git.sesse.net Git - vlc/blob - modules/video_chroma/i420_yuy2.h

   1 /*****************************************************************************
   2  * i420_yuy2.h : YUV to YUV conversion module for vlc
   3  *****************************************************************************
   4  * Copyright (C) 2000, 2001 the VideoLAN team
   5  * $Id$
   6  *
   7  * Authors: Samuel Hocevar <sam@zoy.org>
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program 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
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  22  *****************************************************************************/
  23
  24 #ifdef MODULE_NAME_IS_i420_yuy2_mmx
  25
  26 #define MMX_CALL(MMX_INSTRUCTIONS)                                          \
  27     do {                                                                    \
  28     __asm__ __volatile__(                                                   \
  29         ".align 8 \n\t"                                                     \
  30         MMX_INSTRUCTIONS                                                    \
  31         :                                                                   \
  32         : "r" (p_line1),  "r" (p_line2),  "r" (p_y1),  "r" (p_y2),          \
  33           "r" (p_u), "r" (p_v) );                                           \
  34     p_line1 += 16; p_line2 += 16; p_y1 += 8; p_y2 += 8; p_u += 4; p_v += 4; \
  35     } while(0);                                                             \
  36
  37 #define MMX_YUV420_YUYV "                                                 \n\
  38 movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
  39 movd       (%4), %%mm1  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
  40 movd       (%5), %%mm2  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
  41 punpcklbw %%mm2, %%mm1  #                     v3 u3 v2 u2 v1 u1 v0 u0     \n\
  42 movq      %%mm0, %%mm2  #                     y7 y6 y5 y4 y3 y2 y1 y0     \n\
  43 punpcklbw %%mm1, %%mm2  #                     v1 y3 u1 y2 v0 y1 u0 y0     \n\
  44 movq      %%mm2, (%0)   # Store low YUYV                                  \n\
  45 punpckhbw %%mm1, %%mm0  #                     v3 y7 u3 y6 v2 y5 u2 y4     \n\
  46 movq      %%mm0, 8(%0)  # Store high YUYV                                 \n\
  47 movq       (%3), %%mm0  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
  48 movq      %%mm0, %%mm2  #                     Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
  49 punpcklbw %%mm1, %%mm2  #                     v1 Y3 u1 Y2 v0 Y1 u0 Y0     \n\
  50 movq      %%mm2, (%1)   # Store low YUYV                                  \n\
  51 punpckhbw %%mm1, %%mm0  #                     v3 Y7 u3 Y6 v2 Y5 u2 Y4     \n\
  52 movq      %%mm0, 8(%1)  # Store high YUYV                                 \n\
  53 "
  54
  55 #define MMX_YUV420_YVYU "                                                 \n\
  56 movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
  57 movd       (%4), %%mm2  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
  58 movd       (%5), %%mm1  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
  59 punpcklbw %%mm2, %%mm1  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
  60 movq      %%mm0, %%mm2  #                     y7 y6 y5 y4 y3 y2 y1 y0     \n\
  61 punpcklbw %%mm1, %%mm2  #                     u1 y3 v1 y2 u0 y1 v0 y0     \n\
  62 movq      %%mm2, (%0)   # Store low YUYV                                  \n\
  63 punpckhbw %%mm1, %%mm0  #                     u3 y7 v3 y6 u2 y5 v2 y4     \n\
  64 movq      %%mm0, 8(%0)  # Store high YUYV                                 \n\
  65 movq       (%3), %%mm0  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
  66 movq      %%mm0, %%mm2  #                     Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
  67 punpcklbw %%mm1, %%mm2  #                     u1 Y3 v1 Y2 u0 Y1 v0 Y0     \n\
  68 movq      %%mm2, (%1)   # Store low YUYV                                  \n\
  69 punpckhbw %%mm1, %%mm0  #                     u3 Y7 v3 Y6 u2 Y5 v2 Y4     \n\
  70 movq      %%mm0, 8(%1)  # Store high YUYV                                 \n\
  71 "
  72
  73 #define MMX_YUV420_UYVY "                                                 \n\
  74 movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
  75 movq       (%3), %%mm3  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
  76 movd       (%4), %%mm2  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
  77 movd       (%5), %%mm1  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
  78 punpcklbw %%mm2, %%mm1  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
  79 movq      %%mm1, %%mm2  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
  80 punpcklbw %%mm0, %%mm2  #                     y3 v1 y2 u1 y1 v0 y0 u0     \n\
  81 movq      %%mm2, (%0)   # Store low UYVY                                  \n\
  82 movq      %%mm1, %%mm2  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
  83 punpckhbw %%mm0, %%mm2  #                     y3 v1 y2 u1 y1 v0 y0 u0     \n\
  84 movq      %%mm2, 8(%0)  # Store high UYVY                                 \n\
  85 movq      %%mm1, %%mm2  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
  86 punpcklbw %%mm3, %%mm2  #                     Y3 v1 Y2 u1 Y1 v0 Y0 u0     \n\
  87 movq      %%mm2, (%1)   # Store low UYVY                                  \n\
  88 punpckhbw %%mm3, %%mm1  #                     Y7 v3 Y6 u3 Y5 v2 Y4 u2     \n\
  89 movq      %%mm1, 8(%1)  # Store high UYVY                                 \n\
  90 "
  91
  92 /* FIXME: this code does not work ! Chroma seems to be wrong. */
  93 #define MMX_YUV420_Y211 "                                                 \n\
  94 movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
  95 movq       (%3), %%mm1  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
  96 movd       (%4), %%mm2  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
  97 movd       (%5), %%mm3  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
  98 pand    i_00ffw, %%mm0  # get Y even          00 Y6 00 Y4 00 Y2 00 Y0     \n\
  99 packuswb  %%mm0, %%mm0  # pack Y              y6 y4 y2 y0 y6 y4 y2 y0     \n\
 100 pand    i_00ffw, %%mm2  # get U even          00 u6 00 u4 00 u2 00 u0     \n\
 101 packuswb  %%mm2, %%mm2  # pack U              00 00 u2 u0 00 00 u2 u0     \n\
 102 pand    i_00ffw, %%mm3  # get V even          00 v6 00 v4 00 v2 00 v0     \n\
 103 packuswb  %%mm3, %%mm3  # pack V              00 00 v2 v0 00 00 v2 v0     \n\
 104 punpcklbw %%mm3, %%mm2  #                     00 00 00 00 v2 u2 v0 u0     \n\
 105 psubsw    i_80w, %%mm2  # U,V -= 128                                      \n\
 106 punpcklbw %%mm2, %%mm0  #                     v2 y6 u2 y4 v0 y2 u0 y0     \n\
 107 movq      %%mm0, (%0)   # Store YUYV                                      \n\
 108 pand    i_00ffw, %%mm1  # get Y even          00 Y6 00 Y4 00 Y2 00 Y0     \n\
 109 packuswb  %%mm1, %%mm1  # pack Y              Y6 Y4 Y2 Y0 Y6 Y4 Y2 Y0     \n\
 110 punpcklbw %%mm2, %%mm1  #                     v2 Y6 u2 Y4 v0 Y2 u0 Y0     \n\
 111 movq      %%mm1, (%1)   # Store YUYV                                      \n\
 112 "
 113
 114 #else
 115
 116 #define C_YUV420_YVYU( )                                                    \
 117     *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
 118     *(p_line1)++ =            *(p_line2)++ = *(p_v)++;                      \
 119     *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
 120     *(p_line1)++ =            *(p_line2)++ = *(p_u)++;                      \
 121
 122 #define C_YUV420_UYVY( )                                                    \
 123     *(p_line1)++ =            *(p_line2)++ = *(p_u)++;                      \
 124     *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
 125     *(p_line1)++ =            *(p_line2)++ = *(p_v)++;                      \
 126     *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
 127
 128 #define C_YUV420_Y211( )                                                    \
 129     *(p_line1)++ = *(p_y1); p_y1 += 2;                                      \
 130     *(p_line2)++ = *(p_y2); p_y2 += 2;                                      \
 131     *(p_line1)++ = *(p_line2)++ = *(p_u) - 0x80; p_u += 2;                  \
 132     *(p_line1)++ = *(p_y1); p_y1 += 2;                                      \
 133     *(p_line2)++ = *(p_y2); p_y2 += 2;                                      \
 134     *(p_line1)++ = *(p_line2)++ = *(p_v) - 0x80; p_v += 2;                  \
 135
 136 #endif
 137
 138 /* Used in both MMX and C modules */
 139 #define C_YUV420_YUYV( )                                                    \
 140     *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
 141     *(p_line1)++ =            *(p_line2)++ = *(p_u)++;                      \
 142     *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
 143     *(p_line1)++ =            *(p_line2)++ = *(p_v)++;                      \
 144