2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of Libav.
6 * Libav is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * Libav is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 #if COMPILE_TEMPLATE_MMX2
26 #define PREFETCH "prefetchnta"
28 #define PREFETCH " # nop"
31 #if COMPILE_TEMPLATE_MMX2
32 #define REAL_MOVNTQ(a,b) "movntq " #a ", " #b " \n\t"
34 #define REAL_MOVNTQ(a,b) "movq " #a ", " #b " \n\t"
36 #define MOVNTQ(a,b) REAL_MOVNTQ(a,b)
38 #define YSCALEYUV2YV12X(offset, dest, end, pos) \
40 "movq "VROUNDER_OFFSET"(%0), %%mm3 \n\t"\
41 "movq %%mm3, %%mm4 \n\t"\
42 "lea " offset "(%0), %%"REG_d" \n\t"\
43 "mov (%%"REG_d"), %%"REG_S" \n\t"\
44 ".p2align 4 \n\t" /* FIXME Unroll? */\
46 "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
47 "movq (%%"REG_S", %3, 2), %%mm2 \n\t" /* srcData */\
48 "movq 8(%%"REG_S", %3, 2), %%mm5 \n\t" /* srcData */\
49 "add $16, %%"REG_d" \n\t"\
50 "mov (%%"REG_d"), %%"REG_S" \n\t"\
51 "test %%"REG_S", %%"REG_S" \n\t"\
52 "pmulhw %%mm0, %%mm2 \n\t"\
53 "pmulhw %%mm0, %%mm5 \n\t"\
54 "paddw %%mm2, %%mm3 \n\t"\
55 "paddw %%mm5, %%mm4 \n\t"\
57 "psraw $3, %%mm3 \n\t"\
58 "psraw $3, %%mm4 \n\t"\
59 "packuswb %%mm4, %%mm3 \n\t"\
60 MOVNTQ(%%mm3, (%1, %3))\
63 "movq "VROUNDER_OFFSET"(%0), %%mm3 \n\t"\
64 "movq %%mm3, %%mm4 \n\t"\
65 "lea " offset "(%0), %%"REG_d" \n\t"\
66 "mov (%%"REG_d"), %%"REG_S" \n\t"\
68 :: "r" (&c->redDither),\
69 "r" (dest), "g" ((x86_reg)(end)), "r"((x86_reg)(pos))\
73 static void RENAME(yuv2yuvX)(SwsContext *c, const int16_t *lumFilter,
74 const int16_t **lumSrc, int lumFilterSize,
75 const int16_t *chrFilter, const int16_t **chrUSrc,
76 const int16_t **chrVSrc,
77 int chrFilterSize, const int16_t **alpSrc,
78 uint8_t *dest[4], int dstW, int chrDstW)
80 uint8_t *yDest = dest[0], *uDest = dest[1], *vDest = dest[2],
81 *aDest = CONFIG_SWSCALE_ALPHA ? dest[3] : NULL;
84 x86_reg uv_off = c->uv_off;
85 YSCALEYUV2YV12X(CHR_MMX_FILTER_OFFSET, uDest, chrDstW, 0)
86 YSCALEYUV2YV12X(CHR_MMX_FILTER_OFFSET, vDest - uv_off, chrDstW + uv_off, uv_off)
88 if (CONFIG_SWSCALE_ALPHA && aDest) {
89 YSCALEYUV2YV12X(ALP_MMX_FILTER_OFFSET, aDest, dstW, 0)
92 YSCALEYUV2YV12X(LUM_MMX_FILTER_OFFSET, yDest, dstW, 0)
95 #define YSCALEYUV2YV12X_ACCURATE(offset, dest, end, pos) \
97 "lea " offset "(%0), %%"REG_d" \n\t"\
98 "pxor %%mm4, %%mm4 \n\t"\
99 "pxor %%mm5, %%mm5 \n\t"\
100 "pxor %%mm6, %%mm6 \n\t"\
101 "pxor %%mm7, %%mm7 \n\t"\
102 "mov (%%"REG_d"), %%"REG_S" \n\t"\
105 "movq (%%"REG_S", %3, 2), %%mm0 \n\t" /* srcData */\
106 "movq 8(%%"REG_S", %3, 2), %%mm2 \n\t" /* srcData */\
107 "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S" \n\t"\
108 "movq (%%"REG_S", %3, 2), %%mm1 \n\t" /* srcData */\
109 "movq %%mm0, %%mm3 \n\t"\
110 "punpcklwd %%mm1, %%mm0 \n\t"\
111 "punpckhwd %%mm1, %%mm3 \n\t"\
112 "movq "STR(APCK_COEF)"(%%"REG_d"), %%mm1 \n\t" /* filterCoeff */\
113 "pmaddwd %%mm1, %%mm0 \n\t"\
114 "pmaddwd %%mm1, %%mm3 \n\t"\
115 "paddd %%mm0, %%mm4 \n\t"\
116 "paddd %%mm3, %%mm5 \n\t"\
117 "movq 8(%%"REG_S", %3, 2), %%mm3 \n\t" /* srcData */\
118 "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S" \n\t"\
119 "add $"STR(APCK_SIZE)", %%"REG_d" \n\t"\
120 "test %%"REG_S", %%"REG_S" \n\t"\
121 "movq %%mm2, %%mm0 \n\t"\
122 "punpcklwd %%mm3, %%mm2 \n\t"\
123 "punpckhwd %%mm3, %%mm0 \n\t"\
124 "pmaddwd %%mm1, %%mm2 \n\t"\
125 "pmaddwd %%mm1, %%mm0 \n\t"\
126 "paddd %%mm2, %%mm6 \n\t"\
127 "paddd %%mm0, %%mm7 \n\t"\
129 "psrad $16, %%mm4 \n\t"\
130 "psrad $16, %%mm5 \n\t"\
131 "psrad $16, %%mm6 \n\t"\
132 "psrad $16, %%mm7 \n\t"\
133 "movq "VROUNDER_OFFSET"(%0), %%mm0 \n\t"\
134 "packssdw %%mm5, %%mm4 \n\t"\
135 "packssdw %%mm7, %%mm6 \n\t"\
136 "paddw %%mm0, %%mm4 \n\t"\
137 "paddw %%mm0, %%mm6 \n\t"\
138 "psraw $3, %%mm4 \n\t"\
139 "psraw $3, %%mm6 \n\t"\
140 "packuswb %%mm6, %%mm4 \n\t"\
141 MOVNTQ(%%mm4, (%1, %3))\
144 "lea " offset "(%0), %%"REG_d" \n\t"\
145 "pxor %%mm4, %%mm4 \n\t"\
146 "pxor %%mm5, %%mm5 \n\t"\
147 "pxor %%mm6, %%mm6 \n\t"\
148 "pxor %%mm7, %%mm7 \n\t"\
149 "mov (%%"REG_d"), %%"REG_S" \n\t"\
151 :: "r" (&c->redDither),\
152 "r" (dest), "g" ((x86_reg)(end)), "r"((x86_reg)(pos))\
153 : "%"REG_a, "%"REG_d, "%"REG_S\
156 static void RENAME(yuv2yuvX_ar)(SwsContext *c, const int16_t *lumFilter,
157 const int16_t **lumSrc, int lumFilterSize,
158 const int16_t *chrFilter, const int16_t **chrUSrc,
159 const int16_t **chrVSrc,
160 int chrFilterSize, const int16_t **alpSrc,
161 uint8_t *dest[4], int dstW, int chrDstW)
163 uint8_t *yDest = dest[0], *uDest = dest[1], *vDest = dest[2],
164 *aDest = CONFIG_SWSCALE_ALPHA ? dest[3] : NULL;
167 x86_reg uv_off = c->uv_off;
168 YSCALEYUV2YV12X_ACCURATE(CHR_MMX_FILTER_OFFSET, uDest, chrDstW, 0)
169 YSCALEYUV2YV12X_ACCURATE(CHR_MMX_FILTER_OFFSET, vDest - uv_off, chrDstW + uv_off, uv_off)
171 if (CONFIG_SWSCALE_ALPHA && aDest) {
172 YSCALEYUV2YV12X_ACCURATE(ALP_MMX_FILTER_OFFSET, aDest, dstW, 0)
175 YSCALEYUV2YV12X_ACCURATE(LUM_MMX_FILTER_OFFSET, yDest, dstW, 0)
178 static void RENAME(yuv2yuv1)(SwsContext *c, const int16_t *lumSrc,
179 const int16_t *chrUSrc, const int16_t *chrVSrc,
180 const int16_t *alpSrc,
181 uint8_t *dst[4], int dstW, int chrDstW)
184 const int16_t *src[4]= {
185 lumSrc + dstW, chrUSrc + chrDstW,
186 chrVSrc + chrDstW, alpSrc + dstW
188 x86_reg counter[4]= { dstW, chrDstW, chrDstW, dstW };
193 "mov %2, %%"REG_a" \n\t"
194 ".p2align 4 \n\t" /* FIXME Unroll? */
196 "movq (%0, %%"REG_a", 2), %%mm0 \n\t"
197 "movq 8(%0, %%"REG_a", 2), %%mm1 \n\t"
198 "psraw $7, %%mm0 \n\t"
199 "psraw $7, %%mm1 \n\t"
200 "packuswb %%mm1, %%mm0 \n\t"
201 MOVNTQ(%%mm0, (%1, %%REGa))
202 "add $8, %%"REG_a" \n\t"
204 :: "r" (src[p]), "r" (dst[p] + counter[p]),
212 static void RENAME(yuv2yuv1_ar)(SwsContext *c, const int16_t *lumSrc,
213 const int16_t *chrUSrc, const int16_t *chrVSrc,
214 const int16_t *alpSrc,
215 uint8_t *dst[4], int dstW, int chrDstW)
218 const int16_t *src[4]= {
219 lumSrc + dstW, chrUSrc + chrDstW,
220 chrVSrc + chrDstW, alpSrc + dstW
222 x86_reg counter[4]= { dstW, chrDstW, chrDstW, dstW };
227 "mov %2, %%"REG_a" \n\t"
228 "pcmpeqw %%mm7, %%mm7 \n\t"
229 "psrlw $15, %%mm7 \n\t"
230 "psllw $6, %%mm7 \n\t"
231 ".p2align 4 \n\t" /* FIXME Unroll? */
233 "movq (%0, %%"REG_a", 2), %%mm0 \n\t"
234 "movq 8(%0, %%"REG_a", 2), %%mm1 \n\t"
235 "paddsw %%mm7, %%mm0 \n\t"
236 "paddsw %%mm7, %%mm1 \n\t"
237 "psraw $7, %%mm0 \n\t"
238 "psraw $7, %%mm1 \n\t"
239 "packuswb %%mm1, %%mm0 \n\t"
240 MOVNTQ(%%mm0, (%1, %%REGa))
241 "add $8, %%"REG_a" \n\t"
243 :: "r" (src[p]), "r" (dst[p] + counter[p]),
251 #define YSCALEYUV2PACKEDX_UV \
253 "xor %%"REG_a", %%"REG_a" \n\t"\
257 "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d" \n\t"\
258 "mov (%%"REG_d"), %%"REG_S" \n\t"\
259 "movq "VROUNDER_OFFSET"(%0), %%mm3 \n\t"\
260 "movq %%mm3, %%mm4 \n\t"\
263 "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
264 "movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* UsrcData */\
265 "add %6, %%"REG_S" \n\t" \
266 "movq (%%"REG_S", %%"REG_a"), %%mm5 \n\t" /* VsrcData */\
267 "add $16, %%"REG_d" \n\t"\
268 "mov (%%"REG_d"), %%"REG_S" \n\t"\
269 "pmulhw %%mm0, %%mm2 \n\t"\
270 "pmulhw %%mm0, %%mm5 \n\t"\
271 "paddw %%mm2, %%mm3 \n\t"\
272 "paddw %%mm5, %%mm4 \n\t"\
273 "test %%"REG_S", %%"REG_S" \n\t"\
276 #define YSCALEYUV2PACKEDX_YA(offset,coeff,src1,src2,dst1,dst2) \
277 "lea "offset"(%0), %%"REG_d" \n\t"\
278 "mov (%%"REG_d"), %%"REG_S" \n\t"\
279 "movq "VROUNDER_OFFSET"(%0), "#dst1" \n\t"\
280 "movq "#dst1", "#dst2" \n\t"\
283 "movq 8(%%"REG_d"), "#coeff" \n\t" /* filterCoeff */\
284 "movq (%%"REG_S", %%"REG_a", 2), "#src1" \n\t" /* Y1srcData */\
285 "movq 8(%%"REG_S", %%"REG_a", 2), "#src2" \n\t" /* Y2srcData */\
286 "add $16, %%"REG_d" \n\t"\
287 "mov (%%"REG_d"), %%"REG_S" \n\t"\
288 "pmulhw "#coeff", "#src1" \n\t"\
289 "pmulhw "#coeff", "#src2" \n\t"\
290 "paddw "#src1", "#dst1" \n\t"\
291 "paddw "#src2", "#dst2" \n\t"\
292 "test %%"REG_S", %%"REG_S" \n\t"\
295 #define YSCALEYUV2PACKEDX \
296 YSCALEYUV2PACKEDX_UV \
297 YSCALEYUV2PACKEDX_YA(LUM_MMX_FILTER_OFFSET,%%mm0,%%mm2,%%mm5,%%mm1,%%mm7) \
299 #define YSCALEYUV2PACKEDX_END \
300 :: "r" (&c->redDither), \
301 "m" (dummy), "m" (dummy), "m" (dummy),\
302 "r" (dest), "m" (dstW_reg), "m"(uv_off) \
303 : "%"REG_a, "%"REG_d, "%"REG_S \
306 #define YSCALEYUV2PACKEDX_ACCURATE_UV \
308 "xor %%"REG_a", %%"REG_a" \n\t"\
312 "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d" \n\t"\
313 "mov (%%"REG_d"), %%"REG_S" \n\t"\
314 "pxor %%mm4, %%mm4 \n\t"\
315 "pxor %%mm5, %%mm5 \n\t"\
316 "pxor %%mm6, %%mm6 \n\t"\
317 "pxor %%mm7, %%mm7 \n\t"\
320 "movq (%%"REG_S", %%"REG_a"), %%mm0 \n\t" /* UsrcData */\
321 "add %6, %%"REG_S" \n\t" \
322 "movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* VsrcData */\
323 "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S" \n\t"\
324 "movq (%%"REG_S", %%"REG_a"), %%mm1 \n\t" /* UsrcData */\
325 "movq %%mm0, %%mm3 \n\t"\
326 "punpcklwd %%mm1, %%mm0 \n\t"\
327 "punpckhwd %%mm1, %%mm3 \n\t"\
328 "movq "STR(APCK_COEF)"(%%"REG_d"),%%mm1 \n\t" /* filterCoeff */\
329 "pmaddwd %%mm1, %%mm0 \n\t"\
330 "pmaddwd %%mm1, %%mm3 \n\t"\
331 "paddd %%mm0, %%mm4 \n\t"\
332 "paddd %%mm3, %%mm5 \n\t"\
333 "add %6, %%"REG_S" \n\t" \
334 "movq (%%"REG_S", %%"REG_a"), %%mm3 \n\t" /* VsrcData */\
335 "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S" \n\t"\
336 "add $"STR(APCK_SIZE)", %%"REG_d" \n\t"\
337 "test %%"REG_S", %%"REG_S" \n\t"\
338 "movq %%mm2, %%mm0 \n\t"\
339 "punpcklwd %%mm3, %%mm2 \n\t"\
340 "punpckhwd %%mm3, %%mm0 \n\t"\
341 "pmaddwd %%mm1, %%mm2 \n\t"\
342 "pmaddwd %%mm1, %%mm0 \n\t"\
343 "paddd %%mm2, %%mm6 \n\t"\
344 "paddd %%mm0, %%mm7 \n\t"\
346 "psrad $16, %%mm4 \n\t"\
347 "psrad $16, %%mm5 \n\t"\
348 "psrad $16, %%mm6 \n\t"\
349 "psrad $16, %%mm7 \n\t"\
350 "movq "VROUNDER_OFFSET"(%0), %%mm0 \n\t"\
351 "packssdw %%mm5, %%mm4 \n\t"\
352 "packssdw %%mm7, %%mm6 \n\t"\
353 "paddw %%mm0, %%mm4 \n\t"\
354 "paddw %%mm0, %%mm6 \n\t"\
355 "movq %%mm4, "U_TEMP"(%0) \n\t"\
356 "movq %%mm6, "V_TEMP"(%0) \n\t"\
358 #define YSCALEYUV2PACKEDX_ACCURATE_YA(offset) \
359 "lea "offset"(%0), %%"REG_d" \n\t"\
360 "mov (%%"REG_d"), %%"REG_S" \n\t"\
361 "pxor %%mm1, %%mm1 \n\t"\
362 "pxor %%mm5, %%mm5 \n\t"\
363 "pxor %%mm7, %%mm7 \n\t"\
364 "pxor %%mm6, %%mm6 \n\t"\
367 "movq (%%"REG_S", %%"REG_a", 2), %%mm0 \n\t" /* Y1srcData */\
368 "movq 8(%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* Y2srcData */\
369 "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S" \n\t"\
370 "movq (%%"REG_S", %%"REG_a", 2), %%mm4 \n\t" /* Y1srcData */\
371 "movq %%mm0, %%mm3 \n\t"\
372 "punpcklwd %%mm4, %%mm0 \n\t"\
373 "punpckhwd %%mm4, %%mm3 \n\t"\
374 "movq "STR(APCK_COEF)"(%%"REG_d"), %%mm4 \n\t" /* filterCoeff */\
375 "pmaddwd %%mm4, %%mm0 \n\t"\
376 "pmaddwd %%mm4, %%mm3 \n\t"\
377 "paddd %%mm0, %%mm1 \n\t"\
378 "paddd %%mm3, %%mm5 \n\t"\
379 "movq 8(%%"REG_S", %%"REG_a", 2), %%mm3 \n\t" /* Y2srcData */\
380 "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S" \n\t"\
381 "add $"STR(APCK_SIZE)", %%"REG_d" \n\t"\
382 "test %%"REG_S", %%"REG_S" \n\t"\
383 "movq %%mm2, %%mm0 \n\t"\
384 "punpcklwd %%mm3, %%mm2 \n\t"\
385 "punpckhwd %%mm3, %%mm0 \n\t"\
386 "pmaddwd %%mm4, %%mm2 \n\t"\
387 "pmaddwd %%mm4, %%mm0 \n\t"\
388 "paddd %%mm2, %%mm7 \n\t"\
389 "paddd %%mm0, %%mm6 \n\t"\
391 "psrad $16, %%mm1 \n\t"\
392 "psrad $16, %%mm5 \n\t"\
393 "psrad $16, %%mm7 \n\t"\
394 "psrad $16, %%mm6 \n\t"\
395 "movq "VROUNDER_OFFSET"(%0), %%mm0 \n\t"\
396 "packssdw %%mm5, %%mm1 \n\t"\
397 "packssdw %%mm6, %%mm7 \n\t"\
398 "paddw %%mm0, %%mm1 \n\t"\
399 "paddw %%mm0, %%mm7 \n\t"\
400 "movq "U_TEMP"(%0), %%mm3 \n\t"\
401 "movq "V_TEMP"(%0), %%mm4 \n\t"\
403 #define YSCALEYUV2PACKEDX_ACCURATE \
404 YSCALEYUV2PACKEDX_ACCURATE_UV \
405 YSCALEYUV2PACKEDX_ACCURATE_YA(LUM_MMX_FILTER_OFFSET)
407 #define YSCALEYUV2RGBX \
408 "psubw "U_OFFSET"(%0), %%mm3 \n\t" /* (U-128)8*/\
409 "psubw "V_OFFSET"(%0), %%mm4 \n\t" /* (V-128)8*/\
410 "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\
411 "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\
412 "pmulhw "UG_COEFF"(%0), %%mm3 \n\t"\
413 "pmulhw "VG_COEFF"(%0), %%mm4 \n\t"\
414 /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
415 "pmulhw "UB_COEFF"(%0), %%mm2 \n\t"\
416 "pmulhw "VR_COEFF"(%0), %%mm5 \n\t"\
417 "psubw "Y_OFFSET"(%0), %%mm1 \n\t" /* 8(Y-16)*/\
418 "psubw "Y_OFFSET"(%0), %%mm7 \n\t" /* 8(Y-16)*/\
419 "pmulhw "Y_COEFF"(%0), %%mm1 \n\t"\
420 "pmulhw "Y_COEFF"(%0), %%mm7 \n\t"\
421 /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
422 "paddw %%mm3, %%mm4 \n\t"\
423 "movq %%mm2, %%mm0 \n\t"\
424 "movq %%mm5, %%mm6 \n\t"\
425 "movq %%mm4, %%mm3 \n\t"\
426 "punpcklwd %%mm2, %%mm2 \n\t"\
427 "punpcklwd %%mm5, %%mm5 \n\t"\
428 "punpcklwd %%mm4, %%mm4 \n\t"\
429 "paddw %%mm1, %%mm2 \n\t"\
430 "paddw %%mm1, %%mm5 \n\t"\
431 "paddw %%mm1, %%mm4 \n\t"\
432 "punpckhwd %%mm0, %%mm0 \n\t"\
433 "punpckhwd %%mm6, %%mm6 \n\t"\
434 "punpckhwd %%mm3, %%mm3 \n\t"\
435 "paddw %%mm7, %%mm0 \n\t"\
436 "paddw %%mm7, %%mm6 \n\t"\
437 "paddw %%mm7, %%mm3 \n\t"\
438 /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
439 "packuswb %%mm0, %%mm2 \n\t"\
440 "packuswb %%mm6, %%mm5 \n\t"\
441 "packuswb %%mm3, %%mm4 \n\t"\
443 #define REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) \
444 "movq "#b", "#q2" \n\t" /* B */\
445 "movq "#r", "#t" \n\t" /* R */\
446 "punpcklbw "#g", "#b" \n\t" /* GBGBGBGB 0 */\
447 "punpcklbw "#a", "#r" \n\t" /* ARARARAR 0 */\
448 "punpckhbw "#g", "#q2" \n\t" /* GBGBGBGB 2 */\
449 "punpckhbw "#a", "#t" \n\t" /* ARARARAR 2 */\
450 "movq "#b", "#q0" \n\t" /* GBGBGBGB 0 */\
451 "movq "#q2", "#q3" \n\t" /* GBGBGBGB 2 */\
452 "punpcklwd "#r", "#q0" \n\t" /* ARGBARGB 0 */\
453 "punpckhwd "#r", "#b" \n\t" /* ARGBARGB 1 */\
454 "punpcklwd "#t", "#q2" \n\t" /* ARGBARGB 2 */\
455 "punpckhwd "#t", "#q3" \n\t" /* ARGBARGB 3 */\
457 MOVNTQ( q0, (dst, index, 4))\
458 MOVNTQ( b, 8(dst, index, 4))\
459 MOVNTQ( q2, 16(dst, index, 4))\
460 MOVNTQ( q3, 24(dst, index, 4))\
462 "add $8, "#index" \n\t"\
463 "cmp "#dstw", "#index" \n\t"\
465 #define WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t)
467 static void RENAME(yuv2rgb32_X_ar)(SwsContext *c, const int16_t *lumFilter,
468 const int16_t **lumSrc, int lumFilterSize,
469 const int16_t *chrFilter, const int16_t **chrUSrc,
470 const int16_t **chrVSrc,
471 int chrFilterSize, const int16_t **alpSrc,
472 uint8_t *dest, int dstW, int dstY)
475 x86_reg dstW_reg = dstW;
476 x86_reg uv_off = c->uv_off << 1;
478 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
479 YSCALEYUV2PACKEDX_ACCURATE
481 "movq %%mm2, "U_TEMP"(%0) \n\t"
482 "movq %%mm4, "V_TEMP"(%0) \n\t"
483 "movq %%mm5, "Y_TEMP"(%0) \n\t"
484 YSCALEYUV2PACKEDX_ACCURATE_YA(ALP_MMX_FILTER_OFFSET)
485 "movq "Y_TEMP"(%0), %%mm5 \n\t"
486 "psraw $3, %%mm1 \n\t"
487 "psraw $3, %%mm7 \n\t"
488 "packuswb %%mm7, %%mm1 \n\t"
489 WRITEBGR32(%4, %5, %%REGa, %%mm3, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm2, %%mm6)
490 YSCALEYUV2PACKEDX_END
492 YSCALEYUV2PACKEDX_ACCURATE
494 "pcmpeqd %%mm7, %%mm7 \n\t"
495 WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
496 YSCALEYUV2PACKEDX_END
500 static void RENAME(yuv2rgb32_X)(SwsContext *c, const int16_t *lumFilter,
501 const int16_t **lumSrc, int lumFilterSize,
502 const int16_t *chrFilter, const int16_t **chrUSrc,
503 const int16_t **chrVSrc,
504 int chrFilterSize, const int16_t **alpSrc,
505 uint8_t *dest, int dstW, int dstY)
508 x86_reg dstW_reg = dstW;
509 x86_reg uv_off = c->uv_off << 1;
511 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
514 YSCALEYUV2PACKEDX_YA(ALP_MMX_FILTER_OFFSET, %%mm0, %%mm3, %%mm6, %%mm1, %%mm7)
515 "psraw $3, %%mm1 \n\t"
516 "psraw $3, %%mm7 \n\t"
517 "packuswb %%mm7, %%mm1 \n\t"
518 WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
519 YSCALEYUV2PACKEDX_END
523 "pcmpeqd %%mm7, %%mm7 \n\t"
524 WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
525 YSCALEYUV2PACKEDX_END
529 #define REAL_WRITERGB16(dst, dstw, index) \
530 "pand "MANGLE(bF8)", %%mm2 \n\t" /* B */\
531 "pand "MANGLE(bFC)", %%mm4 \n\t" /* G */\
532 "pand "MANGLE(bF8)", %%mm5 \n\t" /* R */\
533 "psrlq $3, %%mm2 \n\t"\
535 "movq %%mm2, %%mm1 \n\t"\
536 "movq %%mm4, %%mm3 \n\t"\
538 "punpcklbw %%mm7, %%mm3 \n\t"\
539 "punpcklbw %%mm5, %%mm2 \n\t"\
540 "punpckhbw %%mm7, %%mm4 \n\t"\
541 "punpckhbw %%mm5, %%mm1 \n\t"\
543 "psllq $3, %%mm3 \n\t"\
544 "psllq $3, %%mm4 \n\t"\
546 "por %%mm3, %%mm2 \n\t"\
547 "por %%mm4, %%mm1 \n\t"\
549 MOVNTQ(%%mm2, (dst, index, 2))\
550 MOVNTQ(%%mm1, 8(dst, index, 2))\
552 "add $8, "#index" \n\t"\
553 "cmp "#dstw", "#index" \n\t"\
555 #define WRITERGB16(dst, dstw, index) REAL_WRITERGB16(dst, dstw, index)
557 static void RENAME(yuv2rgb565_X_ar)(SwsContext *c, const int16_t *lumFilter,
558 const int16_t **lumSrc, int lumFilterSize,
559 const int16_t *chrFilter, const int16_t **chrUSrc,
560 const int16_t **chrVSrc,
561 int chrFilterSize, const int16_t **alpSrc,
562 uint8_t *dest, int dstW, int dstY)
565 x86_reg dstW_reg = dstW;
566 x86_reg uv_off = c->uv_off << 1;
568 YSCALEYUV2PACKEDX_ACCURATE
570 "pxor %%mm7, %%mm7 \n\t"
571 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
573 "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"
574 "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"
575 "paddusb "RED_DITHER"(%0), %%mm5\n\t"
577 WRITERGB16(%4, %5, %%REGa)
578 YSCALEYUV2PACKEDX_END
581 static void RENAME(yuv2rgb565_X)(SwsContext *c, const int16_t *lumFilter,
582 const int16_t **lumSrc, int lumFilterSize,
583 const int16_t *chrFilter, const int16_t **chrUSrc,
584 const int16_t **chrVSrc,
585 int chrFilterSize, const int16_t **alpSrc,
586 uint8_t *dest, int dstW, int dstY)
589 x86_reg dstW_reg = dstW;
590 x86_reg uv_off = c->uv_off << 1;
594 "pxor %%mm7, %%mm7 \n\t"
595 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
597 "paddusb "BLUE_DITHER"(%0), %%mm2 \n\t"
598 "paddusb "GREEN_DITHER"(%0), %%mm4 \n\t"
599 "paddusb "RED_DITHER"(%0), %%mm5 \n\t"
601 WRITERGB16(%4, %5, %%REGa)
602 YSCALEYUV2PACKEDX_END
605 #define REAL_WRITERGB15(dst, dstw, index) \
606 "pand "MANGLE(bF8)", %%mm2 \n\t" /* B */\
607 "pand "MANGLE(bF8)", %%mm4 \n\t" /* G */\
608 "pand "MANGLE(bF8)", %%mm5 \n\t" /* R */\
609 "psrlq $3, %%mm2 \n\t"\
610 "psrlq $1, %%mm5 \n\t"\
612 "movq %%mm2, %%mm1 \n\t"\
613 "movq %%mm4, %%mm3 \n\t"\
615 "punpcklbw %%mm7, %%mm3 \n\t"\
616 "punpcklbw %%mm5, %%mm2 \n\t"\
617 "punpckhbw %%mm7, %%mm4 \n\t"\
618 "punpckhbw %%mm5, %%mm1 \n\t"\
620 "psllq $2, %%mm3 \n\t"\
621 "psllq $2, %%mm4 \n\t"\
623 "por %%mm3, %%mm2 \n\t"\
624 "por %%mm4, %%mm1 \n\t"\
626 MOVNTQ(%%mm2, (dst, index, 2))\
627 MOVNTQ(%%mm1, 8(dst, index, 2))\
629 "add $8, "#index" \n\t"\
630 "cmp "#dstw", "#index" \n\t"\
632 #define WRITERGB15(dst, dstw, index) REAL_WRITERGB15(dst, dstw, index)
634 static void RENAME(yuv2rgb555_X_ar)(SwsContext *c, const int16_t *lumFilter,
635 const int16_t **lumSrc, int lumFilterSize,
636 const int16_t *chrFilter, const int16_t **chrUSrc,
637 const int16_t **chrVSrc,
638 int chrFilterSize, const int16_t **alpSrc,
639 uint8_t *dest, int dstW, int dstY)
642 x86_reg dstW_reg = dstW;
643 x86_reg uv_off = c->uv_off << 1;
645 YSCALEYUV2PACKEDX_ACCURATE
647 "pxor %%mm7, %%mm7 \n\t"
648 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
650 "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"
651 "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"
652 "paddusb "RED_DITHER"(%0), %%mm5\n\t"
654 WRITERGB15(%4, %5, %%REGa)
655 YSCALEYUV2PACKEDX_END
658 static void RENAME(yuv2rgb555_X)(SwsContext *c, const int16_t *lumFilter,
659 const int16_t **lumSrc, int lumFilterSize,
660 const int16_t *chrFilter, const int16_t **chrUSrc,
661 const int16_t **chrVSrc,
662 int chrFilterSize, const int16_t **alpSrc,
663 uint8_t *dest, int dstW, int dstY)
666 x86_reg dstW_reg = dstW;
667 x86_reg uv_off = c->uv_off << 1;
671 "pxor %%mm7, %%mm7 \n\t"
672 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
674 "paddusb "BLUE_DITHER"(%0), %%mm2 \n\t"
675 "paddusb "GREEN_DITHER"(%0), %%mm4 \n\t"
676 "paddusb "RED_DITHER"(%0), %%mm5 \n\t"
678 WRITERGB15(%4, %5, %%REGa)
679 YSCALEYUV2PACKEDX_END
682 #define WRITEBGR24MMX(dst, dstw, index) \
683 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
684 "movq %%mm2, %%mm1 \n\t" /* B */\
685 "movq %%mm5, %%mm6 \n\t" /* R */\
686 "punpcklbw %%mm4, %%mm2 \n\t" /* GBGBGBGB 0 */\
687 "punpcklbw %%mm7, %%mm5 \n\t" /* 0R0R0R0R 0 */\
688 "punpckhbw %%mm4, %%mm1 \n\t" /* GBGBGBGB 2 */\
689 "punpckhbw %%mm7, %%mm6 \n\t" /* 0R0R0R0R 2 */\
690 "movq %%mm2, %%mm0 \n\t" /* GBGBGBGB 0 */\
691 "movq %%mm1, %%mm3 \n\t" /* GBGBGBGB 2 */\
692 "punpcklwd %%mm5, %%mm0 \n\t" /* 0RGB0RGB 0 */\
693 "punpckhwd %%mm5, %%mm2 \n\t" /* 0RGB0RGB 1 */\
694 "punpcklwd %%mm6, %%mm1 \n\t" /* 0RGB0RGB 2 */\
695 "punpckhwd %%mm6, %%mm3 \n\t" /* 0RGB0RGB 3 */\
697 "movq %%mm0, %%mm4 \n\t" /* 0RGB0RGB 0 */\
698 "movq %%mm2, %%mm6 \n\t" /* 0RGB0RGB 1 */\
699 "movq %%mm1, %%mm5 \n\t" /* 0RGB0RGB 2 */\
700 "movq %%mm3, %%mm7 \n\t" /* 0RGB0RGB 3 */\
702 "psllq $40, %%mm0 \n\t" /* RGB00000 0 */\
703 "psllq $40, %%mm2 \n\t" /* RGB00000 1 */\
704 "psllq $40, %%mm1 \n\t" /* RGB00000 2 */\
705 "psllq $40, %%mm3 \n\t" /* RGB00000 3 */\
707 "punpckhdq %%mm4, %%mm0 \n\t" /* 0RGBRGB0 0 */\
708 "punpckhdq %%mm6, %%mm2 \n\t" /* 0RGBRGB0 1 */\
709 "punpckhdq %%mm5, %%mm1 \n\t" /* 0RGBRGB0 2 */\
710 "punpckhdq %%mm7, %%mm3 \n\t" /* 0RGBRGB0 3 */\
712 "psrlq $8, %%mm0 \n\t" /* 00RGBRGB 0 */\
713 "movq %%mm2, %%mm6 \n\t" /* 0RGBRGB0 1 */\
714 "psllq $40, %%mm2 \n\t" /* GB000000 1 */\
715 "por %%mm2, %%mm0 \n\t" /* GBRGBRGB 0 */\
716 MOVNTQ(%%mm0, (dst))\
718 "psrlq $24, %%mm6 \n\t" /* 0000RGBR 1 */\
719 "movq %%mm1, %%mm5 \n\t" /* 0RGBRGB0 2 */\
720 "psllq $24, %%mm1 \n\t" /* BRGB0000 2 */\
721 "por %%mm1, %%mm6 \n\t" /* BRGBRGBR 1 */\
722 MOVNTQ(%%mm6, 8(dst))\
724 "psrlq $40, %%mm5 \n\t" /* 000000RG 2 */\
725 "psllq $8, %%mm3 \n\t" /* RGBRGB00 3 */\
726 "por %%mm3, %%mm5 \n\t" /* RGBRGBRG 2 */\
727 MOVNTQ(%%mm5, 16(dst))\
729 "add $24, "#dst" \n\t"\
731 "add $8, "#index" \n\t"\
732 "cmp "#dstw", "#index" \n\t"\
735 #define WRITEBGR24MMX2(dst, dstw, index) \
736 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
737 "movq "MANGLE(ff_M24A)", %%mm0 \n\t"\
738 "movq "MANGLE(ff_M24C)", %%mm7 \n\t"\
739 "pshufw $0x50, %%mm2, %%mm1 \n\t" /* B3 B2 B3 B2 B1 B0 B1 B0 */\
740 "pshufw $0x50, %%mm4, %%mm3 \n\t" /* G3 G2 G3 G2 G1 G0 G1 G0 */\
741 "pshufw $0x00, %%mm5, %%mm6 \n\t" /* R1 R0 R1 R0 R1 R0 R1 R0 */\
743 "pand %%mm0, %%mm1 \n\t" /* B2 B1 B0 */\
744 "pand %%mm0, %%mm3 \n\t" /* G2 G1 G0 */\
745 "pand %%mm7, %%mm6 \n\t" /* R1 R0 */\
747 "psllq $8, %%mm3 \n\t" /* G2 G1 G0 */\
748 "por %%mm1, %%mm6 \n\t"\
749 "por %%mm3, %%mm6 \n\t"\
750 MOVNTQ(%%mm6, (dst))\
752 "psrlq $8, %%mm4 \n\t" /* 00 G7 G6 G5 G4 G3 G2 G1 */\
753 "pshufw $0xA5, %%mm2, %%mm1 \n\t" /* B5 B4 B5 B4 B3 B2 B3 B2 */\
754 "pshufw $0x55, %%mm4, %%mm3 \n\t" /* G4 G3 G4 G3 G4 G3 G4 G3 */\
755 "pshufw $0xA5, %%mm5, %%mm6 \n\t" /* R5 R4 R5 R4 R3 R2 R3 R2 */\
757 "pand "MANGLE(ff_M24B)", %%mm1 \n\t" /* B5 B4 B3 */\
758 "pand %%mm7, %%mm3 \n\t" /* G4 G3 */\
759 "pand %%mm0, %%mm6 \n\t" /* R4 R3 R2 */\
761 "por %%mm1, %%mm3 \n\t" /* B5 G4 B4 G3 B3 */\
762 "por %%mm3, %%mm6 \n\t"\
763 MOVNTQ(%%mm6, 8(dst))\
765 "pshufw $0xFF, %%mm2, %%mm1 \n\t" /* B7 B6 B7 B6 B7 B6 B6 B7 */\
766 "pshufw $0xFA, %%mm4, %%mm3 \n\t" /* 00 G7 00 G7 G6 G5 G6 G5 */\
767 "pshufw $0xFA, %%mm5, %%mm6 \n\t" /* R7 R6 R7 R6 R5 R4 R5 R4 */\
769 "pand %%mm7, %%mm1 \n\t" /* B7 B6 */\
770 "pand %%mm0, %%mm3 \n\t" /* G7 G6 G5 */\
771 "pand "MANGLE(ff_M24B)", %%mm6 \n\t" /* R7 R6 R5 */\
773 "por %%mm1, %%mm3 \n\t"\
774 "por %%mm3, %%mm6 \n\t"\
775 MOVNTQ(%%mm6, 16(dst))\
777 "add $24, "#dst" \n\t"\
779 "add $8, "#index" \n\t"\
780 "cmp "#dstw", "#index" \n\t"\
783 #if COMPILE_TEMPLATE_MMX2
785 #define WRITEBGR24(dst, dstw, index) WRITEBGR24MMX2(dst, dstw, index)
788 #define WRITEBGR24(dst, dstw, index) WRITEBGR24MMX(dst, dstw, index)
791 static void RENAME(yuv2bgr24_X_ar)(SwsContext *c, const int16_t *lumFilter,
792 const int16_t **lumSrc, int lumFilterSize,
793 const int16_t *chrFilter, const int16_t **chrUSrc,
794 const int16_t **chrVSrc,
795 int chrFilterSize, const int16_t **alpSrc,
796 uint8_t *dest, int dstW, int dstY)
799 x86_reg dstW_reg = dstW;
800 x86_reg uv_off = c->uv_off << 1;
802 YSCALEYUV2PACKEDX_ACCURATE
804 "pxor %%mm7, %%mm7 \n\t"
805 "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"\n\t" //FIXME optimize
806 "add %4, %%"REG_c" \n\t"
807 WRITEBGR24(%%REGc, %5, %%REGa)
808 :: "r" (&c->redDither),
809 "m" (dummy), "m" (dummy), "m" (dummy),
810 "r" (dest), "m" (dstW_reg), "m"(uv_off)
811 : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S
815 static void RENAME(yuv2bgr24_X)(SwsContext *c, const int16_t *lumFilter,
816 const int16_t **lumSrc, int lumFilterSize,
817 const int16_t *chrFilter, const int16_t **chrUSrc,
818 const int16_t **chrVSrc,
819 int chrFilterSize, const int16_t **alpSrc,
820 uint8_t *dest, int dstW, int dstY)
823 x86_reg dstW_reg = dstW;
824 x86_reg uv_off = c->uv_off << 1;
828 "pxor %%mm7, %%mm7 \n\t"
829 "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c" \n\t" //FIXME optimize
830 "add %4, %%"REG_c" \n\t"
831 WRITEBGR24(%%REGc, %5, %%REGa)
832 :: "r" (&c->redDither),
833 "m" (dummy), "m" (dummy), "m" (dummy),
834 "r" (dest), "m" (dstW_reg), "m"(uv_off)
835 : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S
839 #define REAL_WRITEYUY2(dst, dstw, index) \
840 "packuswb %%mm3, %%mm3 \n\t"\
841 "packuswb %%mm4, %%mm4 \n\t"\
842 "packuswb %%mm7, %%mm1 \n\t"\
843 "punpcklbw %%mm4, %%mm3 \n\t"\
844 "movq %%mm1, %%mm7 \n\t"\
845 "punpcklbw %%mm3, %%mm1 \n\t"\
846 "punpckhbw %%mm3, %%mm7 \n\t"\
848 MOVNTQ(%%mm1, (dst, index, 2))\
849 MOVNTQ(%%mm7, 8(dst, index, 2))\
851 "add $8, "#index" \n\t"\
852 "cmp "#dstw", "#index" \n\t"\
854 #define WRITEYUY2(dst, dstw, index) REAL_WRITEYUY2(dst, dstw, index)
856 static void RENAME(yuv2yuyv422_X_ar)(SwsContext *c, const int16_t *lumFilter,
857 const int16_t **lumSrc, int lumFilterSize,
858 const int16_t *chrFilter, const int16_t **chrUSrc,
859 const int16_t **chrVSrc,
860 int chrFilterSize, const int16_t **alpSrc,
861 uint8_t *dest, int dstW, int dstY)
864 x86_reg dstW_reg = dstW;
865 x86_reg uv_off = c->uv_off << 1;
867 YSCALEYUV2PACKEDX_ACCURATE
868 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
869 "psraw $3, %%mm3 \n\t"
870 "psraw $3, %%mm4 \n\t"
871 "psraw $3, %%mm1 \n\t"
872 "psraw $3, %%mm7 \n\t"
873 WRITEYUY2(%4, %5, %%REGa)
874 YSCALEYUV2PACKEDX_END
877 static void RENAME(yuv2yuyv422_X)(SwsContext *c, const int16_t *lumFilter,
878 const int16_t **lumSrc, int lumFilterSize,
879 const int16_t *chrFilter, const int16_t **chrUSrc,
880 const int16_t **chrVSrc,
881 int chrFilterSize, const int16_t **alpSrc,
882 uint8_t *dest, int dstW, int dstY)
885 x86_reg dstW_reg = dstW;
886 x86_reg uv_off = c->uv_off << 1;
889 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
890 "psraw $3, %%mm3 \n\t"
891 "psraw $3, %%mm4 \n\t"
892 "psraw $3, %%mm1 \n\t"
893 "psraw $3, %%mm7 \n\t"
894 WRITEYUY2(%4, %5, %%REGa)
895 YSCALEYUV2PACKEDX_END
898 #define REAL_YSCALEYUV2RGB_UV(index, c) \
899 "xor "#index", "#index" \n\t"\
902 "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
903 "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
904 "add "UV_OFFx2"("#c"), "#index" \n\t" \
905 "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
906 "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
907 "sub "UV_OFFx2"("#c"), "#index" \n\t" \
908 "psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
909 "psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
910 "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\
911 "pmulhw %%mm0, %%mm2 \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
912 "pmulhw %%mm0, %%mm5 \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
913 "psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
914 "psraw $4, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
915 "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
916 "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
917 "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\
918 "psubw "V_OFFSET"("#c"), %%mm4 \n\t" /* (V-128)8*/\
919 "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\
920 "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\
921 "pmulhw "UG_COEFF"("#c"), %%mm3 \n\t"\
922 "pmulhw "VG_COEFF"("#c"), %%mm4 \n\t"\
923 /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
925 #define REAL_YSCALEYUV2RGB_YA(index, c, b1, b2) \
926 "movq ("#b1", "#index", 2), %%mm0 \n\t" /*buf0[eax]*/\
927 "movq ("#b2", "#index", 2), %%mm1 \n\t" /*buf1[eax]*/\
928 "movq 8("#b1", "#index", 2), %%mm6 \n\t" /*buf0[eax]*/\
929 "movq 8("#b2", "#index", 2), %%mm7 \n\t" /*buf1[eax]*/\
930 "psubw %%mm1, %%mm0 \n\t" /* buf0[eax] - buf1[eax]*/\
931 "psubw %%mm7, %%mm6 \n\t" /* buf0[eax] - buf1[eax]*/\
932 "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
933 "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
934 "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
935 "psraw $4, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
936 "paddw %%mm0, %%mm1 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
937 "paddw %%mm6, %%mm7 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
939 #define REAL_YSCALEYUV2RGB_COEFF(c) \
940 "pmulhw "UB_COEFF"("#c"), %%mm2 \n\t"\
941 "pmulhw "VR_COEFF"("#c"), %%mm5 \n\t"\
942 "psubw "Y_OFFSET"("#c"), %%mm1 \n\t" /* 8(Y-16)*/\
943 "psubw "Y_OFFSET"("#c"), %%mm7 \n\t" /* 8(Y-16)*/\
944 "pmulhw "Y_COEFF"("#c"), %%mm1 \n\t"\
945 "pmulhw "Y_COEFF"("#c"), %%mm7 \n\t"\
946 /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
947 "paddw %%mm3, %%mm4 \n\t"\
948 "movq %%mm2, %%mm0 \n\t"\
949 "movq %%mm5, %%mm6 \n\t"\
950 "movq %%mm4, %%mm3 \n\t"\
951 "punpcklwd %%mm2, %%mm2 \n\t"\
952 "punpcklwd %%mm5, %%mm5 \n\t"\
953 "punpcklwd %%mm4, %%mm4 \n\t"\
954 "paddw %%mm1, %%mm2 \n\t"\
955 "paddw %%mm1, %%mm5 \n\t"\
956 "paddw %%mm1, %%mm4 \n\t"\
957 "punpckhwd %%mm0, %%mm0 \n\t"\
958 "punpckhwd %%mm6, %%mm6 \n\t"\
959 "punpckhwd %%mm3, %%mm3 \n\t"\
960 "paddw %%mm7, %%mm0 \n\t"\
961 "paddw %%mm7, %%mm6 \n\t"\
962 "paddw %%mm7, %%mm3 \n\t"\
963 /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
964 "packuswb %%mm0, %%mm2 \n\t"\
965 "packuswb %%mm6, %%mm5 \n\t"\
966 "packuswb %%mm3, %%mm4 \n\t"\
968 #define YSCALEYUV2RGB_YA(index, c, b1, b2) REAL_YSCALEYUV2RGB_YA(index, c, b1, b2)
970 #define YSCALEYUV2RGB(index, c) \
971 REAL_YSCALEYUV2RGB_UV(index, c) \
972 REAL_YSCALEYUV2RGB_YA(index, c, %0, %1) \
973 REAL_YSCALEYUV2RGB_COEFF(c)
976 * vertical bilinear scale YV12 to RGB
978 static void RENAME(yuv2rgb32_2)(SwsContext *c, const int16_t *buf[2],
979 const int16_t *ubuf[2], const int16_t *vbuf[2],
980 const int16_t *abuf[2], uint8_t *dest,
981 int dstW, int yalpha, int uvalpha, int y)
983 const int16_t *buf0 = buf[0], *buf1 = buf[1],
984 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
986 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
987 const int16_t *abuf0 = abuf[0], *abuf1 = abuf[1];
990 YSCALEYUV2RGB(%%r8, %5)
991 YSCALEYUV2RGB_YA(%%r8, %5, %6, %7)
992 "psraw $3, %%mm1 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
993 "psraw $3, %%mm7 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
994 "packuswb %%mm7, %%mm1 \n\t"
995 WRITEBGR32(%4, 8280(%5), %%r8, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
996 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "r" (dest),
998 "r" (abuf0), "r" (abuf1)
1002 *(const uint16_t **)(&c->u_temp)=abuf0;
1003 *(const uint16_t **)(&c->v_temp)=abuf1;
1005 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1006 "mov %4, %%"REG_b" \n\t"
1007 "push %%"REG_BP" \n\t"
1008 YSCALEYUV2RGB(%%REGBP, %5)
1011 "mov "U_TEMP"(%5), %0 \n\t"
1012 "mov "V_TEMP"(%5), %1 \n\t"
1013 YSCALEYUV2RGB_YA(%%REGBP, %5, %0, %1)
1014 "psraw $3, %%mm1 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
1015 "psraw $3, %%mm7 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
1016 "packuswb %%mm7, %%mm1 \n\t"
1019 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
1020 "pop %%"REG_BP" \n\t"
1021 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1022 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1028 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1029 "mov %4, %%"REG_b" \n\t"
1030 "push %%"REG_BP" \n\t"
1031 YSCALEYUV2RGB(%%REGBP, %5)
1032 "pcmpeqd %%mm7, %%mm7 \n\t"
1033 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1034 "pop %%"REG_BP" \n\t"
1035 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1036 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1042 static void RENAME(yuv2bgr24_2)(SwsContext *c, const int16_t *buf[2],
1043 const int16_t *ubuf[2], const int16_t *vbuf[2],
1044 const int16_t *abuf[2], uint8_t *dest,
1045 int dstW, int yalpha, int uvalpha, int y)
1047 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1048 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1050 //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(
1052 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1053 "mov %4, %%"REG_b" \n\t"
1054 "push %%"REG_BP" \n\t"
1055 YSCALEYUV2RGB(%%REGBP, %5)
1056 "pxor %%mm7, %%mm7 \n\t"
1057 WRITEBGR24(%%REGb, 8280(%5), %%REGBP)
1058 "pop %%"REG_BP" \n\t"
1059 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1060 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1065 static void RENAME(yuv2rgb555_2)(SwsContext *c, const int16_t *buf[2],
1066 const int16_t *ubuf[2], const int16_t *vbuf[2],
1067 const int16_t *abuf[2], uint8_t *dest,
1068 int dstW, int yalpha, int uvalpha, int y)
1070 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1071 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1073 //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(
1075 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1076 "mov %4, %%"REG_b" \n\t"
1077 "push %%"REG_BP" \n\t"
1078 YSCALEYUV2RGB(%%REGBP, %5)
1079 "pxor %%mm7, %%mm7 \n\t"
1080 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1082 "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t"
1083 "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t"
1084 "paddusb "RED_DITHER"(%5), %%mm5 \n\t"
1086 WRITERGB15(%%REGb, 8280(%5), %%REGBP)
1087 "pop %%"REG_BP" \n\t"
1088 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1089 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1094 static void RENAME(yuv2rgb565_2)(SwsContext *c, const int16_t *buf[2],
1095 const int16_t *ubuf[2], const int16_t *vbuf[2],
1096 const int16_t *abuf[2], uint8_t *dest,
1097 int dstW, int yalpha, int uvalpha, int y)
1099 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1100 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1102 //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(
1104 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1105 "mov %4, %%"REG_b" \n\t"
1106 "push %%"REG_BP" \n\t"
1107 YSCALEYUV2RGB(%%REGBP, %5)
1108 "pxor %%mm7, %%mm7 \n\t"
1109 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1111 "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t"
1112 "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t"
1113 "paddusb "RED_DITHER"(%5), %%mm5 \n\t"
1115 WRITERGB16(%%REGb, 8280(%5), %%REGBP)
1116 "pop %%"REG_BP" \n\t"
1117 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1118 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1123 #define REAL_YSCALEYUV2PACKED(index, c) \
1124 "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\
1125 "movq "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm1 \n\t"\
1126 "psraw $3, %%mm0 \n\t"\
1127 "psraw $3, %%mm1 \n\t"\
1128 "movq %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c") \n\t"\
1129 "movq %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c") \n\t"\
1130 "xor "#index", "#index" \n\t"\
1133 "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
1134 "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
1135 "add "UV_OFFx2"("#c"), "#index" \n\t" \
1136 "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
1137 "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
1138 "sub "UV_OFFx2"("#c"), "#index" \n\t" \
1139 "psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
1140 "psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
1141 "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\
1142 "pmulhw %%mm0, %%mm2 \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
1143 "pmulhw %%mm0, %%mm5 \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
1144 "psraw $7, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
1145 "psraw $7, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
1146 "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
1147 "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
1148 "movq (%0, "#index", 2), %%mm0 \n\t" /*buf0[eax]*/\
1149 "movq (%1, "#index", 2), %%mm1 \n\t" /*buf1[eax]*/\
1150 "movq 8(%0, "#index", 2), %%mm6 \n\t" /*buf0[eax]*/\
1151 "movq 8(%1, "#index", 2), %%mm7 \n\t" /*buf1[eax]*/\
1152 "psubw %%mm1, %%mm0 \n\t" /* buf0[eax] - buf1[eax]*/\
1153 "psubw %%mm7, %%mm6 \n\t" /* buf0[eax] - buf1[eax]*/\
1154 "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
1155 "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
1156 "psraw $7, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1157 "psraw $7, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1158 "paddw %%mm0, %%mm1 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
1159 "paddw %%mm6, %%mm7 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
1161 #define YSCALEYUV2PACKED(index, c) REAL_YSCALEYUV2PACKED(index, c)
1163 static void RENAME(yuv2yuyv422_2)(SwsContext *c, const int16_t *buf[2],
1164 const int16_t *ubuf[2], const int16_t *vbuf[2],
1165 const int16_t *abuf[2], uint8_t *dest,
1166 int dstW, int yalpha, int uvalpha, int y)
1168 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1169 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1171 //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(
1173 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1174 "mov %4, %%"REG_b" \n\t"
1175 "push %%"REG_BP" \n\t"
1176 YSCALEYUV2PACKED(%%REGBP, %5)
1177 WRITEYUY2(%%REGb, 8280(%5), %%REGBP)
1178 "pop %%"REG_BP" \n\t"
1179 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1180 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1185 #define REAL_YSCALEYUV2RGB1(index, c) \
1186 "xor "#index", "#index" \n\t"\
1189 "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\
1190 "add "UV_OFFx2"("#c"), "#index" \n\t" \
1191 "movq (%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
1192 "sub "UV_OFFx2"("#c"), "#index" \n\t" \
1193 "psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
1194 "psraw $4, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
1195 "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\
1196 "psubw "V_OFFSET"("#c"), %%mm4 \n\t" /* (V-128)8*/\
1197 "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\
1198 "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\
1199 "pmulhw "UG_COEFF"("#c"), %%mm3 \n\t"\
1200 "pmulhw "VG_COEFF"("#c"), %%mm4 \n\t"\
1201 /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
1202 "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\
1203 "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\
1204 "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1205 "psraw $4, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1206 "pmulhw "UB_COEFF"("#c"), %%mm2 \n\t"\
1207 "pmulhw "VR_COEFF"("#c"), %%mm5 \n\t"\
1208 "psubw "Y_OFFSET"("#c"), %%mm1 \n\t" /* 8(Y-16)*/\
1209 "psubw "Y_OFFSET"("#c"), %%mm7 \n\t" /* 8(Y-16)*/\
1210 "pmulhw "Y_COEFF"("#c"), %%mm1 \n\t"\
1211 "pmulhw "Y_COEFF"("#c"), %%mm7 \n\t"\
1212 /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
1213 "paddw %%mm3, %%mm4 \n\t"\
1214 "movq %%mm2, %%mm0 \n\t"\
1215 "movq %%mm5, %%mm6 \n\t"\
1216 "movq %%mm4, %%mm3 \n\t"\
1217 "punpcklwd %%mm2, %%mm2 \n\t"\
1218 "punpcklwd %%mm5, %%mm5 \n\t"\
1219 "punpcklwd %%mm4, %%mm4 \n\t"\
1220 "paddw %%mm1, %%mm2 \n\t"\
1221 "paddw %%mm1, %%mm5 \n\t"\
1222 "paddw %%mm1, %%mm4 \n\t"\
1223 "punpckhwd %%mm0, %%mm0 \n\t"\
1224 "punpckhwd %%mm6, %%mm6 \n\t"\
1225 "punpckhwd %%mm3, %%mm3 \n\t"\
1226 "paddw %%mm7, %%mm0 \n\t"\
1227 "paddw %%mm7, %%mm6 \n\t"\
1228 "paddw %%mm7, %%mm3 \n\t"\
1229 /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
1230 "packuswb %%mm0, %%mm2 \n\t"\
1231 "packuswb %%mm6, %%mm5 \n\t"\
1232 "packuswb %%mm3, %%mm4 \n\t"\
1234 #define YSCALEYUV2RGB1(index, c) REAL_YSCALEYUV2RGB1(index, c)
1236 // do vertical chrominance interpolation
1237 #define REAL_YSCALEYUV2RGB1b(index, c) \
1238 "xor "#index", "#index" \n\t"\
1241 "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
1242 "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
1243 "add "UV_OFFx2"("#c"), "#index" \n\t" \
1244 "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
1245 "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
1246 "sub "UV_OFFx2"("#c"), "#index" \n\t" \
1247 "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
1248 "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
1249 "psrlw $5, %%mm3 \n\t" /*FIXME might overflow*/\
1250 "psrlw $5, %%mm4 \n\t" /*FIXME might overflow*/\
1251 "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\
1252 "psubw "V_OFFSET"("#c"), %%mm4 \n\t" /* (V-128)8*/\
1253 "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\
1254 "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\
1255 "pmulhw "UG_COEFF"("#c"), %%mm3 \n\t"\
1256 "pmulhw "VG_COEFF"("#c"), %%mm4 \n\t"\
1257 /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
1258 "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\
1259 "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\
1260 "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1261 "psraw $4, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
1262 "pmulhw "UB_COEFF"("#c"), %%mm2 \n\t"\
1263 "pmulhw "VR_COEFF"("#c"), %%mm5 \n\t"\
1264 "psubw "Y_OFFSET"("#c"), %%mm1 \n\t" /* 8(Y-16)*/\
1265 "psubw "Y_OFFSET"("#c"), %%mm7 \n\t" /* 8(Y-16)*/\
1266 "pmulhw "Y_COEFF"("#c"), %%mm1 \n\t"\
1267 "pmulhw "Y_COEFF"("#c"), %%mm7 \n\t"\
1268 /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
1269 "paddw %%mm3, %%mm4 \n\t"\
1270 "movq %%mm2, %%mm0 \n\t"\
1271 "movq %%mm5, %%mm6 \n\t"\
1272 "movq %%mm4, %%mm3 \n\t"\
1273 "punpcklwd %%mm2, %%mm2 \n\t"\
1274 "punpcklwd %%mm5, %%mm5 \n\t"\
1275 "punpcklwd %%mm4, %%mm4 \n\t"\
1276 "paddw %%mm1, %%mm2 \n\t"\
1277 "paddw %%mm1, %%mm5 \n\t"\
1278 "paddw %%mm1, %%mm4 \n\t"\
1279 "punpckhwd %%mm0, %%mm0 \n\t"\
1280 "punpckhwd %%mm6, %%mm6 \n\t"\
1281 "punpckhwd %%mm3, %%mm3 \n\t"\
1282 "paddw %%mm7, %%mm0 \n\t"\
1283 "paddw %%mm7, %%mm6 \n\t"\
1284 "paddw %%mm7, %%mm3 \n\t"\
1285 /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
1286 "packuswb %%mm0, %%mm2 \n\t"\
1287 "packuswb %%mm6, %%mm5 \n\t"\
1288 "packuswb %%mm3, %%mm4 \n\t"\
1290 #define YSCALEYUV2RGB1b(index, c) REAL_YSCALEYUV2RGB1b(index, c)
1292 #define REAL_YSCALEYUV2RGB1_ALPHA(index) \
1293 "movq (%1, "#index", 2), %%mm7 \n\t" /* abuf0[index ] */\
1294 "movq 8(%1, "#index", 2), %%mm1 \n\t" /* abuf0[index+4] */\
1295 "psraw $7, %%mm7 \n\t" /* abuf0[index ] >>7 */\
1296 "psraw $7, %%mm1 \n\t" /* abuf0[index+4] >>7 */\
1297 "packuswb %%mm1, %%mm7 \n\t"
1298 #define YSCALEYUV2RGB1_ALPHA(index) REAL_YSCALEYUV2RGB1_ALPHA(index)
1301 * YV12 to RGB without scaling or interpolating
1303 static void RENAME(yuv2rgb32_1)(SwsContext *c, const int16_t *buf0,
1304 const int16_t *ubuf[2], const int16_t *bguf[2],
1305 const int16_t *abuf0, uint8_t *dest,
1306 int dstW, int uvalpha, int y)
1308 const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1309 const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1311 if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1312 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1314 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1315 "mov %4, %%"REG_b" \n\t"
1316 "push %%"REG_BP" \n\t"
1317 YSCALEYUV2RGB1(%%REGBP, %5)
1318 YSCALEYUV2RGB1_ALPHA(%%REGBP)
1319 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1320 "pop %%"REG_BP" \n\t"
1321 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1322 :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1327 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1328 "mov %4, %%"REG_b" \n\t"
1329 "push %%"REG_BP" \n\t"
1330 YSCALEYUV2RGB1(%%REGBP, %5)
1331 "pcmpeqd %%mm7, %%mm7 \n\t"
1332 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1333 "pop %%"REG_BP" \n\t"
1334 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1335 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1340 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1342 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1343 "mov %4, %%"REG_b" \n\t"
1344 "push %%"REG_BP" \n\t"
1345 YSCALEYUV2RGB1b(%%REGBP, %5)
1346 YSCALEYUV2RGB1_ALPHA(%%REGBP)
1347 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1348 "pop %%"REG_BP" \n\t"
1349 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1350 :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1355 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1356 "mov %4, %%"REG_b" \n\t"
1357 "push %%"REG_BP" \n\t"
1358 YSCALEYUV2RGB1b(%%REGBP, %5)
1359 "pcmpeqd %%mm7, %%mm7 \n\t"
1360 WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1361 "pop %%"REG_BP" \n\t"
1362 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1363 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1370 static void RENAME(yuv2bgr24_1)(SwsContext *c, const int16_t *buf0,
1371 const int16_t *ubuf[2], const int16_t *bguf[2],
1372 const int16_t *abuf0, uint8_t *dest,
1373 int dstW, int uvalpha, int y)
1375 const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1376 const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1378 if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1380 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1381 "mov %4, %%"REG_b" \n\t"
1382 "push %%"REG_BP" \n\t"
1383 YSCALEYUV2RGB1(%%REGBP, %5)
1384 "pxor %%mm7, %%mm7 \n\t"
1385 WRITEBGR24(%%REGb, 8280(%5), %%REGBP)
1386 "pop %%"REG_BP" \n\t"
1387 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1388 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1393 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1394 "mov %4, %%"REG_b" \n\t"
1395 "push %%"REG_BP" \n\t"
1396 YSCALEYUV2RGB1b(%%REGBP, %5)
1397 "pxor %%mm7, %%mm7 \n\t"
1398 WRITEBGR24(%%REGb, 8280(%5), %%REGBP)
1399 "pop %%"REG_BP" \n\t"
1400 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1401 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1407 static void RENAME(yuv2rgb555_1)(SwsContext *c, const int16_t *buf0,
1408 const int16_t *ubuf[2], const int16_t *bguf[2],
1409 const int16_t *abuf0, uint8_t *dest,
1410 int dstW, int uvalpha, int y)
1412 const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1413 const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1415 if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1417 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1418 "mov %4, %%"REG_b" \n\t"
1419 "push %%"REG_BP" \n\t"
1420 YSCALEYUV2RGB1(%%REGBP, %5)
1421 "pxor %%mm7, %%mm7 \n\t"
1422 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1424 "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t"
1425 "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t"
1426 "paddusb "RED_DITHER"(%5), %%mm5 \n\t"
1428 WRITERGB15(%%REGb, 8280(%5), %%REGBP)
1429 "pop %%"REG_BP" \n\t"
1430 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1431 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1436 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1437 "mov %4, %%"REG_b" \n\t"
1438 "push %%"REG_BP" \n\t"
1439 YSCALEYUV2RGB1b(%%REGBP, %5)
1440 "pxor %%mm7, %%mm7 \n\t"
1441 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1443 "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t"
1444 "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t"
1445 "paddusb "RED_DITHER"(%5), %%mm5 \n\t"
1447 WRITERGB15(%%REGb, 8280(%5), %%REGBP)
1448 "pop %%"REG_BP" \n\t"
1449 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1450 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1456 static void RENAME(yuv2rgb565_1)(SwsContext *c, const int16_t *buf0,
1457 const int16_t *ubuf[2], const int16_t *bguf[2],
1458 const int16_t *abuf0, uint8_t *dest,
1459 int dstW, int uvalpha, int y)
1461 const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1462 const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1464 if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1466 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1467 "mov %4, %%"REG_b" \n\t"
1468 "push %%"REG_BP" \n\t"
1469 YSCALEYUV2RGB1(%%REGBP, %5)
1470 "pxor %%mm7, %%mm7 \n\t"
1471 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1473 "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t"
1474 "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t"
1475 "paddusb "RED_DITHER"(%5), %%mm5 \n\t"
1477 WRITERGB16(%%REGb, 8280(%5), %%REGBP)
1478 "pop %%"REG_BP" \n\t"
1479 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1480 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1485 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1486 "mov %4, %%"REG_b" \n\t"
1487 "push %%"REG_BP" \n\t"
1488 YSCALEYUV2RGB1b(%%REGBP, %5)
1489 "pxor %%mm7, %%mm7 \n\t"
1490 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1492 "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t"
1493 "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t"
1494 "paddusb "RED_DITHER"(%5), %%mm5 \n\t"
1496 WRITERGB16(%%REGb, 8280(%5), %%REGBP)
1497 "pop %%"REG_BP" \n\t"
1498 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1499 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1505 #define REAL_YSCALEYUV2PACKED1(index, c) \
1506 "xor "#index", "#index" \n\t"\
1509 "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\
1510 "add "UV_OFFx2"("#c"), "#index" \n\t" \
1511 "movq (%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
1512 "sub "UV_OFFx2"("#c"), "#index" \n\t" \
1513 "psraw $7, %%mm3 \n\t" \
1514 "psraw $7, %%mm4 \n\t" \
1515 "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\
1516 "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\
1517 "psraw $7, %%mm1 \n\t" \
1518 "psraw $7, %%mm7 \n\t" \
1520 #define YSCALEYUV2PACKED1(index, c) REAL_YSCALEYUV2PACKED1(index, c)
1522 #define REAL_YSCALEYUV2PACKED1b(index, c) \
1523 "xor "#index", "#index" \n\t"\
1526 "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
1527 "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
1528 "add "UV_OFFx2"("#c"), "#index" \n\t" \
1529 "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
1530 "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
1531 "sub "UV_OFFx2"("#c"), "#index" \n\t" \
1532 "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
1533 "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
1534 "psrlw $8, %%mm3 \n\t" \
1535 "psrlw $8, %%mm4 \n\t" \
1536 "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\
1537 "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\
1538 "psraw $7, %%mm1 \n\t" \
1539 "psraw $7, %%mm7 \n\t"
1540 #define YSCALEYUV2PACKED1b(index, c) REAL_YSCALEYUV2PACKED1b(index, c)
1542 static void RENAME(yuv2yuyv422_1)(SwsContext *c, const int16_t *buf0,
1543 const int16_t *ubuf[2], const int16_t *bguf[2],
1544 const int16_t *abuf0, uint8_t *dest,
1545 int dstW, int uvalpha, int y)
1547 const int16_t *ubuf0 = ubuf[0], *ubuf1 = ubuf[1];
1548 const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1550 if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1552 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1553 "mov %4, %%"REG_b" \n\t"
1554 "push %%"REG_BP" \n\t"
1555 YSCALEYUV2PACKED1(%%REGBP, %5)
1556 WRITEYUY2(%%REGb, 8280(%5), %%REGBP)
1557 "pop %%"REG_BP" \n\t"
1558 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1559 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1564 "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t"
1565 "mov %4, %%"REG_b" \n\t"
1566 "push %%"REG_BP" \n\t"
1567 YSCALEYUV2PACKED1b(%%REGBP, %5)
1568 WRITEYUY2(%%REGb, 8280(%5), %%REGBP)
1569 "pop %%"REG_BP" \n\t"
1570 "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t"
1571 :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest),
1577 #if !COMPILE_TEMPLATE_MMX2
1578 //FIXME yuy2* can read up to 7 samples too much
1580 static void RENAME(yuy2ToY)(uint8_t *dst, const uint8_t *src,
1581 int width, uint32_t *unused)
1584 "movq "MANGLE(bm01010101)", %%mm2 \n\t"
1585 "mov %0, %%"REG_a" \n\t"
1587 "movq (%1, %%"REG_a",2), %%mm0 \n\t"
1588 "movq 8(%1, %%"REG_a",2), %%mm1 \n\t"
1589 "pand %%mm2, %%mm0 \n\t"
1590 "pand %%mm2, %%mm1 \n\t"
1591 "packuswb %%mm1, %%mm0 \n\t"
1592 "movq %%mm0, (%2, %%"REG_a") \n\t"
1593 "add $8, %%"REG_a" \n\t"
1595 : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst+width)
1600 static void RENAME(yuy2ToUV)(uint8_t *dstU, uint8_t *dstV,
1601 const uint8_t *src1, const uint8_t *src2,
1602 int width, uint32_t *unused)
1605 "movq "MANGLE(bm01010101)", %%mm4 \n\t"
1606 "mov %0, %%"REG_a" \n\t"
1608 "movq (%1, %%"REG_a",4), %%mm0 \n\t"
1609 "movq 8(%1, %%"REG_a",4), %%mm1 \n\t"
1610 "psrlw $8, %%mm0 \n\t"
1611 "psrlw $8, %%mm1 \n\t"
1612 "packuswb %%mm1, %%mm0 \n\t"
1613 "movq %%mm0, %%mm1 \n\t"
1614 "psrlw $8, %%mm0 \n\t"
1615 "pand %%mm4, %%mm1 \n\t"
1616 "packuswb %%mm0, %%mm0 \n\t"
1617 "packuswb %%mm1, %%mm1 \n\t"
1618 "movd %%mm0, (%3, %%"REG_a") \n\t"
1619 "movd %%mm1, (%2, %%"REG_a") \n\t"
1620 "add $4, %%"REG_a" \n\t"
1622 : : "g" ((x86_reg)-width), "r" (src1+width*4), "r" (dstU+width), "r" (dstV+width)
1625 assert(src1 == src2);
1628 /* This is almost identical to the previous, end exists only because
1629 * yuy2ToY/UV)(dst, src+1, ...) would have 100% unaligned accesses. */
1630 static void RENAME(uyvyToY)(uint8_t *dst, const uint8_t *src,
1631 int width, uint32_t *unused)
1634 "mov %0, %%"REG_a" \n\t"
1636 "movq (%1, %%"REG_a",2), %%mm0 \n\t"
1637 "movq 8(%1, %%"REG_a",2), %%mm1 \n\t"
1638 "psrlw $8, %%mm0 \n\t"
1639 "psrlw $8, %%mm1 \n\t"
1640 "packuswb %%mm1, %%mm0 \n\t"
1641 "movq %%mm0, (%2, %%"REG_a") \n\t"
1642 "add $8, %%"REG_a" \n\t"
1644 : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst+width)
1649 static void RENAME(uyvyToUV)(uint8_t *dstU, uint8_t *dstV,
1650 const uint8_t *src1, const uint8_t *src2,
1651 int width, uint32_t *unused)
1654 "movq "MANGLE(bm01010101)", %%mm4 \n\t"
1655 "mov %0, %%"REG_a" \n\t"
1657 "movq (%1, %%"REG_a",4), %%mm0 \n\t"
1658 "movq 8(%1, %%"REG_a",4), %%mm1 \n\t"
1659 "pand %%mm4, %%mm0 \n\t"
1660 "pand %%mm4, %%mm1 \n\t"
1661 "packuswb %%mm1, %%mm0 \n\t"
1662 "movq %%mm0, %%mm1 \n\t"
1663 "psrlw $8, %%mm0 \n\t"
1664 "pand %%mm4, %%mm1 \n\t"
1665 "packuswb %%mm0, %%mm0 \n\t"
1666 "packuswb %%mm1, %%mm1 \n\t"
1667 "movd %%mm0, (%3, %%"REG_a") \n\t"
1668 "movd %%mm1, (%2, %%"REG_a") \n\t"
1669 "add $4, %%"REG_a" \n\t"
1671 : : "g" ((x86_reg)-width), "r" (src1+width*4), "r" (dstU+width), "r" (dstV+width)
1674 assert(src1 == src2);
1677 static av_always_inline void RENAME(nvXXtoUV)(uint8_t *dst1, uint8_t *dst2,
1678 const uint8_t *src, int width)
1681 "movq "MANGLE(bm01010101)", %%mm4 \n\t"
1682 "mov %0, %%"REG_a" \n\t"
1684 "movq (%1, %%"REG_a",2), %%mm0 \n\t"
1685 "movq 8(%1, %%"REG_a",2), %%mm1 \n\t"
1686 "movq %%mm0, %%mm2 \n\t"
1687 "movq %%mm1, %%mm3 \n\t"
1688 "pand %%mm4, %%mm0 \n\t"
1689 "pand %%mm4, %%mm1 \n\t"
1690 "psrlw $8, %%mm2 \n\t"
1691 "psrlw $8, %%mm3 \n\t"
1692 "packuswb %%mm1, %%mm0 \n\t"
1693 "packuswb %%mm3, %%mm2 \n\t"
1694 "movq %%mm0, (%2, %%"REG_a") \n\t"
1695 "movq %%mm2, (%3, %%"REG_a") \n\t"
1696 "add $8, %%"REG_a" \n\t"
1698 : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst1+width), "r" (dst2+width)
1703 static void RENAME(nv12ToUV)(uint8_t *dstU, uint8_t *dstV,
1704 const uint8_t *src1, const uint8_t *src2,
1705 int width, uint32_t *unused)
1707 RENAME(nvXXtoUV)(dstU, dstV, src1, width);
1710 static void RENAME(nv21ToUV)(uint8_t *dstU, uint8_t *dstV,
1711 const uint8_t *src1, const uint8_t *src2,
1712 int width, uint32_t *unused)
1714 RENAME(nvXXtoUV)(dstV, dstU, src1, width);
1716 #endif /* !COMPILE_TEMPLATE_MMX2 */
1718 static av_always_inline void RENAME(bgr24ToY_mmx)(uint8_t *dst, const uint8_t *src,
1719 int width, enum PixelFormat srcFormat)
1722 if(srcFormat == PIX_FMT_BGR24) {
1724 "movq "MANGLE(ff_bgr24toY1Coeff)", %%mm5 \n\t"
1725 "movq "MANGLE(ff_bgr24toY2Coeff)", %%mm6 \n\t"
1730 "movq "MANGLE(ff_rgb24toY1Coeff)", %%mm5 \n\t"
1731 "movq "MANGLE(ff_rgb24toY2Coeff)", %%mm6 \n\t"
1737 "movq "MANGLE(ff_bgr24toYOffset)", %%mm4 \n\t"
1738 "mov %2, %%"REG_a" \n\t"
1739 "pxor %%mm7, %%mm7 \n\t"
1741 PREFETCH" 64(%0) \n\t"
1742 "movd (%0), %%mm0 \n\t"
1743 "movd 2(%0), %%mm1 \n\t"
1744 "movd 6(%0), %%mm2 \n\t"
1745 "movd 8(%0), %%mm3 \n\t"
1747 "punpcklbw %%mm7, %%mm0 \n\t"
1748 "punpcklbw %%mm7, %%mm1 \n\t"
1749 "punpcklbw %%mm7, %%mm2 \n\t"
1750 "punpcklbw %%mm7, %%mm3 \n\t"
1751 "pmaddwd %%mm5, %%mm0 \n\t"
1752 "pmaddwd %%mm6, %%mm1 \n\t"
1753 "pmaddwd %%mm5, %%mm2 \n\t"
1754 "pmaddwd %%mm6, %%mm3 \n\t"
1755 "paddd %%mm1, %%mm0 \n\t"
1756 "paddd %%mm3, %%mm2 \n\t"
1757 "paddd %%mm4, %%mm0 \n\t"
1758 "paddd %%mm4, %%mm2 \n\t"
1759 "psrad $15, %%mm0 \n\t"
1760 "psrad $15, %%mm2 \n\t"
1761 "packssdw %%mm2, %%mm0 \n\t"
1762 "packuswb %%mm0, %%mm0 \n\t"
1763 "movd %%mm0, (%1, %%"REG_a") \n\t"
1764 "add $4, %%"REG_a" \n\t"
1767 : "r" (dst+width), "g" ((x86_reg)-width)
1772 static void RENAME(bgr24ToY)(uint8_t *dst, const uint8_t *src,
1773 int width, uint32_t *unused)
1775 RENAME(bgr24ToY_mmx)(dst, src, width, PIX_FMT_BGR24);
1778 static void RENAME(rgb24ToY)(uint8_t *dst, const uint8_t *src,
1779 int width, uint32_t *unused)
1781 RENAME(bgr24ToY_mmx)(dst, src, width, PIX_FMT_RGB24);
1784 static av_always_inline void RENAME(bgr24ToUV_mmx)(uint8_t *dstU, uint8_t *dstV,
1785 const uint8_t *src, int width,
1786 enum PixelFormat srcFormat)
1789 "movq 24(%4), %%mm6 \n\t"
1790 "mov %3, %%"REG_a" \n\t"
1791 "pxor %%mm7, %%mm7 \n\t"
1793 PREFETCH" 64(%0) \n\t"
1794 "movd (%0), %%mm0 \n\t"
1795 "movd 2(%0), %%mm1 \n\t"
1796 "punpcklbw %%mm7, %%mm0 \n\t"
1797 "punpcklbw %%mm7, %%mm1 \n\t"
1798 "movq %%mm0, %%mm2 \n\t"
1799 "movq %%mm1, %%mm3 \n\t"
1800 "pmaddwd (%4), %%mm0 \n\t"
1801 "pmaddwd 8(%4), %%mm1 \n\t"
1802 "pmaddwd 16(%4), %%mm2 \n\t"
1803 "pmaddwd %%mm6, %%mm3 \n\t"
1804 "paddd %%mm1, %%mm0 \n\t"
1805 "paddd %%mm3, %%mm2 \n\t"
1807 "movd 6(%0), %%mm1 \n\t"
1808 "movd 8(%0), %%mm3 \n\t"
1810 "punpcklbw %%mm7, %%mm1 \n\t"
1811 "punpcklbw %%mm7, %%mm3 \n\t"
1812 "movq %%mm1, %%mm4 \n\t"
1813 "movq %%mm3, %%mm5 \n\t"
1814 "pmaddwd (%4), %%mm1 \n\t"
1815 "pmaddwd 8(%4), %%mm3 \n\t"
1816 "pmaddwd 16(%4), %%mm4 \n\t"
1817 "pmaddwd %%mm6, %%mm5 \n\t"
1818 "paddd %%mm3, %%mm1 \n\t"
1819 "paddd %%mm5, %%mm4 \n\t"
1821 "movq "MANGLE(ff_bgr24toUVOffset)", %%mm3 \n\t"
1822 "paddd %%mm3, %%mm0 \n\t"
1823 "paddd %%mm3, %%mm2 \n\t"
1824 "paddd %%mm3, %%mm1 \n\t"
1825 "paddd %%mm3, %%mm4 \n\t"
1826 "psrad $15, %%mm0 \n\t"
1827 "psrad $15, %%mm2 \n\t"
1828 "psrad $15, %%mm1 \n\t"
1829 "psrad $15, %%mm4 \n\t"
1830 "packssdw %%mm1, %%mm0 \n\t"
1831 "packssdw %%mm4, %%mm2 \n\t"
1832 "packuswb %%mm0, %%mm0 \n\t"
1833 "packuswb %%mm2, %%mm2 \n\t"
1834 "movd %%mm0, (%1, %%"REG_a") \n\t"
1835 "movd %%mm2, (%2, %%"REG_a") \n\t"
1836 "add $4, %%"REG_a" \n\t"
1839 : "r" (dstU+width), "r" (dstV+width), "g" ((x86_reg)-width), "r"(ff_bgr24toUV[srcFormat == PIX_FMT_RGB24])
1844 static void RENAME(bgr24ToUV)(uint8_t *dstU, uint8_t *dstV,
1845 const uint8_t *src1, const uint8_t *src2,
1846 int width, uint32_t *unused)
1848 RENAME(bgr24ToUV_mmx)(dstU, dstV, src1, width, PIX_FMT_BGR24);
1849 assert(src1 == src2);
1852 static void RENAME(rgb24ToUV)(uint8_t *dstU, uint8_t *dstV,
1853 const uint8_t *src1, const uint8_t *src2,
1854 int width, uint32_t *unused)
1857 RENAME(bgr24ToUV_mmx)(dstU, dstV, src1, width, PIX_FMT_RGB24);
1860 #if !COMPILE_TEMPLATE_MMX2
1861 // bilinear / bicubic scaling
1862 static void RENAME(hScale)(SwsContext *c, int16_t *dst, int dstW,
1863 const uint8_t *src, const int16_t *filter,
1864 const int16_t *filterPos, int filterSize)
1866 assert(filterSize % 4 == 0 && filterSize>0);
1867 if (filterSize==4) { // Always true for upscaling, sometimes for down, too.
1868 x86_reg counter= -2*dstW;
1870 filterPos-= counter/2;
1874 "push %%"REG_b" \n\t"
1876 "pxor %%mm7, %%mm7 \n\t"
1877 "push %%"REG_BP" \n\t" // we use 7 regs here ...
1878 "mov %%"REG_a", %%"REG_BP" \n\t"
1881 "movzwl (%2, %%"REG_BP"), %%eax \n\t"
1882 "movzwl 2(%2, %%"REG_BP"), %%ebx \n\t"
1883 "movq (%1, %%"REG_BP", 4), %%mm1 \n\t"
1884 "movq 8(%1, %%"REG_BP", 4), %%mm3 \n\t"
1885 "movd (%3, %%"REG_a"), %%mm0 \n\t"
1886 "movd (%3, %%"REG_b"), %%mm2 \n\t"
1887 "punpcklbw %%mm7, %%mm0 \n\t"
1888 "punpcklbw %%mm7, %%mm2 \n\t"
1889 "pmaddwd %%mm1, %%mm0 \n\t"
1890 "pmaddwd %%mm2, %%mm3 \n\t"
1891 "movq %%mm0, %%mm4 \n\t"
1892 "punpckldq %%mm3, %%mm0 \n\t"
1893 "punpckhdq %%mm3, %%mm4 \n\t"
1894 "paddd %%mm4, %%mm0 \n\t"
1895 "psrad $7, %%mm0 \n\t"
1896 "packssdw %%mm0, %%mm0 \n\t"
1897 "movd %%mm0, (%4, %%"REG_BP") \n\t"
1898 "add $4, %%"REG_BP" \n\t"
1901 "pop %%"REG_BP" \n\t"
1903 "pop %%"REG_b" \n\t"
1906 : "c" (filter), "d" (filterPos), "S" (src), "D" (dst)
1911 } else if (filterSize==8) {
1912 x86_reg counter= -2*dstW;
1914 filterPos-= counter/2;
1918 "push %%"REG_b" \n\t"
1920 "pxor %%mm7, %%mm7 \n\t"
1921 "push %%"REG_BP" \n\t" // we use 7 regs here ...
1922 "mov %%"REG_a", %%"REG_BP" \n\t"
1925 "movzwl (%2, %%"REG_BP"), %%eax \n\t"
1926 "movzwl 2(%2, %%"REG_BP"), %%ebx \n\t"
1927 "movq (%1, %%"REG_BP", 8), %%mm1 \n\t"
1928 "movq 16(%1, %%"REG_BP", 8), %%mm3 \n\t"
1929 "movd (%3, %%"REG_a"), %%mm0 \n\t"
1930 "movd (%3, %%"REG_b"), %%mm2 \n\t"
1931 "punpcklbw %%mm7, %%mm0 \n\t"
1932 "punpcklbw %%mm7, %%mm2 \n\t"
1933 "pmaddwd %%mm1, %%mm0 \n\t"
1934 "pmaddwd %%mm2, %%mm3 \n\t"
1936 "movq 8(%1, %%"REG_BP", 8), %%mm1 \n\t"
1937 "movq 24(%1, %%"REG_BP", 8), %%mm5 \n\t"
1938 "movd 4(%3, %%"REG_a"), %%mm4 \n\t"
1939 "movd 4(%3, %%"REG_b"), %%mm2 \n\t"
1940 "punpcklbw %%mm7, %%mm4 \n\t"
1941 "punpcklbw %%mm7, %%mm2 \n\t"
1942 "pmaddwd %%mm1, %%mm4 \n\t"
1943 "pmaddwd %%mm2, %%mm5 \n\t"
1944 "paddd %%mm4, %%mm0 \n\t"
1945 "paddd %%mm5, %%mm3 \n\t"
1946 "movq %%mm0, %%mm4 \n\t"
1947 "punpckldq %%mm3, %%mm0 \n\t"
1948 "punpckhdq %%mm3, %%mm4 \n\t"
1949 "paddd %%mm4, %%mm0 \n\t"
1950 "psrad $7, %%mm0 \n\t"
1951 "packssdw %%mm0, %%mm0 \n\t"
1952 "movd %%mm0, (%4, %%"REG_BP") \n\t"
1953 "add $4, %%"REG_BP" \n\t"
1956 "pop %%"REG_BP" \n\t"
1958 "pop %%"REG_b" \n\t"
1961 : "c" (filter), "d" (filterPos), "S" (src), "D" (dst)
1967 const uint8_t *offset = src+filterSize;
1968 x86_reg counter= -2*dstW;
1969 //filter-= counter*filterSize/2;
1970 filterPos-= counter/2;
1973 "pxor %%mm7, %%mm7 \n\t"
1976 "mov %2, %%"REG_c" \n\t"
1977 "movzwl (%%"REG_c", %0), %%eax \n\t"
1978 "movzwl 2(%%"REG_c", %0), %%edx \n\t"
1979 "mov %5, %%"REG_c" \n\t"
1980 "pxor %%mm4, %%mm4 \n\t"
1981 "pxor %%mm5, %%mm5 \n\t"
1983 "movq (%1), %%mm1 \n\t"
1984 "movq (%1, %6), %%mm3 \n\t"
1985 "movd (%%"REG_c", %%"REG_a"), %%mm0 \n\t"
1986 "movd (%%"REG_c", %%"REG_d"), %%mm2 \n\t"
1987 "punpcklbw %%mm7, %%mm0 \n\t"
1988 "punpcklbw %%mm7, %%mm2 \n\t"
1989 "pmaddwd %%mm1, %%mm0 \n\t"
1990 "pmaddwd %%mm2, %%mm3 \n\t"
1991 "paddd %%mm3, %%mm5 \n\t"
1992 "paddd %%mm0, %%mm4 \n\t"
1994 "add $4, %%"REG_c" \n\t"
1995 "cmp %4, %%"REG_c" \n\t"
1998 "movq %%mm4, %%mm0 \n\t"
1999 "punpckldq %%mm5, %%mm4 \n\t"
2000 "punpckhdq %%mm5, %%mm0 \n\t"
2001 "paddd %%mm0, %%mm4 \n\t"
2002 "psrad $7, %%mm4 \n\t"
2003 "packssdw %%mm4, %%mm4 \n\t"
2004 "mov %3, %%"REG_a" \n\t"
2005 "movd %%mm4, (%%"REG_a", %0) \n\t"
2009 : "+r" (counter), "+r" (filter)
2010 : "m" (filterPos), "m" (dst), "m"(offset),
2011 "m" (src), "r" ((x86_reg)filterSize*2)
2012 : "%"REG_a, "%"REG_c, "%"REG_d
2016 #endif /* !COMPILE_TEMPLATE_MMX2 */
2018 #if COMPILE_TEMPLATE_MMX2
2019 static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
2020 int dstWidth, const uint8_t *src,
2023 int16_t *filterPos = c->hLumFilterPos;
2024 int16_t *filter = c->hLumFilter;
2025 void *mmx2FilterCode= c->lumMmx2FilterCode;
2028 DECLARE_ALIGNED(8, uint64_t, ebxsave);
2033 "mov %%"REG_b", %5 \n\t"
2035 "pxor %%mm7, %%mm7 \n\t"
2036 "mov %0, %%"REG_c" \n\t"
2037 "mov %1, %%"REG_D" \n\t"
2038 "mov %2, %%"REG_d" \n\t"
2039 "mov %3, %%"REG_b" \n\t"
2040 "xor %%"REG_a", %%"REG_a" \n\t" // i
2041 PREFETCH" (%%"REG_c") \n\t"
2042 PREFETCH" 32(%%"REG_c") \n\t"
2043 PREFETCH" 64(%%"REG_c") \n\t"
2046 #define CALL_MMX2_FILTER_CODE \
2047 "movl (%%"REG_b"), %%esi \n\t"\
2049 "movl (%%"REG_b", %%"REG_a"), %%esi \n\t"\
2050 "add %%"REG_S", %%"REG_c" \n\t"\
2051 "add %%"REG_a", %%"REG_D" \n\t"\
2052 "xor %%"REG_a", %%"REG_a" \n\t"\
2055 #define CALL_MMX2_FILTER_CODE \
2056 "movl (%%"REG_b"), %%esi \n\t"\
2058 "addl (%%"REG_b", %%"REG_a"), %%"REG_c" \n\t"\
2059 "add %%"REG_a", %%"REG_D" \n\t"\
2060 "xor %%"REG_a", %%"REG_a" \n\t"\
2062 #endif /* ARCH_X86_64 */
2064 CALL_MMX2_FILTER_CODE
2065 CALL_MMX2_FILTER_CODE
2066 CALL_MMX2_FILTER_CODE
2067 CALL_MMX2_FILTER_CODE
2068 CALL_MMX2_FILTER_CODE
2069 CALL_MMX2_FILTER_CODE
2070 CALL_MMX2_FILTER_CODE
2071 CALL_MMX2_FILTER_CODE
2074 "mov %5, %%"REG_b" \n\t"
2076 :: "m" (src), "m" (dst), "m" (filter), "m" (filterPos),
2077 "m" (mmx2FilterCode)
2081 : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
2087 for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--)
2088 dst[i] = src[srcW-1]*128;
2091 static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2,
2092 int dstWidth, const uint8_t *src1,
2093 const uint8_t *src2, int srcW, int xInc)
2095 int16_t *filterPos = c->hChrFilterPos;
2096 int16_t *filter = c->hChrFilter;
2097 void *mmx2FilterCode= c->chrMmx2FilterCode;
2100 DECLARE_ALIGNED(8, uint64_t, ebxsave);
2105 "mov %%"REG_b", %7 \n\t"
2107 "pxor %%mm7, %%mm7 \n\t"
2108 "mov %0, %%"REG_c" \n\t"
2109 "mov %1, %%"REG_D" \n\t"
2110 "mov %2, %%"REG_d" \n\t"
2111 "mov %3, %%"REG_b" \n\t"
2112 "xor %%"REG_a", %%"REG_a" \n\t" // i
2113 PREFETCH" (%%"REG_c") \n\t"
2114 PREFETCH" 32(%%"REG_c") \n\t"
2115 PREFETCH" 64(%%"REG_c") \n\t"
2117 CALL_MMX2_FILTER_CODE
2118 CALL_MMX2_FILTER_CODE
2119 CALL_MMX2_FILTER_CODE
2120 CALL_MMX2_FILTER_CODE
2121 "xor %%"REG_a", %%"REG_a" \n\t" // i
2122 "mov %5, %%"REG_c" \n\t" // src
2123 "mov %6, %%"REG_D" \n\t" // buf2
2124 PREFETCH" (%%"REG_c") \n\t"
2125 PREFETCH" 32(%%"REG_c") \n\t"
2126 PREFETCH" 64(%%"REG_c") \n\t"
2128 CALL_MMX2_FILTER_CODE
2129 CALL_MMX2_FILTER_CODE
2130 CALL_MMX2_FILTER_CODE
2131 CALL_MMX2_FILTER_CODE
2134 "mov %7, %%"REG_b" \n\t"
2136 :: "m" (src1), "m" (dst1), "m" (filter), "m" (filterPos),
2137 "m" (mmx2FilterCode), "m" (src2), "m"(dst2)
2141 : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
2147 for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) {
2148 dst1[i] = src1[srcW-1]*128;
2149 dst2[i] = src2[srcW-1]*128;
2152 #endif /* COMPILE_TEMPLATE_MMX2 */
2154 static av_cold void RENAME(sws_init_swScale)(SwsContext *c)
2156 enum PixelFormat srcFormat = c->srcFormat,
2157 dstFormat = c->dstFormat;
2159 if (!is16BPS(dstFormat) && !is9_OR_10BPS(dstFormat) &&
2160 dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21) {
2161 if (!(c->flags & SWS_BITEXACT)) {
2162 if (c->flags & SWS_ACCURATE_RND) {
2163 c->yuv2yuv1 = RENAME(yuv2yuv1_ar );
2164 c->yuv2yuvX = RENAME(yuv2yuvX_ar );
2165 if (!(c->flags & SWS_FULL_CHR_H_INT)) {
2166 switch (c->dstFormat) {
2167 case PIX_FMT_RGB32: c->yuv2packedX = RENAME(yuv2rgb32_X_ar); break;
2168 case PIX_FMT_BGR24: c->yuv2packedX = RENAME(yuv2bgr24_X_ar); break;
2169 case PIX_FMT_RGB555: c->yuv2packedX = RENAME(yuv2rgb555_X_ar); break;
2170 case PIX_FMT_RGB565: c->yuv2packedX = RENAME(yuv2rgb565_X_ar); break;
2171 case PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X_ar); break;
2176 c->yuv2yuv1 = RENAME(yuv2yuv1 );
2177 c->yuv2yuvX = RENAME(yuv2yuvX );
2178 if (!(c->flags & SWS_FULL_CHR_H_INT)) {
2179 switch (c->dstFormat) {
2180 case PIX_FMT_RGB32: c->yuv2packedX = RENAME(yuv2rgb32_X); break;
2181 case PIX_FMT_BGR24: c->yuv2packedX = RENAME(yuv2bgr24_X); break;
2182 case PIX_FMT_RGB555: c->yuv2packedX = RENAME(yuv2rgb555_X); break;
2183 case PIX_FMT_RGB565: c->yuv2packedX = RENAME(yuv2rgb565_X); break;
2184 case PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X); break;
2190 if (!(c->flags & SWS_FULL_CHR_H_INT)) {
2191 switch (c->dstFormat) {
2193 c->yuv2packed1 = RENAME(yuv2rgb32_1);
2194 c->yuv2packed2 = RENAME(yuv2rgb32_2);
2197 c->yuv2packed1 = RENAME(yuv2bgr24_1);
2198 c->yuv2packed2 = RENAME(yuv2bgr24_2);
2200 case PIX_FMT_RGB555:
2201 c->yuv2packed1 = RENAME(yuv2rgb555_1);
2202 c->yuv2packed2 = RENAME(yuv2rgb555_2);
2204 case PIX_FMT_RGB565:
2205 c->yuv2packed1 = RENAME(yuv2rgb565_1);
2206 c->yuv2packed2 = RENAME(yuv2rgb565_2);
2208 case PIX_FMT_YUYV422:
2209 c->yuv2packed1 = RENAME(yuv2yuyv422_1);
2210 c->yuv2packed2 = RENAME(yuv2yuyv422_2);
2218 if (c->scalingBpp == 8) {
2219 #if !COMPILE_TEMPLATE_MMX2
2220 c->hScale = RENAME(hScale );
2221 #endif /* !COMPILE_TEMPLATE_MMX2 */
2223 // Use the new MMX scaler if the MMX2 one can't be used (it is faster than the x86 ASM one).
2224 #if COMPILE_TEMPLATE_MMX2
2225 if (c->flags & SWS_FAST_BILINEAR && c->canMMX2BeUsed)
2227 c->hyscale_fast = RENAME(hyscale_fast);
2228 c->hcscale_fast = RENAME(hcscale_fast);
2230 #endif /* COMPILE_TEMPLATE_MMX2 */
2231 c->hyscale_fast = NULL;
2232 c->hcscale_fast = NULL;
2233 #if COMPILE_TEMPLATE_MMX2
2235 #endif /* COMPILE_TEMPLATE_MMX2 */
2238 #if !COMPILE_TEMPLATE_MMX2
2240 case PIX_FMT_YUYV422 : c->chrToYV12 = RENAME(yuy2ToUV); break;
2241 case PIX_FMT_UYVY422 : c->chrToYV12 = RENAME(uyvyToUV); break;
2242 case PIX_FMT_NV12 : c->chrToYV12 = RENAME(nv12ToUV); break;
2243 case PIX_FMT_NV21 : c->chrToYV12 = RENAME(nv21ToUV); break;
2246 #endif /* !COMPILE_TEMPLATE_MMX2 */
2247 if (!c->chrSrcHSubSample) {
2249 case PIX_FMT_BGR24 : c->chrToYV12 = RENAME(bgr24ToUV); break;
2250 case PIX_FMT_RGB24 : c->chrToYV12 = RENAME(rgb24ToUV); break;
2255 switch (srcFormat) {
2256 #if !COMPILE_TEMPLATE_MMX2
2257 case PIX_FMT_YUYV422 :
2258 case PIX_FMT_Y400A : c->lumToYV12 = RENAME(yuy2ToY); break;
2259 case PIX_FMT_UYVY422 : c->lumToYV12 = RENAME(uyvyToY); break;
2260 #endif /* !COMPILE_TEMPLATE_MMX2 */
2261 case PIX_FMT_BGR24 : c->lumToYV12 = RENAME(bgr24ToY); break;
2262 case PIX_FMT_RGB24 : c->lumToYV12 = RENAME(rgb24ToY); break;
2265 #if !COMPILE_TEMPLATE_MMX2
2267 switch (srcFormat) {
2268 case PIX_FMT_Y400A : c->alpToYV12 = RENAME(yuy2ToY); break;
2272 #endif /* !COMPILE_TEMPLATE_MMX2 */