2 * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg 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 * FFmpeg 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 FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 #include "libswscale/swscale.h"
24 #include "libswscale/swscale_internal.h"
25 #include "libavutil/attributes.h"
26 #include "libavutil/avassert.h"
27 #include "libavutil/intreadwrite.h"
28 #include "libavutil/x86/asm.h"
29 #include "libavutil/x86/cpu.h"
30 #include "libavutil/cpu.h"
31 #include "libavutil/pixdesc.h"
37 DECLARE_ASM_CONST(8, uint64_t, bF8)= 0xF8F8F8F8F8F8F8F8LL;
38 DECLARE_ASM_CONST(8, uint64_t, bFC)= 0xFCFCFCFCFCFCFCFCLL;
39 DECLARE_ASM_CONST(8, uint64_t, w10)= 0x0010001000100010LL;
40 DECLARE_ASM_CONST(8, uint64_t, w02)= 0x0002000200020002LL;
42 const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
44 0x0200020002000200LL,};
46 const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
48 0x0004000400040004LL,};
50 DECLARE_ASM_CONST(8, uint64_t, b16Mask)= 0x001F001F001F001FLL;
51 DECLARE_ASM_CONST(8, uint64_t, g16Mask)= 0x07E007E007E007E0LL;
52 DECLARE_ASM_CONST(8, uint64_t, r16Mask)= 0xF800F800F800F800LL;
53 DECLARE_ASM_CONST(8, uint64_t, b15Mask)= 0x001F001F001F001FLL;
54 DECLARE_ASM_CONST(8, uint64_t, g15Mask)= 0x03E003E003E003E0LL;
55 DECLARE_ASM_CONST(8, uint64_t, r15Mask)= 0x7C007C007C007C00LL;
57 DECLARE_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL;
58 DECLARE_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL;
59 DECLARE_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL;
62 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000000210041000DULL;
63 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000FFEEFFDC0038ULL;
64 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00000038FFD2FFF8ULL;
66 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000020E540830C8BULL;
67 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000ED0FDAC23831ULL;
68 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00003831D0E6F6EAULL;
69 #endif /* FAST_BGR2YV12 */
70 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL;
71 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
72 DECLARE_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL;
78 #define COMPILE_TEMPLATE_MMXEXT 0
79 #define RENAME(a) a ## _MMX
80 #include "swscale_template.c"
84 #if HAVE_MMXEXT_INLINE
86 #undef COMPILE_TEMPLATE_MMXEXT
87 #define COMPILE_TEMPLATE_MMXEXT 1
88 #define RENAME(a) a ## _MMXEXT
89 #include "swscale_template.c"
92 void updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
93 int lastInLumBuf, int lastInChrBuf)
95 const int dstH= c->dstH;
96 const int flags= c->flags;
97 int16_t **lumPixBuf= c->lumPixBuf;
98 int16_t **chrUPixBuf= c->chrUPixBuf;
99 int16_t **alpPixBuf= c->alpPixBuf;
100 const int vLumBufSize= c->vLumBufSize;
101 const int vChrBufSize= c->vChrBufSize;
102 int32_t *vLumFilterPos= c->vLumFilterPos;
103 int32_t *vChrFilterPos= c->vChrFilterPos;
104 int16_t *vLumFilter= c->vLumFilter;
105 int16_t *vChrFilter= c->vChrFilter;
106 int32_t *lumMmxFilter= c->lumMmxFilter;
107 int32_t *chrMmxFilter= c->chrMmxFilter;
108 int32_t av_unused *alpMmxFilter= c->alpMmxFilter;
109 const int vLumFilterSize= c->vLumFilterSize;
110 const int vChrFilterSize= c->vChrFilterSize;
111 const int chrDstY= dstY>>c->chrDstVSubSample;
112 const int firstLumSrcY= vLumFilterPos[dstY]; //First line needed as input
113 const int firstChrSrcY= vChrFilterPos[chrDstY]; //First line needed as input
115 c->blueDither= ff_dither8[dstY&1];
116 if (c->dstFormat == AV_PIX_FMT_RGB555 || c->dstFormat == AV_PIX_FMT_BGR555)
117 c->greenDither= ff_dither8[dstY&1];
119 c->greenDither= ff_dither4[dstY&1];
120 c->redDither= ff_dither8[(dstY+1)&1];
121 if (dstY < dstH - 2) {
122 const int16_t **lumSrcPtr= (const int16_t **)(void*) lumPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
123 const int16_t **chrUSrcPtr= (const int16_t **)(void*) chrUPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
124 const int16_t **alpSrcPtr= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? (const int16_t **)(void*) alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
127 if (firstLumSrcY < 0 || firstLumSrcY + vLumFilterSize > c->srcH) {
128 const int16_t **tmpY = (const int16_t **) lumPixBuf + 2 * vLumBufSize;
129 int neg = -firstLumSrcY, i, end = FFMIN(c->srcH - firstLumSrcY, vLumFilterSize);
130 for (i = 0; i < neg; i++)
131 tmpY[i] = lumSrcPtr[neg];
132 for ( ; i < end; i++)
133 tmpY[i] = lumSrcPtr[i];
134 for ( ; i < vLumFilterSize; i++)
139 const int16_t **tmpA = (const int16_t **) alpPixBuf + 2 * vLumBufSize;
140 for (i = 0; i < neg; i++)
141 tmpA[i] = alpSrcPtr[neg];
142 for ( ; i < end; i++)
143 tmpA[i] = alpSrcPtr[i];
144 for ( ; i < vLumFilterSize; i++)
145 tmpA[i] = tmpA[i - 1];
149 if (firstChrSrcY < 0 || firstChrSrcY + vChrFilterSize > c->chrSrcH) {
150 const int16_t **tmpU = (const int16_t **) chrUPixBuf + 2 * vChrBufSize;
151 int neg = -firstChrSrcY, i, end = FFMIN(c->chrSrcH - firstChrSrcY, vChrFilterSize);
152 for (i = 0; i < neg; i++) {
153 tmpU[i] = chrUSrcPtr[neg];
155 for ( ; i < end; i++) {
156 tmpU[i] = chrUSrcPtr[i];
158 for ( ; i < vChrFilterSize; i++) {
159 tmpU[i] = tmpU[i - 1];
164 if (flags & SWS_ACCURATE_RND) {
165 int s= APCK_SIZE / 8;
166 for (i=0; i<vLumFilterSize; i+=2) {
167 *(const void**)&lumMmxFilter[s*i ]= lumSrcPtr[i ];
168 *(const void**)&lumMmxFilter[s*i+APCK_PTR2/4 ]= lumSrcPtr[i+(vLumFilterSize>1)];
169 lumMmxFilter[s*i+APCK_COEF/4 ]=
170 lumMmxFilter[s*i+APCK_COEF/4+1]= vLumFilter[dstY*vLumFilterSize + i ]
171 + (vLumFilterSize>1 ? vLumFilter[dstY*vLumFilterSize + i + 1]<<16 : 0);
172 if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
173 *(const void**)&alpMmxFilter[s*i ]= alpSrcPtr[i ];
174 *(const void**)&alpMmxFilter[s*i+APCK_PTR2/4 ]= alpSrcPtr[i+(vLumFilterSize>1)];
175 alpMmxFilter[s*i+APCK_COEF/4 ]=
176 alpMmxFilter[s*i+APCK_COEF/4+1]= lumMmxFilter[s*i+APCK_COEF/4 ];
179 for (i=0; i<vChrFilterSize; i+=2) {
180 *(const void**)&chrMmxFilter[s*i ]= chrUSrcPtr[i ];
181 *(const void**)&chrMmxFilter[s*i+APCK_PTR2/4 ]= chrUSrcPtr[i+(vChrFilterSize>1)];
182 chrMmxFilter[s*i+APCK_COEF/4 ]=
183 chrMmxFilter[s*i+APCK_COEF/4+1]= vChrFilter[chrDstY*vChrFilterSize + i ]
184 + (vChrFilterSize>1 ? vChrFilter[chrDstY*vChrFilterSize + i + 1]<<16 : 0);
187 for (i=0; i<vLumFilterSize; i++) {
188 *(const void**)&lumMmxFilter[4*i+0]= lumSrcPtr[i];
191 ((uint16_t)vLumFilter[dstY*vLumFilterSize + i])*0x10001U;
192 if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
193 *(const void**)&alpMmxFilter[4*i+0]= alpSrcPtr[i];
195 alpMmxFilter[4*i+3]= lumMmxFilter[4*i+2];
198 for (i=0; i<vChrFilterSize; i++) {
199 *(const void**)&chrMmxFilter[4*i+0]= chrUSrcPtr[i];
202 ((uint16_t)vChrFilter[chrDstY*vChrFilterSize + i])*0x10001U;
209 static void yuv2yuvX_sse3(const int16_t *filter, int filterSize,
210 const int16_t **src, uint8_t *dest, int dstW,
211 const uint8_t *dither, int offset)
213 if(((int)dest) & 15){
214 return yuv2yuvX_MMXEXT(filter, filterSize, src, dest, dstW, dither, offset);
217 __asm__ volatile("movq (%0), %%xmm3\n\t"
218 "movdqa %%xmm3, %%xmm4\n\t"
219 "psrlq $24, %%xmm3\n\t"
220 "psllq $40, %%xmm4\n\t"
221 "por %%xmm4, %%xmm3\n\t"
225 __asm__ volatile("movq (%0), %%xmm3\n\t"
230 "pxor %%xmm0, %%xmm0\n\t"
231 "punpcklbw %%xmm0, %%xmm3\n\t"
232 "psraw $4, %%xmm3\n\t"
233 "movdqa %%xmm3, %%xmm4\n\t"
234 "movdqa %%xmm3, %%xmm7\n\t"
236 "mov %0, %%"REG_d" \n\t"\
237 "mov (%%"REG_d"), %%"REG_S" \n\t"\
238 ".p2align 4 \n\t" /* FIXME Unroll? */\
240 "movddup 8(%%"REG_d"), %%xmm0 \n\t" /* filterCoeff */\
241 "movdqa (%%"REG_S", %%"REG_c", 2), %%xmm2 \n\t" /* srcData */\
242 "movdqa 16(%%"REG_S", %%"REG_c", 2), %%xmm5 \n\t" /* srcData */\
243 "add $16, %%"REG_d" \n\t"\
244 "mov (%%"REG_d"), %%"REG_S" \n\t"\
245 "test %%"REG_S", %%"REG_S" \n\t"\
246 "pmulhw %%xmm0, %%xmm2 \n\t"\
247 "pmulhw %%xmm0, %%xmm5 \n\t"\
248 "paddw %%xmm2, %%xmm3 \n\t"\
249 "paddw %%xmm5, %%xmm4 \n\t"\
251 "psraw $3, %%xmm3 \n\t"\
252 "psraw $3, %%xmm4 \n\t"\
253 "packuswb %%xmm4, %%xmm3 \n\t"
254 "movntdq %%xmm3, (%1, %%"REG_c")\n\t"
255 "add $16, %%"REG_c" \n\t"\
256 "cmp %2, %%"REG_c" \n\t"\
257 "movdqa %%xmm7, %%xmm3\n\t"
258 "movdqa %%xmm7, %%xmm4\n\t"
259 "mov %0, %%"REG_d" \n\t"\
260 "mov (%%"REG_d"), %%"REG_S" \n\t"\
263 "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset)
264 : "%"REG_d, "%"REG_S, "%"REG_c
269 #endif /* HAVE_INLINE_ASM */
271 #define SCALE_FUNC(filter_n, from_bpc, to_bpc, opt) \
272 extern void ff_hscale ## from_bpc ## to ## to_bpc ## _ ## filter_n ## _ ## opt( \
273 SwsContext *c, int16_t *data, \
274 int dstW, const uint8_t *src, \
275 const int16_t *filter, \
276 const int32_t *filterPos, int filterSize)
278 #define SCALE_FUNCS(filter_n, opt) \
279 SCALE_FUNC(filter_n, 8, 15, opt); \
280 SCALE_FUNC(filter_n, 9, 15, opt); \
281 SCALE_FUNC(filter_n, 10, 15, opt); \
282 SCALE_FUNC(filter_n, 12, 15, opt); \
283 SCALE_FUNC(filter_n, 14, 15, opt); \
284 SCALE_FUNC(filter_n, 16, 15, opt); \
285 SCALE_FUNC(filter_n, 8, 19, opt); \
286 SCALE_FUNC(filter_n, 9, 19, opt); \
287 SCALE_FUNC(filter_n, 10, 19, opt); \
288 SCALE_FUNC(filter_n, 12, 19, opt); \
289 SCALE_FUNC(filter_n, 14, 19, opt); \
290 SCALE_FUNC(filter_n, 16, 19, opt)
292 #define SCALE_FUNCS_MMX(opt) \
293 SCALE_FUNCS(4, opt); \
294 SCALE_FUNCS(8, opt); \
297 #define SCALE_FUNCS_SSE(opt) \
298 SCALE_FUNCS(4, opt); \
299 SCALE_FUNCS(8, opt); \
300 SCALE_FUNCS(X4, opt); \
304 SCALE_FUNCS_MMX(mmx);
306 SCALE_FUNCS_SSE(sse2);
307 SCALE_FUNCS_SSE(ssse3);
308 SCALE_FUNCS_SSE(sse4);
310 #define VSCALEX_FUNC(size, opt) \
311 extern void ff_yuv2planeX_ ## size ## _ ## opt(const int16_t *filter, int filterSize, \
312 const int16_t **src, uint8_t *dest, int dstW, \
313 const uint8_t *dither, int offset)
314 #define VSCALEX_FUNCS(opt) \
315 VSCALEX_FUNC(8, opt); \
316 VSCALEX_FUNC(9, opt); \
317 VSCALEX_FUNC(10, opt)
320 VSCALEX_FUNCS(mmxext);
324 VSCALEX_FUNC(16, sse4);
327 #define VSCALE_FUNC(size, opt) \
328 extern void ff_yuv2plane1_ ## size ## _ ## opt(const int16_t *src, uint8_t *dst, int dstW, \
329 const uint8_t *dither, int offset)
330 #define VSCALE_FUNCS(opt1, opt2) \
331 VSCALE_FUNC(8, opt1); \
332 VSCALE_FUNC(9, opt2); \
333 VSCALE_FUNC(10, opt2); \
334 VSCALE_FUNC(16, opt1)
337 VSCALE_FUNCS(mmx, mmxext);
339 VSCALE_FUNCS(sse2, sse2);
340 VSCALE_FUNC(16, sse4);
341 VSCALE_FUNCS(avx, avx);
343 #define INPUT_Y_FUNC(fmt, opt) \
344 extern void ff_ ## fmt ## ToY_ ## opt(uint8_t *dst, const uint8_t *src, \
345 const uint8_t *unused1, const uint8_t *unused2, \
346 int w, uint32_t *unused)
347 #define INPUT_UV_FUNC(fmt, opt) \
348 extern void ff_ ## fmt ## ToUV_ ## opt(uint8_t *dstU, uint8_t *dstV, \
349 const uint8_t *unused0, \
350 const uint8_t *src1, \
351 const uint8_t *src2, \
352 int w, uint32_t *unused)
353 #define INPUT_FUNC(fmt, opt) \
354 INPUT_Y_FUNC(fmt, opt); \
355 INPUT_UV_FUNC(fmt, opt)
356 #define INPUT_FUNCS(opt) \
357 INPUT_FUNC(uyvy, opt); \
358 INPUT_FUNC(yuyv, opt); \
359 INPUT_UV_FUNC(nv12, opt); \
360 INPUT_UV_FUNC(nv21, opt); \
361 INPUT_FUNC(rgba, opt); \
362 INPUT_FUNC(bgra, opt); \
363 INPUT_FUNC(argb, opt); \
364 INPUT_FUNC(abgr, opt); \
365 INPUT_FUNC(rgb24, opt); \
366 INPUT_FUNC(bgr24, opt)
375 av_cold void ff_sws_init_swScale_mmx(SwsContext *c)
377 int cpu_flags = av_get_cpu_flags();
380 if (cpu_flags & AV_CPU_FLAG_MMX)
381 sws_init_swScale_MMX(c);
382 #if HAVE_MMXEXT_INLINE
383 if (cpu_flags & AV_CPU_FLAG_MMXEXT)
384 sws_init_swScale_MMXEXT(c);
385 if (cpu_flags & AV_CPU_FLAG_SSE3){
386 if(c->use_mmx_vfilter && !(c->flags & SWS_ACCURATE_RND))
387 c->yuv2planeX = yuv2yuvX_sse3;
390 #endif /* HAVE_INLINE_ASM */
392 #define ASSIGN_SCALE_FUNC2(hscalefn, filtersize, opt1, opt2) do { \
393 if (c->srcBpc == 8) { \
394 hscalefn = c->dstBpc <= 14 ? ff_hscale8to15_ ## filtersize ## _ ## opt2 : \
395 ff_hscale8to19_ ## filtersize ## _ ## opt1; \
396 } else if (c->srcBpc == 9) { \
397 hscalefn = c->dstBpc <= 14 ? ff_hscale9to15_ ## filtersize ## _ ## opt2 : \
398 ff_hscale9to19_ ## filtersize ## _ ## opt1; \
399 } else if (c->srcBpc == 10) { \
400 hscalefn = c->dstBpc <= 14 ? ff_hscale10to15_ ## filtersize ## _ ## opt2 : \
401 ff_hscale10to19_ ## filtersize ## _ ## opt1; \
402 } else if (c->srcBpc == 12) { \
403 hscalefn = c->dstBpc <= 14 ? ff_hscale12to15_ ## filtersize ## _ ## opt2 : \
404 ff_hscale12to19_ ## filtersize ## _ ## opt1; \
405 } else if (c->srcBpc == 14 || ((c->srcFormat==AV_PIX_FMT_PAL8||isAnyRGB(c->srcFormat)) && av_pix_fmt_desc_get(c->srcFormat)->comp[0].depth_minus1<15)) { \
406 hscalefn = c->dstBpc <= 14 ? ff_hscale14to15_ ## filtersize ## _ ## opt2 : \
407 ff_hscale14to19_ ## filtersize ## _ ## opt1; \
408 } else { /* c->srcBpc == 16 */ \
409 av_assert0(c->srcBpc == 16);\
410 hscalefn = c->dstBpc <= 14 ? ff_hscale16to15_ ## filtersize ## _ ## opt2 : \
411 ff_hscale16to19_ ## filtersize ## _ ## opt1; \
414 #define ASSIGN_MMX_SCALE_FUNC(hscalefn, filtersize, opt1, opt2) \
415 switch (filtersize) { \
416 case 4: ASSIGN_SCALE_FUNC2(hscalefn, 4, opt1, opt2); break; \
417 case 8: ASSIGN_SCALE_FUNC2(hscalefn, 8, opt1, opt2); break; \
418 default: ASSIGN_SCALE_FUNC2(hscalefn, X, opt1, opt2); break; \
420 #define ASSIGN_VSCALEX_FUNC(vscalefn, opt, do_16_case, condition_8bit) \
422 case 16: do_16_case; break; \
423 case 10: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_10_ ## opt; break; \
424 case 9: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_9_ ## opt; break; \
425 default: if (condition_8bit) /*vscalefn = ff_yuv2planeX_8_ ## opt;*/ break; \
427 #define ASSIGN_VSCALE_FUNC(vscalefn, opt1, opt2, opt2chk) \
429 case 16: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2plane1_16_ ## opt1; break; \
430 case 10: if (!isBE(c->dstFormat) && opt2chk) vscalefn = ff_yuv2plane1_10_ ## opt2; break; \
431 case 9: if (!isBE(c->dstFormat) && opt2chk) vscalefn = ff_yuv2plane1_9_ ## opt2; break; \
432 case 8: vscalefn = ff_yuv2plane1_8_ ## opt1; break; \
433 default: av_assert0(c->dstBpc>8); \
435 #define case_rgb(x, X, opt) \
436 case AV_PIX_FMT_ ## X: \
437 c->lumToYV12 = ff_ ## x ## ToY_ ## opt; \
438 if (!c->chrSrcHSubSample) \
439 c->chrToYV12 = ff_ ## x ## ToUV_ ## opt; \
442 if (EXTERNAL_MMX(cpu_flags)) {
443 ASSIGN_MMX_SCALE_FUNC(c->hyScale, c->hLumFilterSize, mmx, mmx);
444 ASSIGN_MMX_SCALE_FUNC(c->hcScale, c->hChrFilterSize, mmx, mmx);
445 ASSIGN_VSCALE_FUNC(c->yuv2plane1, mmx, mmxext, cpu_flags & AV_CPU_FLAG_MMXEXT);
447 switch (c->srcFormat) {
448 case AV_PIX_FMT_Y400A:
449 c->lumToYV12 = ff_yuyvToY_mmx;
451 c->alpToYV12 = ff_uyvyToY_mmx;
453 case AV_PIX_FMT_YUYV422:
454 c->lumToYV12 = ff_yuyvToY_mmx;
455 c->chrToYV12 = ff_yuyvToUV_mmx;
457 case AV_PIX_FMT_UYVY422:
458 c->lumToYV12 = ff_uyvyToY_mmx;
459 c->chrToYV12 = ff_uyvyToUV_mmx;
461 case AV_PIX_FMT_NV12:
462 c->chrToYV12 = ff_nv12ToUV_mmx;
464 case AV_PIX_FMT_NV21:
465 c->chrToYV12 = ff_nv21ToUV_mmx;
467 case_rgb(rgb24, RGB24, mmx);
468 case_rgb(bgr24, BGR24, mmx);
469 case_rgb(bgra, BGRA, mmx);
470 case_rgb(rgba, RGBA, mmx);
471 case_rgb(abgr, ABGR, mmx);
472 case_rgb(argb, ARGB, mmx);
477 if (EXTERNAL_MMXEXT(cpu_flags)) {
478 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, mmxext, , 1);
480 #endif /* ARCH_X86_32 */
481 #define ASSIGN_SSE_SCALE_FUNC(hscalefn, filtersize, opt1, opt2) \
482 switch (filtersize) { \
483 case 4: ASSIGN_SCALE_FUNC2(hscalefn, 4, opt1, opt2); break; \
484 case 8: ASSIGN_SCALE_FUNC2(hscalefn, 8, opt1, opt2); break; \
485 default: if (filtersize & 4) ASSIGN_SCALE_FUNC2(hscalefn, X4, opt1, opt2); \
486 else ASSIGN_SCALE_FUNC2(hscalefn, X8, opt1, opt2); \
489 if (EXTERNAL_SSE2(cpu_flags)) {
490 ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse2, sse2);
491 ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse2, sse2);
492 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse2, ,
493 HAVE_ALIGNED_STACK || ARCH_X86_64);
494 ASSIGN_VSCALE_FUNC(c->yuv2plane1, sse2, sse2, 1);
496 switch (c->srcFormat) {
497 case AV_PIX_FMT_Y400A:
498 c->lumToYV12 = ff_yuyvToY_sse2;
500 c->alpToYV12 = ff_uyvyToY_sse2;
502 case AV_PIX_FMT_YUYV422:
503 c->lumToYV12 = ff_yuyvToY_sse2;
504 c->chrToYV12 = ff_yuyvToUV_sse2;
506 case AV_PIX_FMT_UYVY422:
507 c->lumToYV12 = ff_uyvyToY_sse2;
508 c->chrToYV12 = ff_uyvyToUV_sse2;
510 case AV_PIX_FMT_NV12:
511 c->chrToYV12 = ff_nv12ToUV_sse2;
513 case AV_PIX_FMT_NV21:
514 c->chrToYV12 = ff_nv21ToUV_sse2;
516 case_rgb(rgb24, RGB24, sse2);
517 case_rgb(bgr24, BGR24, sse2);
518 case_rgb(bgra, BGRA, sse2);
519 case_rgb(rgba, RGBA, sse2);
520 case_rgb(abgr, ABGR, sse2);
521 case_rgb(argb, ARGB, sse2);
526 if (EXTERNAL_SSSE3(cpu_flags)) {
527 ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, ssse3, ssse3);
528 ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, ssse3, ssse3);
529 switch (c->srcFormat) {
530 case_rgb(rgb24, RGB24, ssse3);
531 case_rgb(bgr24, BGR24, ssse3);
536 if (EXTERNAL_SSE4(cpu_flags)) {
537 /* Xto15 don't need special sse4 functions */
538 ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse4, ssse3);
539 ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse4, ssse3);
540 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse4,
541 if (!isBE(c->dstFormat)) c->yuv2planeX = ff_yuv2planeX_16_sse4,
542 HAVE_ALIGNED_STACK || ARCH_X86_64);
543 if (c->dstBpc == 16 && !isBE(c->dstFormat))
544 c->yuv2plane1 = ff_yuv2plane1_16_sse4;
547 if (EXTERNAL_AVX(cpu_flags)) {
548 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, avx, ,
549 HAVE_ALIGNED_STACK || ARCH_X86_64);
550 ASSIGN_VSCALE_FUNC(c->yuv2plane1, avx, avx, 1);
552 switch (c->srcFormat) {
553 case AV_PIX_FMT_YUYV422:
554 c->chrToYV12 = ff_yuyvToUV_avx;
556 case AV_PIX_FMT_UYVY422:
557 c->chrToYV12 = ff_uyvyToUV_avx;
559 case AV_PIX_FMT_NV12:
560 c->chrToYV12 = ff_nv12ToUV_avx;
562 case AV_PIX_FMT_NV21:
563 c->chrToYV12 = ff_nv21ToUV_avx;
565 case_rgb(rgb24, RGB24, avx);
566 case_rgb(bgr24, BGR24, avx);
567 case_rgb(bgra, BGRA, avx);
568 case_rgb(rgba, RGBA, avx);
569 case_rgb(abgr, ABGR, avx);
570 case_rgb(argb, ARGB, avx);