A1", %%mm1 \n\t" \
293 "movq %%mm1, %%mm3 \n\t" \
294 "paddw %%mm1, %%mm1 \n\t" \
295 "paddw %%mm3, %%mm1 \n\t" \
296 "movd 8+"A2", %%mm3 \n\t" \
298 "pmullw %%mm6, %%mm3 \n\t" \
299 "psubw %%mm1, %%mm3 \n\t" \
300 "movd 8+"A3", %%mm1 \n\t" \
302 "pmullw %%mm5, %%mm1 \n\t" \
303 "paddw %%mm1, %%mm3 \n\t" \
304 "movd 8+"A4", %%mm1 \n\t" \
306 "psllw $2, %%mm1 \n\t" \
307 "psubw %%mm1, %%mm3 \n\t" \
308 "paddw %%mm7, %%mm3 \n\t" \
309 "psraw %6, %%mm3 \n\t" \
310 "movq %%mm3, 16(%2) \n\t" \
315 : "+r"(h), "+r" (src), "+r" (dst) \
316 : "r"(src_stride), "r"(3*src_stride), \
317 "m"(rnd), "m"(shift) \
329 #define MSPEL_FILTER13_HOR_16B(NAME, A1, A2, A3, A4, OP, OPNAME) \
331 OPNAME ## vc1_hor_16b_ ## NAME ## _mmx(uint8_t *dst, x86_reg stride, \
332 const int16_t *src, int rnd) \
336 rnd -= (-4+58+13-3)*256; \
338 LOAD_ROUNDER_MMX("%4") \
339 "movq "MANGLE(ff_pw_18)", %%mm6 \n\t" \
340 "movq "MANGLE(ff_pw_53)", %%mm5 \n\t" \
343 MSPEL_FILTER13_CORE(DONT_UNPACK, "movq 2", A1, A2, A3, A4) \
344 NORMALIZE_MMX("$7") \
346 "paddw "MANGLE(ff_pw_128)", %%mm3 \n\t" \
347 "paddw "MANGLE(ff_pw_128)", %%mm4 \n\t" \
348 TRANSFER_DO_PACK(OP) \
353 : "+r"(h), "+r" (src), "+r" (dst) \
354 : "r"(stride), "m"(rnd) \
367 #define MSPEL_FILTER13_8B(NAME, A1, A2, A3, A4, OP, OPNAME) \
369 OPNAME ## vc1_## NAME ## _mmx(uint8_t *dst, const uint8_t *src, \
370 x86_reg stride, int rnd, x86_reg offset) \
376 LOAD_ROUNDER_MMX("%6") \
377 "movq "MANGLE(ff_pw_53)", %%mm5 \n\t" \
378 "movq "MANGLE(ff_pw_18)", %%mm6 \n\t" \
381 MSPEL_FILTER13_CORE(DO_UNPACK, "movd 1", A1, A2, A3, A4) \
382 NORMALIZE_MMX("$6") \
383 TRANSFER_DO_PACK(OP) \
388 : "+r"(h), "+r" (src), "+r" (dst) \
389 : "r"(offset), "r"(3*offset), "g"(stride), "m"(rnd) \
395 MSPEL_FILTER13_8B (
shift1,
"0(%1,%4 )",
"0(%1,%3,2)",
"0(%1,%3 )",
"0(%1 )",
OP_PUT, put_)
396 MSPEL_FILTER13_8B (
shift1, "0(%1,%4 )", "0(%1,%3,2)", "0(%1,%3 )", "0(%1 )", OP_AVG, avg_)
397 MSPEL_FILTER13_VER_16B(shift1, "0(%1,%4 )", "0(%1,%3,2)", "0(%1,%3 )", "0(%1 )")
398 MSPEL_FILTER13_HOR_16B(shift1, "2*3(%1)", "2*2(%1)", "2*1(%1)", "2*0(%1)",
OP_PUT, put_)
399 MSPEL_FILTER13_HOR_16B(shift1, "2*3(%1)", "2*2(%1)", "2*1(%1)", "2*0(%1)", OP_AVG, avg_)
402 MSPEL_FILTER13_8B (shift3, "0(%1 )", "0(%1,%3 )", "0(%1,%3,2)", "0(%1,%4 )", OP_PUT, put_)
403 MSPEL_FILTER13_8B (shift3, "0(%1 )", "0(%1,%3 )", "0(%1,%3,2)", "0(%1,%4 )", OP_AVG, avg_)
404 MSPEL_FILTER13_VER_16B(shift3, "0(%1 )", "0(%1,%3 )", "0(%1,%3,2)", "0(%1,%4 )")
405 MSPEL_FILTER13_HOR_16B(shift3, "2*0(%1)", "2*1(%1)", "2*2(%1)", "2*3(%1)", OP_PUT, put_)
406 MSPEL_FILTER13_HOR_16B(shift3, "2*0(%1)", "2*1(%1)", "2*2(%1)", "2*3(%1)", OP_AVG, avg_)
408 typedef
void (*vc1_mspel_mc_filter_ver_16bits)(int16_t *dst, const
uint8_t *src,
x86_reg src_stride,
int rnd, int64_t shift);
409 typedef
void (*vc1_mspel_mc_filter_hor_16bits)(
uint8_t *dst,
x86_reg dst_stride, const int16_t *src,
int rnd);
423 #define VC1_MSPEL_MC(OP)\
424 static void OP ## vc1_mspel_mc(uint8_t *dst, const uint8_t *src, int stride,\
425 int hmode, int vmode, int rnd)\
427 static const vc1_mspel_mc_filter_ver_16bits vc1_put_shift_ver_16bits[] =\
428 { NULL, vc1_put_ver_16b_shift1_mmx, vc1_put_ver_16b_shift2_mmx, vc1_put_ver_16b_shift3_mmx };\
429 static const vc1_mspel_mc_filter_hor_16bits vc1_put_shift_hor_16bits[] =\
430 { NULL, OP ## vc1_hor_16b_shift1_mmx, OP ## vc1_hor_16b_shift2_mmx, OP ## vc1_hor_16b_shift3_mmx };\
431 static const vc1_mspel_mc_filter_8bits vc1_put_shift_8bits[] =\
432 { NULL, OP ## vc1_shift1_mmx, OP ## vc1_shift2_mmx, OP ## vc1_shift3_mmx };\
435 "pxor %%mm0, %%mm0 \n\t"\
441 static const int shift_value[] = { 0, 5, 1, 5 };\
442 int shift = (shift_value[hmode]+shift_value[vmode])>>1;\
444 DECLARE_ALIGNED(16, int16_t, tmp)[12*8];\
446 r = (1<<(shift-1)) + rnd-1;\
447 vc1_put_shift_ver_16bits[vmode](tmp, src-1, stride, r, shift);\
449 vc1_put_shift_hor_16bits[hmode](dst, stride, tmp+1, 64-rnd);\
453 vc1_put_shift_8bits[vmode](dst, src, stride, 1-rnd, stride);\
459 vc1_put_shift_8bits[hmode](dst, src, stride, rnd, 1);\
466 #define DECLARE_FUNCTION(a, b) \
467 static void put_vc1_mspel_mc ## a ## b ## _mmx(uint8_t *dst, \
468 const uint8_t *src, \
472 put_vc1_mspel_mc(dst, src, stride, a, b, rnd); \
474 static void avg_vc1_mspel_mc ## a ## b ## _mmxext(uint8_t *dst, \
475 const uint8_t *src, \
479 avg_vc1_mspel_mc(dst, src, stride, a, b, rnd); \
482 DECLARE_FUNCTION(0, 1)
483 DECLARE_FUNCTION(0, 2)
484 DECLARE_FUNCTION(0, 3)
486 DECLARE_FUNCTION(1, 0)
487 DECLARE_FUNCTION(1, 1)
488 DECLARE_FUNCTION(1, 2)
489 DECLARE_FUNCTION(1, 3)
491 DECLARE_FUNCTION(2, 0)
492 DECLARE_FUNCTION(2, 1)
493 DECLARE_FUNCTION(2, 2)
494 DECLARE_FUNCTION(2, 3)
496 DECLARE_FUNCTION(3, 0)
497 DECLARE_FUNCTION(3, 1)
498 DECLARE_FUNCTION(3, 2)
499 DECLARE_FUNCTION(3, 3)
501 static
void vc1_inv_trans_4x4_dc_mmxext(
uint8_t *dest,
int linesize,
505 dc = (17 * dc + 4) >> 3;
506 dc = (17 * dc + 64) >> 7;
508 "movd %0, %%mm0 \n\t"
509 "pshufw $0, %%mm0, %%mm0 \n\t"
510 "pxor %%mm1, %%mm1 \n\t"
511 "psubw %%mm0, %%mm1 \n\t"
512 "packuswb %%mm0, %%mm0 \n\t"
513 "packuswb %%mm1, %%mm1 \n\t"
517 "movd %0, %%mm2 \n\t"
518 "movd %1, %%mm3 \n\t"
519 "movd %2, %%mm4 \n\t"
520 "movd %3, %%mm5 \n\t"
521 "paddusb %%mm0, %%mm2 \n\t"
522 "paddusb %%mm0, %%mm3 \n\t"
523 "paddusb %%mm0, %%mm4 \n\t"
524 "paddusb %%mm0, %%mm5 \n\t"
525 "psubusb %%mm1, %%mm2 \n\t"
526 "psubusb %%mm1, %%mm3 \n\t"
527 "psubusb %%mm1, %%mm4 \n\t"
528 "psubusb %%mm1, %%mm5 \n\t"
529 "movd %%mm2, %0 \n\t"
530 "movd %%mm3, %1 \n\t"
531 "movd %%mm4, %2 \n\t"
532 "movd %%mm5, %3 \n\t"
533 :
"+m"(*(uint32_t*)(dest+0*linesize)),
534 "+m"(*(uint32_t*)(dest+1*linesize)),
535 "+m"(*(uint32_t*)(dest+2*linesize)),
536 "+m"(*(uint32_t*)(dest+3*linesize))
540 static void vc1_inv_trans_4x8_dc_mmxext(
uint8_t *dest,
int linesize,
544 dc = (17 * dc + 4) >> 3;
545 dc = (12 * dc + 64) >> 7;
547 "movd %0, %%mm0 \n\t"
548 "pshufw $0, %%mm0, %%mm0 \n\t"
549 "pxor %%mm1, %%mm1 \n\t"
550 "psubw %%mm0, %%mm1 \n\t"
551 "packuswb %%mm0, %%mm0 \n\t"
552 "packuswb %%mm1, %%mm1 \n\t"
556 "movd %0, %%mm2 \n\t"
557 "movd %1, %%mm3 \n\t"
558 "movd %2, %%mm4 \n\t"
559 "movd %3, %%mm5 \n\t"
560 "paddusb %%mm0, %%mm2 \n\t"
561 "paddusb %%mm0, %%mm3 \n\t"
562 "paddusb %%mm0, %%mm4 \n\t"
563 "paddusb %%mm0, %%mm5 \n\t"
564 "psubusb %%mm1, %%mm2 \n\t"
565 "psubusb %%mm1, %%mm3 \n\t"
566 "psubusb %%mm1, %%mm4 \n\t"
567 "psubusb %%mm1, %%mm5 \n\t"
568 "movd %%mm2, %0 \n\t"
569 "movd %%mm3, %1 \n\t"
570 "movd %%mm4, %2 \n\t"
571 "movd %%mm5, %3 \n\t"
572 :
"+m"(*(uint32_t*)(dest+0*linesize)),
573 "+m"(*(uint32_t*)(dest+1*linesize)),
574 "+m"(*(uint32_t*)(dest+2*linesize)),
575 "+m"(*(uint32_t*)(dest+3*linesize))
579 "movd %0, %%mm2 \n\t"
580 "movd %1, %%mm3 \n\t"
581 "movd %2, %%mm4 \n\t"
582 "movd %3, %%mm5 \n\t"
583 "paddusb %%mm0, %%mm2 \n\t"
584 "paddusb %%mm0, %%mm3 \n\t"
585 "paddusb %%mm0, %%mm4 \n\t"
586 "paddusb %%mm0, %%mm5 \n\t"
587 "psubusb %%mm1, %%mm2 \n\t"
588 "psubusb %%mm1, %%mm3 \n\t"
589 "psubusb %%mm1, %%mm4 \n\t"
590 "psubusb %%mm1, %%mm5 \n\t"
591 "movd %%mm2, %0 \n\t"
592 "movd %%mm3, %1 \n\t"
593 "movd %%mm4, %2 \n\t"
594 "movd %%mm5, %3 \n\t"
595 :
"+m"(*(uint32_t*)(dest+0*linesize)),
596 "+m"(*(uint32_t*)(dest+1*linesize)),
597 "+m"(*(uint32_t*)(dest+2*linesize)),
598 "+m"(*(uint32_t*)(dest+3*linesize))
602 static void vc1_inv_trans_8x4_dc_mmxext(
uint8_t *dest,
int linesize,
606 dc = ( 3 * dc + 1) >> 1;
607 dc = (17 * dc + 64) >> 7;
609 "movd %0, %%mm0 \n\t"
610 "pshufw $0, %%mm0, %%mm0 \n\t"
611 "pxor %%mm1, %%mm1 \n\t"
612 "psubw %%mm0, %%mm1 \n\t"
613 "packuswb %%mm0, %%mm0 \n\t"
614 "packuswb %%mm1, %%mm1 \n\t"
618 "movq %0, %%mm2 \n\t"
619 "movq %1, %%mm3 \n\t"
620 "movq %2, %%mm4 \n\t"
621 "movq %3, %%mm5 \n\t"
622 "paddusb %%mm0, %%mm2 \n\t"
623 "paddusb %%mm0, %%mm3 \n\t"
624 "paddusb %%mm0, %%mm4 \n\t"
625 "paddusb %%mm0, %%mm5 \n\t"
626 "psubusb %%mm1, %%mm2 \n\t"
627 "psubusb %%mm1, %%mm3 \n\t"
628 "psubusb %%mm1, %%mm4 \n\t"
629 "psubusb %%mm1, %%mm5 \n\t"
630 "movq %%mm2, %0 \n\t"
631 "movq %%mm3, %1 \n\t"
632 "movq %%mm4, %2 \n\t"
633 "movq %%mm5, %3 \n\t"
634 :
"+m"(*(uint32_t*)(dest+0*linesize)),
635 "+m"(*(uint32_t*)(dest+1*linesize)),
636 "+m"(*(uint32_t*)(dest+2*linesize)),
637 "+m"(*(uint32_t*)(dest+3*linesize))
641 static void vc1_inv_trans_8x8_dc_mmxext(
uint8_t *dest,
int linesize,
645 dc = (3 * dc + 1) >> 1;
646 dc = (3 * dc + 16) >> 5;
648 "movd %0, %%mm0 \n\t"
649 "pshufw $0, %%mm0, %%mm0 \n\t"
650 "pxor %%mm1, %%mm1 \n\t"
651 "psubw %%mm0, %%mm1 \n\t"
652 "packuswb %%mm0, %%mm0 \n\t"
653 "packuswb %%mm1, %%mm1 \n\t"
657 "movq %0, %%mm2 \n\t"
658 "movq %1, %%mm3 \n\t"
659 "movq %2, %%mm4 \n\t"
660 "movq %3, %%mm5 \n\t"
661 "paddusb %%mm0, %%mm2 \n\t"
662 "paddusb %%mm0, %%mm3 \n\t"
663 "paddusb %%mm0, %%mm4 \n\t"
664 "paddusb %%mm0, %%mm5 \n\t"
665 "psubusb %%mm1, %%mm2 \n\t"
666 "psubusb %%mm1, %%mm3 \n\t"
667 "psubusb %%mm1, %%mm4 \n\t"
668 "psubusb %%mm1, %%mm5 \n\t"
669 "movq %%mm2, %0 \n\t"
670 "movq %%mm3, %1 \n\t"
671 "movq %%mm4, %2 \n\t"
672 "movq %%mm5, %3 \n\t"
673 :
"+m"(*(uint32_t*)(dest+0*linesize)),
674 "+m"(*(uint32_t*)(dest+1*linesize)),
675 "+m"(*(uint32_t*)(dest+2*linesize)),
676 "+m"(*(uint32_t*)(dest+3*linesize))
680 "movq %0, %%mm2 \n\t"
681 "movq %1, %%mm3 \n\t"
682 "movq %2, %%mm4 \n\t"
683 "movq %3, %%mm5 \n\t"
684 "paddusb %%mm0, %%mm2 \n\t"
685 "paddusb %%mm0, %%mm3 \n\t"
686 "paddusb %%mm0, %%mm4 \n\t"
687 "paddusb %%mm0, %%mm5 \n\t"
688 "psubusb %%mm1, %%mm2 \n\t"
689 "psubusb %%mm1, %%mm3 \n\t"
690 "psubusb %%mm1, %%mm4 \n\t"
691 "psubusb %%mm1, %%mm5 \n\t"
692 "movq %%mm2, %0 \n\t"
693 "movq %%mm3, %1 \n\t"
694 "movq %%mm4, %2 \n\t"
695 "movq %%mm5, %3 \n\t"
696 :
"+m"(*(uint32_t*)(dest+0*linesize)),
697 "+m"(*(uint32_t*)(dest+1*linesize)),
698 "+m"(*(uint32_t*)(dest+2*linesize)),
699 "+m"(*(uint32_t*)(dest+3*linesize))
703 static void put_vc1_mspel_mc00_mmx(
uint8_t *dst,
const uint8_t *src,
704 ptrdiff_t stride,
int rnd)
op_pixels_func avg_vc1_mspel_pixels_tab[16]
void ff_put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
memory handling functions
void ff_vc1dsp_init_mmx(VC1DSPContext *dsp)
void(* vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, int16_t *block)
void(* vc1_inv_trans_8x8_dc)(uint8_t *dest, int line_size, int16_t *block)
static const int shift1[6]
void(* vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, int16_t *block)
common internal API header
void ff_vc1dsp_init_mmxext(VC1DSPContext *dsp)
#define VC1_MSPEL_MC(OP, OPNAME)
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(constuint8_t *) pi-0x80)*(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(constuint8_t *) pi-0x80)*(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(constint16_t *) pi >>8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t,*(constint16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t,*(constint16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(constint32_t *) pi >>24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t,*(constint32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t,*(constint32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(constfloat *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(constfloat *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(constfloat *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(constdouble *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(constdouble *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(constdouble *) pi *(1U<< 31))))#defineSET_CONV_FUNC_GROUP(ofmt, ifmt) staticvoidset_generic_function(AudioConvert *ac){}voidff_audio_convert_free(AudioConvert **ac){if(!*ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);}AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enumAVSampleFormatout_fmt, enumAVSampleFormatin_fmt, intchannels, intsample_rate, intapply_map){AudioConvert *ac;intin_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) returnNULL;ac->avr=avr;ac->out_fmt=out_fmt;ac->in_fmt=in_fmt;ac->channels=channels;ac->apply_map=apply_map;if(avr->dither_method!=AV_RESAMPLE_DITHER_NONE &&av_get_packed_sample_fmt(out_fmt)==AV_SAMPLE_FMT_S16 &&av_get_bytes_per_sample(in_fmt)>2){ac->dc=ff_dither_alloc(avr, out_fmt, in_fmt, channels, sample_rate, apply_map);if(!ac->dc){av_free(ac);returnNULL;}returnac;}in_planar=ff_sample_fmt_is_planar(in_fmt, channels);out_planar=ff_sample_fmt_is_planar(out_fmt, channels);if(in_planar==out_planar){ac->func_type=CONV_FUNC_TYPE_FLAT;ac->planes=in_planar?ac->channels:1;}elseif(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;elseac->func_type=CONV_FUNC_TYPE_DEINTERLEAVE;set_generic_function(ac);if(ARCH_AARCH64) ff_audio_convert_init_aarch64(ac);if(ARCH_ARM) ff_audio_convert_init_arm(ac);if(ARCH_X86) ff_audio_convert_init_x86(ac);returnac;}intff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in){intuse_generic=1;intlen=in->nb_samples;intp;if(ac->dc){av_dlog(ac->avr,"%dsamples-audio_convert:%sto%s(dithered)\n", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));returnff_convert_dither(ac-> dc
void(* vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, int16_t *block)
op_pixels_func put_vc1_mspel_pixels_tab[16]