2 * software YUV to RGB converter
4 * Copyright (C) 2001-2007 Michael Niedermayer
5 * (c) 2010 Konstantin Shishkov
7 * This file is part of FFmpeg.
9 * FFmpeg is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public
11 * License as published by the Free Software Foundation; either
12 * version 2.1 of the License, or (at your option) any later version.
14 * FFmpeg is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with FFmpeg; if not, write to the Free Software
21 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
26 #include "libavutil/x86/asm.h"
27 #include "libswscale/swscale_internal.h"
33 #if COMPILE_TEMPLATE_MMXEXT
34 #define MOVNTQ "movntq"
35 #define SFENCE "sfence"
38 #define SFENCE " # nop"
46 #define YUV2RGB_LOOP(depth) \
47 h_size = (c->dstW + 7) & ~7; \
48 if (h_size * depth > FFABS(dstStride[0])) \
51 vshift = c->srcFormat != AV_PIX_FMT_YUV422P; \
53 __asm__ volatile ("pxor %mm4, %mm4\n\t"); \
54 for (y = 0; y < srcSliceH; y++) { \
55 uint8_t *image = dst[0] + (y + srcSliceY) * dstStride[0]; \
56 const uint8_t *py = src[0] + y * srcStride[0]; \
57 const uint8_t *pu = src[1] + (y >> vshift) * srcStride[1]; \
58 const uint8_t *pv = src[2] + (y >> vshift) * srcStride[2]; \
59 x86_reg index = -h_size / 2; \
61 #define YUV2RGB_INITIAL_LOAD \
63 "movq (%5, %0, 2), %%mm6\n\t" \
64 "movd (%2, %0), %%mm0\n\t" \
65 "movd (%3, %0), %%mm1\n\t" \
69 * Conversion is performed in usual way:
70 * R = Y' * Ycoef + Vred * V'
71 * G = Y' * Ycoef + Vgreen * V' + Ugreen * U'
72 * B = Y' * Ycoef + Ublue * U'
74 * where X' = X * 8 - Xoffset (multiplication is performed to increase
76 * Since it operates in YUV420 colorspace, Y component is additionally
77 * split into Y1 and Y2 for even and odd pixels.
80 * mm0 - U (4 elems), mm1 - V (4 elems), mm6 - Y (8 elems), mm4 - zero register
82 * mm1 - R, mm2 - G, mm0 - B
85 /* convert Y, U, V into Y1', Y2', U', V' */ \
86 "movq %%mm6, %%mm7\n\t" \
87 "punpcklbw %%mm4, %%mm0\n\t" \
88 "punpcklbw %%mm4, %%mm1\n\t" \
89 "pand "MANGLE(mmx_00ffw)", %%mm6\n\t" \
90 "psrlw $8, %%mm7\n\t" \
91 "psllw $3, %%mm0\n\t" \
92 "psllw $3, %%mm1\n\t" \
93 "psllw $3, %%mm6\n\t" \
94 "psllw $3, %%mm7\n\t" \
95 "psubsw "U_OFFSET"(%4), %%mm0\n\t" \
96 "psubsw "V_OFFSET"(%4), %%mm1\n\t" \
97 "psubw "Y_OFFSET"(%4), %%mm6\n\t" \
98 "psubw "Y_OFFSET"(%4), %%mm7\n\t" \
100 /* multiply by coefficients */ \
101 "movq %%mm0, %%mm2\n\t" \
102 "movq %%mm1, %%mm3\n\t" \
103 "pmulhw "UG_COEFF"(%4), %%mm2\n\t" \
104 "pmulhw "VG_COEFF"(%4), %%mm3\n\t" \
105 "pmulhw "Y_COEFF" (%4), %%mm6\n\t" \
106 "pmulhw "Y_COEFF" (%4), %%mm7\n\t" \
107 "pmulhw "UB_COEFF"(%4), %%mm0\n\t" \
108 "pmulhw "VR_COEFF"(%4), %%mm1\n\t" \
109 "paddsw %%mm3, %%mm2\n\t" \
110 /* now: mm0 = UB, mm1 = VR, mm2 = CG */ \
111 /* mm6 = Y1, mm7 = Y2 */ \
114 "movq %%mm7, %%mm3\n\t" \
115 "movq %%mm7, %%mm5\n\t" \
116 "paddsw %%mm0, %%mm3\n\t" \
117 "paddsw %%mm1, %%mm5\n\t" \
118 "paddsw %%mm2, %%mm7\n\t" \
119 "paddsw %%mm6, %%mm0\n\t" \
120 "paddsw %%mm6, %%mm1\n\t" \
121 "paddsw %%mm6, %%mm2\n\t" \
123 #define RGB_PACK_INTERLEAVE \
124 /* pack and interleave even/odd pixels */ \
125 "packuswb %%mm1, %%mm0\n\t" \
126 "packuswb %%mm5, %%mm3\n\t" \
127 "packuswb %%mm2, %%mm2\n\t" \
128 "movq %%mm0, %%mm1\n\n" \
129 "packuswb %%mm7, %%mm7\n\t" \
130 "punpcklbw %%mm3, %%mm0\n\t" \
131 "punpckhbw %%mm3, %%mm1\n\t" \
132 "punpcklbw %%mm7, %%mm2\n\t" \
134 #define YUV2RGB_ENDLOOP(depth) \
135 "movq 8 (%5, %0, 2), %%mm6\n\t" \
136 "movd 4 (%3, %0), %%mm1\n\t" \
137 "movd 4 (%2, %0), %%mm0\n\t" \
138 "add $"AV_STRINGIFY(depth * 8)", %1\n\t" \
142 #if COMPILE_TEMPLATE_MMXEXT
143 #undef RGB_PACK24_B_OPERANDS
144 #define RGB_PACK24_B_OPERANDS NAMED_CONSTRAINTS_ARRAY_ADD(mask1101,mask0110,mask0100,mask0010,mask1001)
146 #undef RGB_PACK24_B_OPERANDS
147 #define RGB_PACK24_B_OPERANDS
150 #define YUV2RGB_OPERANDS \
151 : "+r" (index), "+r" (image) \
152 : "r" (pu - index), "r" (pv - index), "r"(&c->redDither), \
154 NAMED_CONSTRAINTS_ADD(mmx_00ffw,pb_03,pb_07,mmx_redmask,pb_e0) \
155 RGB_PACK24_B_OPERANDS \
160 #define YUV2RGB_OPERANDS_ALPHA \
161 : "+r" (index), "+r" (image) \
162 : "r" (pu - index), "r" (pv - index), "r"(&c->redDither), \
163 "r" (py - 2*index), "r" (pa - 2*index) \
164 NAMED_CONSTRAINTS_ADD(mmx_00ffw) \
169 #define YUV2RGB_ENDFUNC \
170 __asm__ volatile (SFENCE"\n\t" \
177 #define RGB_PACK16(gmask, is15) \
178 "pand "MANGLE(mmx_redmask)", %%mm0\n\t" \
179 "pand "MANGLE(mmx_redmask)", %%mm1\n\t" \
180 "movq %%mm2, %%mm3\n\t" \
181 "psllw $"AV_STRINGIFY(3-is15)", %%mm2\n\t" \
182 "psrlw $"AV_STRINGIFY(5+is15)", %%mm3\n\t" \
183 "psrlw $3, %%mm0\n\t" \
184 IF##is15("psrlw $1, %%mm1\n\t") \
185 "pand "MANGLE(pb_e0)", %%mm2\n\t" \
186 "pand "MANGLE(gmask)", %%mm3\n\t" \
187 "por %%mm2, %%mm0\n\t" \
188 "por %%mm3, %%mm1\n\t" \
189 "movq %%mm0, %%mm2\n\t" \
190 "punpcklbw %%mm1, %%mm0\n\t" \
191 "punpckhbw %%mm1, %%mm2\n\t" \
192 MOVNTQ " %%mm0, (%1)\n\t" \
193 MOVNTQ " %%mm2, 8(%1)\n\t" \
196 "paddusb "BLUE_DITHER"(%4), %%mm0\n\t" \
197 "paddusb "GREEN_DITHER"(%4), %%mm2\n\t" \
198 "paddusb "RED_DITHER"(%4), %%mm1\n\t" \
200 #if !COMPILE_TEMPLATE_MMXEXT
201 static inline int RENAME(yuv420_rgb15)(SwsContext *c, const uint8_t *src[],
203 int srcSliceY, int srcSliceH,
204 uint8_t *dst[], int dstStride[])
206 int y, h_size, vshift;
211 c->blueDither = ff_dither8[y & 1];
212 c->greenDither = ff_dither8[y & 1];
213 c->redDither = ff_dither8[(y + 1) & 1];
229 static inline int RENAME(yuv420_rgb16)(SwsContext *c, const uint8_t *src[],
231 int srcSliceY, int srcSliceH,
232 uint8_t *dst[], int dstStride[])
234 int y, h_size, vshift;
239 c->blueDither = ff_dither8[y & 1];
240 c->greenDither = ff_dither4[y & 1];
241 c->redDither = ff_dither8[(y + 1) & 1];
256 #endif /* !COMPILE_TEMPLATE_MMXEXT */
258 #define RGB_PACK24(blue, red)\
259 "packuswb %%mm3, %%mm0 \n" /* R0 R2 R4 R6 R1 R3 R5 R7 */\
260 "packuswb %%mm5, %%mm1 \n" /* B0 B2 B4 B6 B1 B3 B5 B7 */\
261 "packuswb %%mm7, %%mm2 \n" /* G0 G2 G4 G6 G1 G3 G5 G7 */\
262 "movq %%mm"red", %%mm3 \n"\
263 "movq %%mm"blue", %%mm6 \n"\
264 "psrlq $32, %%mm"red" \n" /* R1 R3 R5 R7 */\
265 "punpcklbw %%mm2, %%mm3 \n" /* R0 G0 R2 G2 R4 G4 R6 G6 */\
266 "punpcklbw %%mm"red", %%mm6 \n" /* B0 R1 B2 R3 B4 R5 B6 R7 */\
267 "movq %%mm3, %%mm5 \n"\
268 "punpckhbw %%mm"blue", %%mm2 \n" /* G1 B1 G3 B3 G5 B5 G7 B7 */\
269 "punpcklwd %%mm6, %%mm3 \n" /* R0 G0 B0 R1 R2 G2 B2 R3 */\
270 "punpckhwd %%mm6, %%mm5 \n" /* R4 G4 B4 R5 R6 G6 B6 R7 */\
273 #if COMPILE_TEMPLATE_MMXEXT
274 DECLARE_ASM_CONST(8, int16_t, mask1101[4]) = {-1,-1, 0,-1};
275 DECLARE_ASM_CONST(8, int16_t, mask0010[4]) = { 0, 0,-1, 0};
276 DECLARE_ASM_CONST(8, int16_t, mask0110[4]) = { 0,-1,-1, 0};
277 DECLARE_ASM_CONST(8, int16_t, mask1001[4]) = {-1, 0, 0,-1};
278 DECLARE_ASM_CONST(8, int16_t, mask0100[4]) = { 0,-1, 0, 0};
280 #define RGB_PACK24_B\
281 "pshufw $0xc6, %%mm2, %%mm1 \n"\
282 "pshufw $0x84, %%mm3, %%mm6 \n"\
283 "pshufw $0x38, %%mm5, %%mm7 \n"\
284 "pand "MANGLE(mask1101)", %%mm6 \n" /* R0 G0 B0 R1 -- -- R2 G2 */\
285 "movq %%mm1, %%mm0 \n"\
286 "pand "MANGLE(mask0110)", %%mm7 \n" /* -- -- R6 G6 B6 R7 -- -- */\
287 "movq %%mm1, %%mm2 \n"\
288 "pand "MANGLE(mask0100)", %%mm1 \n" /* -- -- G3 B3 -- -- -- -- */\
289 "psrlq $48, %%mm3 \n" /* B2 R3 -- -- -- -- -- -- */\
290 "pand "MANGLE(mask0010)", %%mm0 \n" /* -- -- -- -- G1 B1 -- -- */\
291 "psllq $32, %%mm5 \n" /* -- -- -- -- R4 G4 B4 R5 */\
292 "pand "MANGLE(mask1001)", %%mm2 \n" /* G5 B5 -- -- -- -- G7 B7 */\
293 "por %%mm3, %%mm1 \n"\
294 "por %%mm6, %%mm0 \n"\
295 "por %%mm5, %%mm1 \n"\
296 "por %%mm7, %%mm2 \n"\
297 MOVNTQ" %%mm0, (%1) \n"\
298 MOVNTQ" %%mm1, 8(%1) \n"\
299 MOVNTQ" %%mm2, 16(%1) \n"\
303 #define RGB_PACK24_B\
304 "movd %%mm3, (%1) \n" /* R0 G0 B0 R1 */\
305 "movd %%mm2, 4(%1) \n" /* G1 B1 */\
306 "psrlq $32, %%mm3 \n"\
307 "psrlq $16, %%mm2 \n"\
308 "movd %%mm3, 6(%1) \n" /* R2 G2 B2 R3 */\
309 "movd %%mm2, 10(%1) \n" /* G3 B3 */\
310 "psrlq $16, %%mm2 \n"\
311 "movd %%mm5, 12(%1) \n" /* R4 G4 B4 R5 */\
312 "movd %%mm2, 16(%1) \n" /* G5 B5 */\
313 "psrlq $32, %%mm5 \n"\
314 "movd %%mm2, 20(%1) \n" /* -- -- G7 B7 */\
315 "movd %%mm5, 18(%1) \n" /* R6 G6 B6 R7 */\
319 static inline int RENAME(yuv420_rgb24)(SwsContext *c, const uint8_t *src[],
321 int srcSliceY, int srcSliceH,
322 uint8_t *dst[], int dstStride[])
324 int y, h_size, vshift;
330 RGB_PACK24(REG_BLUE, REG_RED)
337 static inline int RENAME(yuv420_bgr24)(SwsContext *c, const uint8_t *src[],
339 int srcSliceY, int srcSliceH,
340 uint8_t *dst[], int dstStride[])
342 int y, h_size, vshift;
348 RGB_PACK24(REG_RED, REG_BLUE)
356 #define SET_EMPTY_ALPHA \
357 "pcmpeqd %%mm"REG_ALPHA", %%mm"REG_ALPHA"\n\t" /* set alpha to 0xFF */ \
360 "movq (%6, %0, 2), %%mm"REG_ALPHA"\n\t" \
362 #define RGB_PACK32(red, green, blue, alpha) \
363 "movq %%mm"blue", %%mm5\n\t" \
364 "movq %%mm"red", %%mm6\n\t" \
365 "punpckhbw %%mm"green", %%mm5\n\t" \
366 "punpcklbw %%mm"green", %%mm"blue"\n\t" \
367 "punpckhbw %%mm"alpha", %%mm6\n\t" \
368 "punpcklbw %%mm"alpha", %%mm"red"\n\t" \
369 "movq %%mm"blue", %%mm"green"\n\t" \
370 "movq %%mm5, %%mm"alpha"\n\t" \
371 "punpcklwd %%mm"red", %%mm"blue"\n\t" \
372 "punpckhwd %%mm"red", %%mm"green"\n\t" \
373 "punpcklwd %%mm6, %%mm5\n\t" \
374 "punpckhwd %%mm6, %%mm"alpha"\n\t" \
375 MOVNTQ " %%mm"blue", 0(%1)\n\t" \
376 MOVNTQ " %%mm"green", 8(%1)\n\t" \
377 MOVNTQ " %%mm5, 16(%1)\n\t" \
378 MOVNTQ " %%mm"alpha", 24(%1)\n\t" \
380 #if !COMPILE_TEMPLATE_MMXEXT
381 static inline int RENAME(yuv420_rgb32)(SwsContext *c, const uint8_t *src[],
383 int srcSliceY, int srcSliceH,
384 uint8_t *dst[], int dstStride[])
386 int y, h_size, vshift;
394 RGB_PACK32(REG_RED, REG_GREEN, REG_BLUE, REG_ALPHA)
401 #if HAVE_7REGS && CONFIG_SWSCALE_ALPHA
402 static inline int RENAME(yuva420_rgb32)(SwsContext *c, const uint8_t *src[],
404 int srcSliceY, int srcSliceH,
405 uint8_t *dst[], int dstStride[])
407 int y, h_size, vshift;
411 const uint8_t *pa = src[3] + y * srcStride[3];
416 RGB_PACK32(REG_RED, REG_GREEN, REG_BLUE, REG_ALPHA)
419 YUV2RGB_OPERANDS_ALPHA
424 static inline int RENAME(yuv420_bgr32)(SwsContext *c, const uint8_t *src[],
426 int srcSliceY, int srcSliceH,
427 uint8_t *dst[], int dstStride[])
429 int y, h_size, vshift;
437 RGB_PACK32(REG_BLUE, REG_GREEN, REG_RED, REG_ALPHA)
444 #if HAVE_7REGS && CONFIG_SWSCALE_ALPHA
445 static inline int RENAME(yuva420_bgr32)(SwsContext *c, const uint8_t *src[],
447 int srcSliceY, int srcSliceH,
448 uint8_t *dst[], int dstStride[])
450 int y, h_size, vshift;
454 const uint8_t *pa = src[3] + y * srcStride[3];
459 RGB_PACK32(REG_BLUE, REG_GREEN, REG_RED, REG_ALPHA)
462 YUV2RGB_OPERANDS_ALPHA
467 #endif /* !COMPILE_TEMPLATE_MMXEXT */