falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
3330 Commits
32 Branches
283MB
Tree: d890d4aed1
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'd890d4aed1'
${ noResults }
FFmpeg/postproc/yuv2rgb_template.c

541 lines
20KB
Raw Blame History 

  1. 

  2. /*

  3.  * yuv2rgb_mmx.c, Software YUV to RGB coverter with Intel MMX "technology"

  4.  *

  5.  * Copyright (C) 2000, Silicon Integrated System Corp.

  6.  * All Rights Reserved.

  7.  *

  8.  * Author: Olie Lho <ollie@sis.com.tw>

  9.  *

  10.  * This file is part of mpeg2dec, a free MPEG-2 video decoder

  11.  *

  12.  * mpeg2dec is free software; you can redistribute it and/or modify

  13.  * it under the terms of the GNU General Public License as published by

  14.  * the Free Software Foundation; either version 2, or (at your option)

  15.  * any later version.

  16.  *

  17.  * mpeg2dec is distributed in the hope that it will be useful,

  18.  * but WITHOUT ANY WARRANTY; without even the implied warranty of

  19.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

  20.  * GNU General Public License for more details.

  21.  *

  22.  * You should have received a copy of the GNU General Public License

  23.  * along with GNU Make; see the file COPYING. If not, write to

  24.  * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.

  25.  *

  26.  * 15,24 bpp and dithering from Michael Niedermayer (michaelni@gmx.at)

  27.  * MMX/MMX2 Template stuff from Michael Niedermayer (needed for fast movntq support)

  28.  * context / deglobalize stuff by Michael Niedermayer

  29.  */

  30. 

  31. #undef MOVNTQ

  32. #undef EMMS

  33. #undef SFENCE

  34. 

  35. #ifdef HAVE_3DNOW

  36. /* On K6 femms is faster of emms. On K7 femms is directly mapped on emms. */

  37. #define EMMS     "femms"

  38. #else

  39. #define EMMS     "emms"

  40. #endif

  41. 

  42. #ifdef HAVE_MMX2

  43. #define MOVNTQ "movntq"

  44. #define SFENCE "sfence"

  45. #else

  46. #define MOVNTQ "movq"

  47. #define SFENCE "/nop"

  48. #endif

  49. 

  50. #define YUV2RGB \

  51. 		     /* Do the multiply part of the conversion for even and odd pixels,

  52. 			register usage:

  53. 			mm0 -> Cblue, mm1 -> Cred, mm2 -> Cgreen even pixels,

  54. 			mm3 -> Cblue, mm4 -> Cred, mm5 -> Cgreen odd pixels,

  55. 			mm6 -> Y even, mm7 -> Y odd */\

  56. 		     /* convert the chroma part */\

  57. 		     "punpcklbw %%mm4, %%mm0;" /* scatter 4 Cb 00 u3 00 u2 00 u1 00 u0 */ \

  58. 		     "punpcklbw %%mm4, %%mm1;" /* scatter 4 Cr 00 v3 00 v2 00 v1 00 v0 */ \

  59. \

  60. 		     "psllw $3, %%mm0;" /* Promote precision */ \

  61. 		     "psllw $3, %%mm1;" /* Promote precision */ \

  62. \

  63. 		     "psubsw "U_OFFSET"(%4), %%mm0;" /* Cb -= 128 */ \

  64. 		     "psubsw "V_OFFSET"(%4), %%mm1;" /* Cr -= 128 */ \

  65. \

  66. 		     "movq %%mm0, %%mm2;" /* Copy 4 Cb 00 u3 00 u2 00 u1 00 u0 */ \

  67. 		     "movq %%mm1, %%mm3;" /* Copy 4 Cr 00 v3 00 v2 00 v1 00 v0 */ \

  68. \

  69. 		     "pmulhw "UG_COEFF"(%4), %%mm2;" /* Mul Cb with green coeff -> Cb green */ \

  70. 		     "pmulhw "VG_COEFF"(%4), %%mm3;" /* Mul Cr with green coeff -> Cr green */ \

  71. \

  72. 		     "pmulhw "UB_COEFF"(%4), %%mm0;" /* Mul Cb -> Cblue 00 b3 00 b2 00 b1 00 b0 */\

  73. 		     "pmulhw "VR_COEFF"(%4), %%mm1;" /* Mul Cr -> Cred 00 r3 00 r2 00 r1 00 r0 */\

  74. \

  75. 		     "paddsw %%mm3, %%mm2;" /* Cb green + Cr green -> Cgreen */\

  76. \

  77. 		     /* convert the luma part */\

  78. 		     "movq %%mm6, %%mm7;" /* Copy 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */\

  79. 		     "pand "MANGLE(mmx_00ffw)", %%mm6;" /* get Y even 00 Y6 00 Y4 00 Y2 00 Y0 */\

  80. \

  81. 		     "psrlw $8, %%mm7;" /* get Y odd 00 Y7 00 Y5 00 Y3 00 Y1 */\

  82. \

  83. 		     "psllw $3, %%mm6;" /* Promote precision */\

  84. 		     "psllw $3, %%mm7;" /* Promote precision */\

  85. \

  86. 		     "psubw "Y_OFFSET"(%4), %%mm6;" /* Y -= 16 */\

  87. 		     "psubw "Y_OFFSET"(%4), %%mm7;" /* Y -= 16 */\

  88. \

  89. 		     "pmulhw "Y_COEFF"(%4), %%mm6;" /* Mul 4 Y even 00 y6 00 y4 00 y2 00 y0 */\

  90. 		     "pmulhw "Y_COEFF"(%4), %%mm7;" /* Mul 4 Y odd 00 y7 00 y5 00 y3 00 y1 */\

  91. \

  92. 		     /* Do the addition part of the conversion for even and odd pixels,

  93. 			register usage:

  94. 			mm0 -> Cblue, mm1 -> Cred, mm2 -> Cgreen even pixels,

  95. 			mm3 -> Cblue, mm4 -> Cred, mm5 -> Cgreen odd pixels,

  96. 			mm6 -> Y even, mm7 -> Y odd */\

  97. 		     "movq %%mm0, %%mm3;" /* Copy Cblue */\

  98. 		     "movq %%mm1, %%mm4;" /* Copy Cred */\

  99. 		     "movq %%mm2, %%mm5;" /* Copy Cgreen */\

  100. \

  101. 		     "paddsw %%mm6, %%mm0;" /* Y even + Cblue 00 B6 00 B4 00 B2 00 B0 */\

  102. 		     "paddsw %%mm7, %%mm3;" /* Y odd + Cblue 00 B7 00 B5 00 B3 00 B1 */\

  103. \

  104. 		     "paddsw %%mm6, %%mm1;" /* Y even + Cred 00 R6 00 R4 00 R2 00 R0 */\

  105. 		     "paddsw %%mm7, %%mm4;" /* Y odd + Cred 00 R7 00 R5 00 R3 00 R1 */\

  106. \

  107. 		     "paddsw %%mm6, %%mm2;" /* Y even + Cgreen 00 G6 00 G4 00 G2 00 G0 */\

  108. 		     "paddsw %%mm7, %%mm5;" /* Y odd + Cgreen 00 G7 00 G5 00 G3 00 G1 */\

  109. \

  110. 		     /* Limit RGB even to 0..255 */\

  111. 		     "packuswb %%mm0, %%mm0;" /* B6 B4 B2 B0  B6 B4 B2 B0 */\

  112. 		     "packuswb %%mm1, %%mm1;" /* R6 R4 R2 R0  R6 R4 R2 R0 */\

  113. 		     "packuswb %%mm2, %%mm2;" /* G6 G4 G2 G0  G6 G4 G2 G0 */\

  114. \

  115. 		     /* Limit RGB odd to 0..255 */\

  116. 		     "packuswb %%mm3, %%mm3;" /* B7 B5 B3 B1  B7 B5 B3 B1 */\

  117. 		     "packuswb %%mm4, %%mm4;" /* R7 R5 R3 R1  R7 R5 R3 R1 */\

  118. 		     "packuswb %%mm5, %%mm5;" /* G7 G5 G3 G1  G7 G5 G3 G1 */\

  119. \

  120. 		     /* Interleave RGB even and odd */\

  121. 		     "punpcklbw %%mm3, %%mm0;" /* B7 B6 B5 B4 B3 B2 B1 B0 */\

  122. 		     "punpcklbw %%mm4, %%mm1;" /* R7 R6 R5 R4 R3 R2 R1 R0 */\

  123. 		     "punpcklbw %%mm5, %%mm2;" /* G7 G6 G5 G4 G3 G2 G1 G0 */\

  124. 

  125. 

  126. static inline int RENAME(yuv420_rgb16)(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,

  127.              int srcSliceH, uint8_t* dst[], int dstStride[]){

  128.     int y, h_size;

  129. 

  130.     if(c->srcFormat == IMGFMT_422P){

  131. 	srcStride[1] *= 2;

  132. 	srcStride[2] *= 2;

  133.     }

  134. 

  135.     h_size= (c->dstW+7)&~7;

  136.     if(h_size*2 > dstStride[0]) h_size-=8;

  137.     

  138.     __asm__ __volatile__ ("pxor %mm4, %mm4;" /* zero mm4 */ );

  139. //printf("%X %X %X %X %X %X %X %X %X %X\n", (int)&c->redDither, (int)&b5Dither, (int)src[0], (int)src[1], (int)src[2], (int)dst[0],

  140. //srcStride[0],srcStride[1],srcStride[2],dstStride[0]);

  141.     for (y= 0; y<srcSliceH; y++ ) {

  142. 	uint8_t *_image = dst[0] + (y+srcSliceY)*dstStride[0];

  143. 	uint8_t *_py = src[0] + y*srcStride[0];

  144. 	uint8_t *_pu = src[1] + (y>>1)*srcStride[1];

  145. 	uint8_t *_pv = src[2] + (y>>1)*srcStride[2];

  146. 	int index= -h_size/2;

  147. 

  148. 	b5Dither= dither8[y&1];

  149. 	g6Dither= dither4[y&1];

  150. 	g5Dither= dither8[y&1];

  151. 	r5Dither= dither8[(y+1)&1];

  152. 	    /* this mmx assembly code deals with SINGLE scan line at a time, it convert 8

  153. 	       pixels in each iteration */

  154. 	    __asm__ __volatile__ (

  155. 	/* load data for start of next scan line */

  156. 		     "movd (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  157. 		     "movd (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  158. 		     "movq (%5, %0, 2), %%mm6;" /* Load 8  Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  159. //		    ".balign 16			\n\t"

  160. 		    "1:				\n\t"

  161. /* no speed diference on my p3@500 with prefetch,

  162.  * if it is faster for anyone with -benchmark then tell me

  163. 			PREFETCH" 64(%0) \n\t"

  164. 			PREFETCH" 64(%1) \n\t"

  165. 			PREFETCH" 64(%2) \n\t"

  166. */

  167. YUV2RGB

  168. 

  169. #ifdef DITHER1XBPP

  170. 			"paddusb "MANGLE(b5Dither)", %%mm0;"

  171. 			"paddusb "MANGLE(g6Dither)", %%mm2;"

  172. 			"paddusb "MANGLE(r5Dither)", %%mm1;"

  173. #endif

  174. 		     /* mask unneeded bits off */

  175. 		     "pand "MANGLE(mmx_redmask)", %%mm0;" /* b7b6b5b4 b3_0_0_0 b7b6b5b4 b3_0_0_0 */

  176. 		     "pand "MANGLE(mmx_grnmask)", %%mm2;" /* g7g6g5g4 g3g2_0_0 g7g6g5g4 g3g2_0_0 */

  177. 		     "pand "MANGLE(mmx_redmask)", %%mm1;" /* r7r6r5r4 r3_0_0_0 r7r6r5r4 r3_0_0_0 */

  178. 

  179. 		     "psrlw $3,%%mm0;" /* 0_0_0_b7 b6b5b4b3 0_0_0_b7 b6b5b4b3 */

  180. 		     "pxor %%mm4, %%mm4;" /* zero mm4 */

  181. 

  182. 		     "movq %%mm0, %%mm5;" /* Copy B7-B0 */

  183. 		     "movq %%mm2, %%mm7;" /* Copy G7-G0 */

  184. 

  185. 		     /* convert rgb24 plane to rgb16 pack for pixel 0-3 */

  186. 		     "punpcklbw %%mm4, %%mm2;" /* 0_0_0_0 0_0_0_0 g7g6g5g4 g3g2_0_0 */

  187. 		     "punpcklbw %%mm1, %%mm0;" /* r7r6r5r4 r3_0_0_0 0_0_0_b7 b6b5b4b3 */

  188. 

  189. 		     "psllw $3, %%mm2;" /* 0_0_0_0 0_g7g6g5 g4g3g2_0 0_0_0_0 */

  190. 		     "por %%mm2, %%mm0;" /* r7r6r5r4 r3g7g6g5 g4g3g2b7 b6b5b4b3 */

  191. 

  192. 		     "movq 8 (%5, %0, 2), %%mm6;" /* Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  193. 		     MOVNTQ " %%mm0, (%1);" /* store pixel 0-3 */

  194. 

  195. 		     /* convert rgb24 plane to rgb16 pack for pixel 0-3 */

  196. 		     "punpckhbw %%mm4, %%mm7;" /* 0_0_0_0 0_0_0_0 g7g6g5g4 g3g2_0_0 */

  197. 		     "punpckhbw %%mm1, %%mm5;" /* r7r6r5r4 r3_0_0_0 0_0_0_b7 b6b5b4b3 */

  198. 

  199. 		     "psllw $3, %%mm7;" /* 0_0_0_0 0_g7g6g5 g4g3g2_0 0_0_0_0 */

  200. 		     "movd 4 (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  201. 

  202. 		     "por %%mm7, %%mm5;" /* r7r6r5r4 r3g7g6g5 g4g3g2b7 b6b5b4b3 */

  203. 		     "movd 4 (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  204. 

  205. 		     MOVNTQ " %%mm5, 8 (%1);" /* store pixel 4-7 */

  206. 		     

  207. 		     "addl $16, %1			\n\t"

  208. 		     "addl $4, %0			\n\t"

  209. 		     " js 1b				\n\t"

  210. 		     

  211. 		     : "+r" (index), "+r" (_image)

  212. 		     : "r" (_pu - index), "r" (_pv - index), "r"(&c->redDither), "r" (_py - 2*index)

  213. 		     );

  214.     }

  215. 

  216.     __asm__ __volatile__ (EMMS);

  217.     

  218.     return srcSliceH;

  219. }

  220. 

  221. static inline int RENAME(yuv420_rgb15)(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,

  222.              int srcSliceH, uint8_t* dst[], int dstStride[]){

  223.     int y, h_size;

  224. 

  225.     if(c->srcFormat == IMGFMT_422P){

  226. 	srcStride[1] *= 2;

  227. 	srcStride[2] *= 2;

  228.     }

  229. 

  230.     h_size= (c->dstW+7)&~7;

  231.     if(h_size*2 > dstStride[0]) h_size-=8;

  232.     

  233.     __asm__ __volatile__ ("pxor %mm4, %mm4;" /* zero mm4 */ );

  234. //printf("%X %X %X %X %X %X %X %X %X %X\n", (int)&c->redDither, (int)&b5Dither, (int)src[0], (int)src[1], (int)src[2], (int)dst[0],

  235. //srcStride[0],srcStride[1],srcStride[2],dstStride[0]);

  236.     for (y= 0; y<srcSliceH; y++ ) {

  237. 	uint8_t *_image = dst[0] + (y+srcSliceY)*dstStride[0];

  238. 	uint8_t *_py = src[0] + y*srcStride[0];

  239. 	uint8_t *_pu = src[1] + (y>>1)*srcStride[1];

  240. 	uint8_t *_pv = src[2] + (y>>1)*srcStride[2];

  241. 	int index= -h_size/2;

  242. 

  243. 	b5Dither= dither8[y&1];

  244. 	g6Dither= dither4[y&1];

  245. 	g5Dither= dither8[y&1];

  246. 	r5Dither= dither8[(y+1)&1];

  247. 	    /* this mmx assembly code deals with SINGLE scan line at a time, it convert 8

  248. 	       pixels in each iteration */

  249. 	    __asm__ __volatile__ (

  250. 	/* load data for start of next scan line */

  251. 		     "movd (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  252. 		     "movd (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  253. 		     "movq (%5, %0, 2), %%mm6;" /* Load 8  Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  254. //		    ".balign 16			\n\t"

  255. 		    "1:				\n\t"

  256. YUV2RGB

  257. 

  258. #ifdef DITHER1XBPP

  259. 			"paddusb "MANGLE(b5Dither)", %%mm0	\n\t"

  260. 			"paddusb "MANGLE(g5Dither)", %%mm2	\n\t"

  261. 			"paddusb "MANGLE(r5Dither)", %%mm1	\n\t"

  262. #endif

  263. 

  264. 		     /* mask unneeded bits off */

  265. 		     "pand "MANGLE(mmx_redmask)", %%mm0;" /* b7b6b5b4 b3_0_0_0 b7b6b5b4 b3_0_0_0 */

  266. 		     "pand "MANGLE(mmx_redmask)", %%mm2;" /* g7g6g5g4 g3_0_0_0 g7g6g5g4 g3_0_0_0 */

  267. 		     "pand "MANGLE(mmx_redmask)", %%mm1;" /* r7r6r5r4 r3_0_0_0 r7r6r5r4 r3_0_0_0 */

  268. 

  269. 		     "psrlw $3,%%mm0;" /* 0_0_0_b7 b6b5b4b3 0_0_0_b7 b6b5b4b3 */

  270. 		     "psrlw $1,%%mm1;"            /* 0_r7r6r5  r4r3_0_0 0_r7r6r5 r4r3_0_0 */

  271. 		     "pxor %%mm4, %%mm4;" /* zero mm4 */

  272. 

  273. 		     "movq %%mm0, %%mm5;" /* Copy B7-B0 */

  274. 		     "movq %%mm2, %%mm7;" /* Copy G7-G0 */

  275. 

  276. 		     /* convert rgb24 plane to rgb16 pack for pixel 0-3 */

  277. 		     "punpcklbw %%mm4, %%mm2;" /* 0_0_0_0 0_0_0_0 g7g6g5g4 g3_0_0_0 */

  278. 		     "punpcklbw %%mm1, %%mm0;" /* r7r6r5r4 r3_0_0_0 0_0_0_b7 b6b5b4b3 */

  279. 

  280. 		     "psllw $2, %%mm2;" /* 0_0_0_0 0_0_g7g6 g5g4g3_0 0_0_0_0 */

  281. 		     "por %%mm2, %%mm0;" /* 0_r7r6r5 r4r3g7g6 g5g4g3b7 b6b5b4b3 */

  282. 

  283. 		     "movq 8 (%5, %0, 2), %%mm6;" /* Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  284. 		     MOVNTQ " %%mm0, (%1);" /* store pixel 0-3 */

  285. 

  286. 		     /* convert rgb24 plane to rgb16 pack for pixel 0-3 */

  287. 		     "punpckhbw %%mm4, %%mm7;" /* 0_0_0_0 0_0_0_0 0_g7g6g5 g4g3_0_0 */

  288. 		     "punpckhbw %%mm1, %%mm5;" /* r7r6r5r4 r3_0_0_0 0_0_0_b7 b6b5b4b3 */

  289. 

  290. 		     "psllw $2, %%mm7;" /* 0_0_0_0 0_0_g7g6 g5g4g3_0 0_0_0_0 */

  291. 		     "movd 4 (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  292. 

  293. 		     "por %%mm7, %%mm5;" /* 0_r7r6r5 r4r3g7g6 g5g4g3b7 b6b5b4b3 */

  294. 		     "movd 4 (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  295. 

  296. 		     MOVNTQ " %%mm5, 8 (%1);" /* store pixel 4-7 */

  297. 		     

  298. 		     "addl $16, %1			\n\t"

  299. 		     "addl $4, %0			\n\t"

  300. 		     " js 1b				\n\t"

  301. 		     : "+r" (index), "+r" (_image)

  302. 		     : "r" (_pu - index), "r" (_pv - index), "r"(&c->redDither), "r" (_py - 2*index)

  303. 		     );

  304.     }

  305. 

  306.     __asm__ __volatile__ (EMMS);

  307.     return srcSliceH;

  308. }

  309. 

  310. static inline int RENAME(yuv420_rgb24)(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,

  311.              int srcSliceH, uint8_t* dst[], int dstStride[]){

  312.     int y, h_size;

  313. 

  314.     if(c->srcFormat == IMGFMT_422P){

  315. 	srcStride[1] *= 2;

  316. 	srcStride[2] *= 2;

  317.     }

  318. 

  319.     h_size= (c->dstW+7)&~7;

  320.     if(h_size*3 > dstStride[0]) h_size-=8;

  321.     

  322.     __asm__ __volatile__ ("pxor %mm4, %mm4;" /* zero mm4 */ );

  323. 

  324.     for (y= 0; y<srcSliceH; y++ ) {

  325. 	uint8_t *_image = dst[0] + (y+srcSliceY)*dstStride[0];

  326. 	uint8_t *_py = src[0] + y*srcStride[0];

  327. 	uint8_t *_pu = src[1] + (y>>1)*srcStride[1];

  328. 	uint8_t *_pv = src[2] + (y>>1)*srcStride[2];

  329. 	int index= -h_size/2;

  330. 

  331. 	    /* this mmx assembly code deals with SINGLE scan line at a time, it convert 8

  332. 	       pixels in each iteration */

  333. 	    __asm__ __volatile__ (

  334. 	/* load data for start of next scan line */

  335. 		     "movd (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  336. 		     "movd (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  337. 		     "movq (%5, %0, 2), %%mm6;" /* Load 8  Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  338. //		    ".balign 16			\n\t"

  339. 		    "1:				\n\t"

  340. YUV2RGB

  341. 	/* mm0=B, %%mm2=G, %%mm1=R */

  342. #ifdef HAVE_MMX2

  343. 			"movq "MANGLE(M24A)", %%mm4	\n\t"

  344. 			"movq "MANGLE(M24C)", %%mm7	\n\t"

  345. 			"pshufw $0x50, %%mm0, %%mm5	\n\t" /* B3 B2 B3 B2  B1 B0 B1 B0 */

  346. 			"pshufw $0x50, %%mm2, %%mm3	\n\t" /* G3 G2 G3 G2  G1 G0 G1 G0 */

  347. 			"pshufw $0x00, %%mm1, %%mm6	\n\t" /* R1 R0 R1 R0  R1 R0 R1 R0 */

  348. 

  349. 			"pand %%mm4, %%mm5		\n\t" /*    B2        B1       B0 */

  350. 			"pand %%mm4, %%mm3		\n\t" /*    G2        G1       G0 */

  351. 			"pand %%mm7, %%mm6		\n\t" /*       R1        R0       */

  352. 

  353. 			"psllq $8, %%mm3		\n\t" /* G2        G1       G0    */

  354. 			"por %%mm5, %%mm6		\n\t"

  355. 			"por %%mm3, %%mm6		\n\t"

  356. 			MOVNTQ" %%mm6, (%1)		\n\t"

  357. 

  358. 			"psrlq $8, %%mm2		\n\t" /* 00 G7 G6 G5  G4 G3 G2 G1 */

  359. 			"pshufw $0xA5, %%mm0, %%mm5	\n\t" /* B5 B4 B5 B4  B3 B2 B3 B2 */

  360. 			"pshufw $0x55, %%mm2, %%mm3	\n\t" /* G4 G3 G4 G3  G4 G3 G4 G3 */

  361. 			"pshufw $0xA5, %%mm1, %%mm6	\n\t" /* R5 R4 R5 R4  R3 R2 R3 R2 */

  362. 

  363. 			"pand "MANGLE(M24B)", %%mm5	\n\t" /* B5       B4        B3    */

  364. 			"pand %%mm7, %%mm3		\n\t" /*       G4        G3       */

  365. 			"pand %%mm4, %%mm6		\n\t" /*    R4        R3       R2 */

  366. 

  367. 			"por %%mm5, %%mm3		\n\t" /* B5    G4 B4     G3 B3    */

  368. 			"por %%mm3, %%mm6		\n\t"

  369. 			MOVNTQ" %%mm6, 8(%1)		\n\t"

  370. 

  371. 			"pshufw $0xFF, %%mm0, %%mm5	\n\t" /* B7 B6 B7 B6  B7 B6 B6 B7 */

  372. 			"pshufw $0xFA, %%mm2, %%mm3	\n\t" /* 00 G7 00 G7  G6 G5 G6 G5 */

  373. 			"pshufw $0xFA, %%mm1, %%mm6	\n\t" /* R7 R6 R7 R6  R5 R4 R5 R4 */

  374. 			"movd 4 (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  375. 

  376. 			"pand %%mm7, %%mm5		\n\t" /*       B7        B6       */

  377. 			"pand %%mm4, %%mm3		\n\t" /*    G7        G6       G5 */

  378. 			"pand "MANGLE(M24B)", %%mm6	\n\t" /* R7       R6        R5    */

  379. 			"movd 4 (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  380. \

  381. 			"por %%mm5, %%mm3		\n\t"

  382. 			"por %%mm3, %%mm6		\n\t"

  383. 			MOVNTQ" %%mm6, 16(%1)		\n\t"

  384. 			"movq 8 (%5, %0, 2), %%mm6;" /* Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  385. 			"pxor %%mm4, %%mm4		\n\t"

  386. 

  387. #else

  388. 

  389. 			"pxor %%mm4, %%mm4		\n\t"

  390. 			"movq %%mm0, %%mm5		\n\t" /* B */

  391. 			"movq %%mm1, %%mm6		\n\t" /* R */

  392. 			"punpcklbw %%mm2, %%mm0		\n\t" /* GBGBGBGB 0 */

  393. 			"punpcklbw %%mm4, %%mm1		\n\t" /* 0R0R0R0R 0 */

  394. 			"punpckhbw %%mm2, %%mm5		\n\t" /* GBGBGBGB 2 */

  395. 			"punpckhbw %%mm4, %%mm6		\n\t" /* 0R0R0R0R 2 */

  396. 			"movq %%mm0, %%mm7		\n\t" /* GBGBGBGB 0 */

  397. 			"movq %%mm5, %%mm3		\n\t" /* GBGBGBGB 2 */

  398. 			"punpcklwd %%mm1, %%mm7		\n\t" /* 0RGB0RGB 0 */

  399. 			"punpckhwd %%mm1, %%mm0		\n\t" /* 0RGB0RGB 1 */

  400. 			"punpcklwd %%mm6, %%mm5		\n\t" /* 0RGB0RGB 2 */

  401. 			"punpckhwd %%mm6, %%mm3		\n\t" /* 0RGB0RGB 3 */

  402. 

  403. 			"movq %%mm7, %%mm2		\n\t" /* 0RGB0RGB 0 */

  404. 			"movq %%mm0, %%mm6		\n\t" /* 0RGB0RGB 1 */

  405. 			"movq %%mm5, %%mm1		\n\t" /* 0RGB0RGB 2 */

  406. 			"movq %%mm3, %%mm4		\n\t" /* 0RGB0RGB 3 */

  407. 

  408. 			"psllq $40, %%mm7		\n\t" /* RGB00000 0 */

  409. 			"psllq $40, %%mm0		\n\t" /* RGB00000 1 */

  410. 			"psllq $40, %%mm5		\n\t" /* RGB00000 2 */

  411. 			"psllq $40, %%mm3		\n\t" /* RGB00000 3 */

  412. 

  413. 			"punpckhdq %%mm2, %%mm7		\n\t" /* 0RGBRGB0 0 */

  414. 			"punpckhdq %%mm6, %%mm0		\n\t" /* 0RGBRGB0 1 */

  415. 			"punpckhdq %%mm1, %%mm5		\n\t" /* 0RGBRGB0 2 */

  416. 			"punpckhdq %%mm4, %%mm3		\n\t" /* 0RGBRGB0 3 */

  417. 

  418. 			"psrlq $8, %%mm7		\n\t" /* 00RGBRGB 0 */

  419. 			"movq %%mm0, %%mm6		\n\t" /* 0RGBRGB0 1 */

  420. 			"psllq $40, %%mm0		\n\t" /* GB000000 1 */

  421. 			"por %%mm0, %%mm7		\n\t" /* GBRGBRGB 0 */

  422. 			MOVNTQ" %%mm7, (%1)		\n\t"

  423. 

  424. 			"movd 4 (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  425. 

  426. 			"psrlq $24, %%mm6		\n\t" /* 0000RGBR 1 */

  427. 			"movq %%mm5, %%mm1		\n\t" /* 0RGBRGB0 2 */

  428. 			"psllq $24, %%mm5		\n\t" /* BRGB0000 2 */

  429. 			"por %%mm5, %%mm6		\n\t" /* BRGBRGBR 1 */

  430. 			MOVNTQ" %%mm6, 8(%1)		\n\t"

  431. 

  432. 			"movq 8 (%5, %0, 2), %%mm6;" /* Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  433. 

  434. 			"psrlq $40, %%mm1		\n\t" /* 000000RG 2 */

  435. 			"psllq $8, %%mm3		\n\t" /* RGBRGB00 3 */

  436. 			"por %%mm3, %%mm1		\n\t" /* RGBRGBRG 2 */

  437. 			MOVNTQ" %%mm1, 16(%1)		\n\t"

  438. 

  439. 			"movd 4 (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  440. 			"pxor %%mm4, %%mm4		\n\t"

  441. #endif

  442. 		     

  443. 		     "addl $24, %1			\n\t"

  444. 		     "addl $4, %0			\n\t"

  445. 		     " js 1b				\n\t"

  446. 		     

  447. 		     : "+r" (index), "+r" (_image)

  448. 		     : "r" (_pu - index), "r" (_pv - index), "r"(&c->redDither), "r" (_py - 2*index)

  449. 		     );

  450.     }

  451. 

  452.     __asm__ __volatile__ (EMMS);

  453.     return srcSliceH;

  454. }

  455. 

  456. static inline int RENAME(yuv420_rgb32)(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,

  457.              int srcSliceH, uint8_t* dst[], int dstStride[]){

  458.     int y, h_size;

  459. 

  460.     if(c->srcFormat == IMGFMT_422P){

  461. 	srcStride[1] *= 2;

  462. 	srcStride[2] *= 2;

  463.     }

  464. 

  465.     h_size= (c->dstW+7)&~7;

  466.     if(h_size*4 > dstStride[0]) h_size-=8;

  467.     

  468.     __asm__ __volatile__ ("pxor %mm4, %mm4;" /* zero mm4 */ );

  469. 

  470.     for (y= 0; y<srcSliceH; y++ ) {

  471. 	uint8_t *_image = dst[0] + (y+srcSliceY)*dstStride[0];

  472. 	uint8_t *_py = src[0] + y*srcStride[0];

  473. 	uint8_t *_pu = src[1] + (y>>1)*srcStride[1];

  474. 	uint8_t *_pv = src[2] + (y>>1)*srcStride[2];

  475. 	int index= -h_size/2;

  476. 

  477. 	    /* this mmx assembly code deals with SINGLE scan line at a time, it convert 8

  478. 	       pixels in each iteration */

  479. 	    __asm__ __volatile__ (

  480. 	/* load data for start of next scan line */

  481. 		     "movd (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  482. 		     "movd (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  483. 		     "movq (%5, %0, 2), %%mm6;" /* Load 8  Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  484. //		    ".balign 16			\n\t"

  485. 		    "1:				\n\t"

  486. YUV2RGB

  487. 		     /* convert RGB plane to RGB packed format,

  488. 			mm0 -> B, mm1 -> R, mm2 -> G, mm3 -> 0,

  489. 			mm4 -> GB, mm5 -> AR pixel 4-7,

  490. 			mm6 -> GB, mm7 -> AR pixel 0-3 */

  491. 		     "pxor %%mm3, %%mm3;" /* zero mm3 */

  492. 

  493. 		     "movq %%mm0, %%mm6;" /* B7 B6 B5 B4 B3 B2 B1 B0 */

  494. 		     "movq %%mm1, %%mm7;" /* R7 R6 R5 R4 R3 R2 R1 R0 */

  495. 

  496. 		     "movq %%mm0, %%mm4;" /* B7 B6 B5 B4 B3 B2 B1 B0 */

  497. 		     "movq %%mm1, %%mm5;" /* R7 R6 R5 R4 R3 R2 R1 R0 */

  498. 

  499. 		     "punpcklbw %%mm2, %%mm6;" /* G3 B3 G2 B2 G1 B1 G0 B0 */

  500. 		     "punpcklbw %%mm3, %%mm7;" /* 00 R3 00 R2 00 R1 00 R0 */

  501. 

  502. 		     "punpcklwd %%mm7, %%mm6;" /* 00 R1 B1 G1 00 R0 B0 G0 */

  503. 		     MOVNTQ " %%mm6, (%1);" /* Store ARGB1 ARGB0 */

  504. 

  505. 		     "movq %%mm0, %%mm6;" /* B7 B6 B5 B4 B3 B2 B1 B0 */

  506. 		     "punpcklbw %%mm2, %%mm6;" /* G3 B3 G2 B2 G1 B1 G0 B0 */

  507. 

  508. 		     "punpckhwd %%mm7, %%mm6;" /* 00 R3 G3 B3 00 R2 B3 G2 */

  509. 		     MOVNTQ " %%mm6, 8 (%1);" /* Store ARGB3 ARGB2 */

  510. 

  511. 		     "punpckhbw %%mm2, %%mm4;" /* G7 B7 G6 B6 G5 B5 G4 B4 */

  512. 		     "punpckhbw %%mm3, %%mm5;" /* 00 R7 00 R6 00 R5 00 R4 */

  513. 

  514. 		     "punpcklwd %%mm5, %%mm4;" /* 00 R5 B5 G5 00 R4 B4 G4 */

  515. 		     MOVNTQ " %%mm4, 16 (%1);" /* Store ARGB5 ARGB4 */

  516. 

  517. 		     "movq %%mm0, %%mm4;" /* B7 B6 B5 B4 B3 B2 B1 B0 */

  518. 		     "punpckhbw %%mm2, %%mm4;" /* G7 B7 G6 B6 G5 B5 G4 B4 */

  519. 

  520. 		     "punpckhwd %%mm5, %%mm4;" /* 00 R7 G7 B7 00 R6 B6 G6 */

  521. 		     MOVNTQ " %%mm4, 24 (%1);" /* Store ARGB7 ARGB6 */

  522. 

  523. 		     "movd 4 (%2, %0), %%mm0;" /* Load 4 Cb 00 00 00 00 u3 u2 u1 u0 */

  524. 		     "movd 4 (%3, %0), %%mm1;" /* Load 4 Cr 00 00 00 00 v3 v2 v1 v0 */

  525. 

  526. 		     "pxor %%mm4, %%mm4;" /* zero mm4 */

  527. 		     "movq 8 (%5, %0, 2), %%mm6;" /* Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */

  528. 

  529. 		     "addl $32, %1			\n\t"

  530. 		     "addl $4, %0			\n\t"

  531. 		     " js 1b				\n\t"

  532. 		     

  533. 		     : "+r" (index), "+r" (_image)

  534. 		     : "r" (_pu - index), "r" (_pv - index), "r"(&c->redDither), "r" (_py - 2*index)

  535. 		     );

  536.     }

  537. 

  538.     __asm__ __volatile__ (EMMS);

  539.     return srcSliceH;

  540. }