i420_yuy2.h 26 KB
Newer Older
1 2 3
/*****************************************************************************
 * i420_yuy2.h : YUV to YUV conversion module for vlc
 *****************************************************************************
4
 * Copyright (C) 2000, 2001 the VideoLAN team
5
 * $Id$
6 7
 *
 * Authors: Samuel Hocevar <sam@zoy.org>
8
 *          Damien Fouilleul <damien@videolan.org>
9 10 11 12 13 14 15 16 17 18 19 20 21
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 * 
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
Antoine Cellerier's avatar
Antoine Cellerier committed
22
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
23 24 25 26
 *****************************************************************************/

#ifdef MODULE_NAME_IS_i420_yuy2_mmx

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
#if defined(CAN_COMPILE_MMX)

/* MMX assembly */
 
#define MMX_CALL(MMX_INSTRUCTIONS)          \
    do {                                    \
    __asm__ __volatile__(                   \
        ".p2align 3 \n\t"                   \
        MMX_INSTRUCTIONS                    \
        :                                   \
        : "r" (p_line1), "r" (p_line2),     \
          "r" (p_y1), "r" (p_y2),           \
          "r" (p_u), "r" (p_v) );           \
        p_line1 += 16; p_line2 += 16;       \
        p_y1 += 8; p_y2 += 8;               \
        p_u += 4; p_v += 4;                 \
43
    } while(0)
44

45 46
#define MMX_END __asm__ __volatile__ ( "emms" )

47 48 49
#define MMX_YUV420_YUYV "                                                 \n\
movd       (%4), %%mm1  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
movd       (%5), %%mm2  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
50 51
movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
movq       (%3), %%mm3  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
52 53 54 55 56 57
punpcklbw %%mm2, %%mm1  #                     v3 u3 v2 u2 v1 u1 v0 u0     \n\
movq      %%mm0, %%mm2  #                     y7 y6 y5 y4 y3 y2 y1 y0     \n\
punpcklbw %%mm1, %%mm2  #                     v1 y3 u1 y2 v0 y1 u0 y0     \n\
movq      %%mm2, (%0)   # Store low YUYV                                  \n\
punpckhbw %%mm1, %%mm0  #                     v3 y7 u3 y6 v2 y5 u2 y4     \n\
movq      %%mm0, 8(%0)  # Store high YUYV                                 \n\
58 59 60 61 62
movq      %%mm3, %%mm4  #                     Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
punpcklbw %%mm1, %%mm4  #                     v1 Y3 u1 Y2 v0 Y1 u0 Y0     \n\
movq      %%mm4, (%1)   # Store low YUYV                                  \n\
punpckhbw %%mm1, %%mm3  #                     v3 Y7 u3 Y6 v2 Y5 u2 Y4     \n\
movq      %%mm3, 8(%1)  # Store high YUYV                                 \n\
63 64 65 66 67
"

#define MMX_YUV420_YVYU "                                                 \n\
movd       (%4), %%mm2  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
movd       (%5), %%mm1  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
68 69
movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
movq       (%3), %%mm3  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
70 71 72 73 74 75
punpcklbw %%mm2, %%mm1  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
movq      %%mm0, %%mm2  #                     y7 y6 y5 y4 y3 y2 y1 y0     \n\
punpcklbw %%mm1, %%mm2  #                     u1 y3 v1 y2 u0 y1 v0 y0     \n\
movq      %%mm2, (%0)   # Store low YUYV                                  \n\
punpckhbw %%mm1, %%mm0  #                     u3 y7 v3 y6 u2 y5 v2 y4     \n\
movq      %%mm0, 8(%0)  # Store high YUYV                                 \n\
76 77 78 79 80
movq      %%mm3, %%mm4  #                     Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
punpcklbw %%mm1, %%mm4  #                     u1 Y3 v1 Y2 u0 Y1 v0 Y0     \n\
movq      %%mm4, (%1)   # Store low YUYV                                  \n\
punpckhbw %%mm1, %%mm3  #                     u3 Y7 v3 Y6 u2 Y5 v2 Y4     \n\
movq      %%mm3, 8(%1)  # Store high YUYV                                 \n\
81 82 83
"

#define MMX_YUV420_UYVY "                                                 \n\
84 85
movd       (%4), %%mm1  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
movd       (%5), %%mm2  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
86 87
movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
movq       (%3), %%mm3  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
88 89
punpcklbw %%mm2, %%mm1  #                     v3 u3 v2 u2 v1 u1 v0 u0     \n\
movq      %%mm1, %%mm2  #                     v3 u3 v2 u2 v1 u1 v0 u0     \n\
90 91 92 93 94
punpcklbw %%mm0, %%mm2  #                     y3 v1 y2 u1 y1 v0 y0 u0     \n\
movq      %%mm2, (%0)   # Store low UYVY                                  \n\
movq      %%mm1, %%mm2  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
punpckhbw %%mm0, %%mm2  #                     y3 v1 y2 u1 y1 v0 y0 u0     \n\
movq      %%mm2, 8(%0)  # Store high UYVY                                 \n\
95 96 97
movq      %%mm1, %%mm4  #                     u3 v3 u2 v2 u1 v1 u0 v0     \n\
punpcklbw %%mm3, %%mm4  #                     Y3 v1 Y2 u1 Y1 v0 Y0 u0     \n\
movq      %%mm4, (%1)   # Store low UYVY                                  \n\
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
punpckhbw %%mm3, %%mm1  #                     Y7 v3 Y6 u3 Y5 v2 Y4 u2     \n\
movq      %%mm1, 8(%1)  # Store high UYVY                                 \n\
"

/* FIXME: this code does not work ! Chroma seems to be wrong. */
#define MMX_YUV420_Y211 "                                                 \n\
movq       (%2), %%mm0  # Load 8 Y            y7 y6 y5 y4 y3 y2 y1 y0     \n\
movq       (%3), %%mm1  # Load 8 Y            Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
movd       (%4), %%mm2  # Load 4 Cb           00 00 00 00 u3 u2 u1 u0     \n\
movd       (%5), %%mm3  # Load 4 Cr           00 00 00 00 v3 v2 v1 v0     \n\
pand    i_00ffw, %%mm0  # get Y even          00 Y6 00 Y4 00 Y2 00 Y0     \n\
packuswb  %%mm0, %%mm0  # pack Y              y6 y4 y2 y0 y6 y4 y2 y0     \n\
pand    i_00ffw, %%mm2  # get U even          00 u6 00 u4 00 u2 00 u0     \n\
packuswb  %%mm2, %%mm2  # pack U              00 00 u2 u0 00 00 u2 u0     \n\
pand    i_00ffw, %%mm3  # get V even          00 v6 00 v4 00 v2 00 v0     \n\
packuswb  %%mm3, %%mm3  # pack V              00 00 v2 v0 00 00 v2 v0     \n\
punpcklbw %%mm3, %%mm2  #                     00 00 00 00 v2 u2 v0 u0     \n\
psubsw    i_80w, %%mm2  # U,V -= 128                                      \n\
punpcklbw %%mm2, %%mm0  #                     v2 y6 u2 y4 v0 y2 u0 y0     \n\
movq      %%mm0, (%0)   # Store YUYV                                      \n\
pand    i_00ffw, %%mm1  # get Y even          00 Y6 00 Y4 00 Y2 00 Y0     \n\
packuswb  %%mm1, %%mm1  # pack Y              Y6 Y4 Y2 Y0 Y6 Y4 Y2 Y0     \n\
punpcklbw %%mm2, %%mm1  #                     v2 Y6 u2 Y4 v0 Y2 u0 Y0     \n\
movq      %%mm1, (%1)   # Store YUYV                                      \n\
"
123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192
#elif defined(HAVE_MMX_INTRINSICS)

/* MMX intrinsics */

#include <mmintrin.h>

#define MMX_CALL(MMX_INSTRUCTIONS)          \
    do {                                    \
        __m64 mm0, mm1, mm2, mm3, mm4;      \
        MMX_INSTRUCTIONS                    \
        p_line1 += 16; p_line2 += 16;       \
        p_y1 += 8; p_y2 += 8;               \
        p_u += 4; p_v += 4;                 \
    } while(0)

#define MMX_END _mm_empty()
    
#define MMX_YUV420_YUYV                     \
    mm1 = _mm_cvtsi32_si64((int)*p_u);      \
    mm2 = _mm_cvtsi32_si64((int)*p_v);      \
    mm0 = (__m64)*(uint64_t*)p_y1;          \
    mm3 = (__m64)*(uint64_t*)p_y2;          \
    mm1 = _mm_unpacklo_pi8(mm1, mm2);       \
    mm2 = mm0;                              \
    mm2 = _mm_unpacklo_pi8(mm2, mm1);       \
    *(uin64_t)p_line1 = (uint64)mm2;        \
    mm0 = _mm_unpackhi_pi8(mm0, mm1);       \
    *(uin64_t)(p_line1 + 4) = (uint64)mm0;  \
    mm4 = mm3;                              \
    mm4 = _mm_unpacklo_pi8(mm4, mm1);       \
    *(uin64_t)p_line2 = (uint64)mm4;        \
    mm3 = _mm_unpackhi_pi8(mm3, mm1);       \
    *(uin64_t)(p_line2 + 4) = (uint64)mm4;

#define MMX_YUV420_YVYU                     \
    mm2 = _mm_cvtsi32_si64((int)*p_u);      \
    mm1 = _mm_cvtsi32_si64((int)*p_v);      \
    mm0 = (__m64)*(uint64_t*)p_y1;          \
    mm3 = (__m64)*(uint64_t*)p_y2;          \
    mm1 = _mm_unpacklo_pi8(mm1, mm2);       \
    mm2 = mm0;                              \
    mm2 = _mm_unpacklo_pi8(mm2, mm1);       \
    *(uin64_t)p_line1 = (uint64)mm2;        \
    mm0 = _mm_unpackhi_pi8(mm0, mm1);       \
    *(uin64_t)(p_line1 + 4) = (uint64)mm0;  \
    mm4 = mm3;                              \
    mm4 = _mm_unpacklo_pi8(mm4, mm1);       \
    *(uin64_t)p_line2 = (uint64)mm4;        \
    mm3 = _mm_unpackhi_pi8(mm3, mm1);       \
    *(uin64_t)(p_line2 + 4) = (uint64)mm4;

#define MMX_YUV420_UYVY                     \
    mm1 = _mm_cvtsi32_si64((int)*p_u);      \
    mm2 = _mm_cvtsi32_si64((int)*p_v);      \
    mm0 = (__m64)*(uint64_t*)p_y1;          \
    mm3 = (__m64)*(uint64_t*)p_y2;          \
    mm1 = _mm_unpacklo_pi8(mm1, mm2);       \
    mm2 = mm1;                              \
    mm2 = _mm_unpacklo_pi8(mm2, mm0);       \
    *(uin64_t)p_line1 = (uint64)mm2;        \
    mm2 = mm1;                              \
    mm2 = _mm_unpackhi_pi8(mm2, mm0);       \
    *(uin64_t)(p_line1 + 4) = (uint64)mm2;  \
    mm4 = mm1;                              \
    mm4 = _mm_unpacklo_pi8(mm4, mm3);       \
    *(uin64_t)p_line2 = (uint64)mm4;        \
    mm1 = _mm_unpackhi_pi8(mm1, mm3);       \
    *(uin64_t)(p_line2 + 4) = (uint64)mm1;

#endif
193

194 195
#elif defined( MODULE_NAME_IS_i420_yuy2_sse2 )

196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
#if defined(CAN_COMPILE_SSE2)

/* SSE2 assembly */

#define SSE2_CALL(SSE2_INSTRUCTIONS)    \
    do {                                \
    __asm__ __volatile__(               \
        ".p2align 3 \n\t"               \
        SSE2_INSTRUCTIONS               \
        :                               \
        : "r" (p_line1), "r" (p_line2), \
          "r" (p_y1),  "r" (p_y2),      \
          "r" (p_u), "r" (p_v) );       \
        p_line1 += 32; p_line2 += 32;   \
        p_y1 += 16; p_y2 += 16;         \
        p_u += 8; p_v += 8;             \
212 213
    } while(0)

214 215
#define SSE2_END  __asm__ __volatile__ ( "sfence" ::: "memory" )

216 217 218
#define SSE2_YUV420_YUYV_ALIGNED "                                        \n\
movq        (%4), %%xmm1  # Load 8 Cb         u7 u6 u5 u4 u3 u2 u1 u0     \n\
movq        (%5), %%xmm2  # Load 8 Cr         v7 06 v5 v4 v3 v2 v1 v0     \n\
219 220
movdqa      (%2), %%xmm0  # Load 16 Y         y15 y14 y13 .. y2 y1 y0     \n\
movdqa      (%3), %%xmm3  # Load 16 Y         Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
221 222 223
punpcklbw %%xmm2, %%xmm1  #                   v7 u7 v6 u6 .. u1 v0 u0     \n\
movdqa    %%xmm0, %%xmm2  #                   y15 y14 y13 .. y2 y1 y0     \n\
punpcklbw %%xmm1, %%xmm2  #                   v3 y7 u3 .. v0 y1 u0 y0     \n\
224
movntdq   %%xmm2, (%0)    # Store low YUYV                                \n\
225
punpckhbw %%xmm1, %%xmm0  #                   v3 y7 u3 y6 v2 y5 u2 y4     \n\
226
movntdq   %%xmm0, 16(%0)  # Store high YUYV                               \n\
227 228
movdqa    %%xmm3, %%xmm4  #                   Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
punpcklbw %%xmm1, %%xmm4  #                   v1 Y3 u1 Y2 v0 Y1 u0 Y0     \n\
229
movntdq   %%xmm4, (%1)    # Store low YUYV                                \n\
230
punpckhbw %%xmm1, %%xmm3  #                   v3 Y7 u3 Y6 v2 Y5 u2 Y4     \n\
231
movntdq   %%xmm3, 16(%1)  # Store high YUYV                               \n\
232 233 234 235 236
"

#define SSE2_YUV420_YUYV_UNALIGNED "                                      \n\
movq        (%4), %%xmm1  # Load 8 Cb         00 00 00 00 u3 u2 u1 u0     \n\
movq        (%5), %%xmm2  # Load 8 Cr         00 00 00 00 v3 v2 v1 v0     \n\
237 238
movdqu      (%2), %%xmm0  # Load 16 Y         y7 y6 y5 y4 y3 y2 y1 y0     \n\
movdqu      (%3), %%xmm3  # Load 16 Y         Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
239 240
prefetchnta (%0)          # Tell CPU not to cache output YUYV data        \n\
prefetchnta (%1)          # Tell CPU not to cache output YUYV data        \n\
241 242 243 244 245 246
punpcklbw %%xmm2, %%xmm1  #                   v3 u3 v2 u2 v1 u1 v0 u0     \n\
movdqa    %%xmm0, %%xmm2  #                   y7 y6 y5 y4 y3 y2 y1 y0     \n\
punpcklbw %%xmm1, %%xmm2  #                   v1 y3 u1 y2 v0 y1 u0 y0     \n\
movdqu    %%xmm2, (%0)    # Store low YUYV                                \n\
punpckhbw %%xmm1, %%xmm0  #                   v3 y7 u3 y6 v2 y5 u2 y4     \n\
movdqu    %%xmm0, 16(%0)  # Store high YUYV                               \n\
247 248 249 250 251
movdqa    %%xmm3, %%xmm4  #                   Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0     \n\
punpcklbw %%xmm1, %%xmm4  #                   v1 Y3 u1 Y2 v0 Y1 u0 Y0     \n\
movdqu    %%xmm4, (%1)    # Store low YUYV                                \n\
punpckhbw %%xmm1, %%xmm3  #                   v3 Y7 u3 Y6 v2 Y5 u2 Y4     \n\
movdqu    %%xmm3, 16(%1)  # Store high YUYV                               \n\
252 253
"

254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
#define SSE2_YUV420_YVYU_ALIGNED "                                        \n\
movq        (%4), %%xmm2  # Load 8 Cb           00 00 00 00 u3 u2 u1 u0   \n\
movq        (%5), %%xmm1  # Load 8 Cr           00 00 00 00 v3 v2 v1 v0   \n\
movdqa      (%2), %%xmm0  # Load 16 Y           y7 y6 y5 y4 y3 y2 y1 y0   \n\
movdqa      (%3), %%xmm3  # Load 16 Y           Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0   \n\
punpcklbw %%xmm2, %%xmm1  #                     u3 v3 u2 v2 u1 v1 u0 v0   \n\
movdqa    %%xmm0, %%xmm2  #                     y7 y6 y5 y4 y3 y2 y1 y0   \n\
punpcklbw %%xmm1, %%xmm2  #                     u1 y3 v1 y2 u0 y1 v0 y0   \n\
movntdq   %%xmm2, (%0)    # Store low YUYV                                \n\
punpckhbw %%xmm1, %%xmm0  #                     u3 y7 v3 y6 u2 y5 v2 y4   \n\
movntdq   %%xmm0, 16(%0)  # Store high YUYV                               \n\
movdqa    %%xmm3, %%xmm4  #                     Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0   \n\
punpcklbw %%xmm1, %%xmm4  #                     u1 Y3 v1 Y2 u0 Y1 v0 Y0   \n\
movntdq   %%xmm4, (%1)    # Store low YUYV                                \n\
punpckhbw %%xmm1, %%xmm3  #                     u3 Y7 v3 Y6 u2 Y5 v2 Y4   \n\
movntdq   %%xmm3, 16(%1)  # Store high YUYV                               \n\
270 271
"

272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
#define SSE2_YUV420_YVYU_UNALIGNED "                                      \n\
movq        (%4), %%xmm2  # Load 8 Cb           00 00 00 00 u3 u2 u1 u0   \n\
movq        (%5), %%xmm1  # Load 8 Cr           00 00 00 00 v3 v2 v1 v0   \n\
movdqu      (%2), %%xmm0  # Load 16 Y           y7 y6 y5 y4 y3 y2 y1 y0   \n\
movdqu      (%3), %%xmm3  # Load 16 Y           Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0   \n\
prefetchnta (%0)          # Tell CPU not to cache output YVYU data        \n\
prefetchnta (%1)          # Tell CPU not to cache output YVYU data        \n\
punpcklbw %%xmm2, %%xmm1  #                     u3 v3 u2 v2 u1 v1 u0 v0   \n\
movdqu    %%xmm0, %%xmm2  #                     y7 y6 y5 y4 y3 y2 y1 y0   \n\
punpcklbw %%xmm1, %%xmm2  #                     u1 y3 v1 y2 u0 y1 v0 y0   \n\
movdqu    %%xmm2, (%0)    # Store low YUYV                                \n\
punpckhbw %%xmm1, %%xmm0  #                     u3 y7 v3 y6 u2 y5 v2 y4   \n\
movdqu    %%xmm0, 16(%0)  # Store high YUYV                               \n\
movdqu    %%xmm3, %%xmm4  #                     Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0   \n\
punpcklbw %%xmm1, %%xmm4  #                     u1 Y3 v1 Y2 u0 Y1 v0 Y0   \n\
movdqu    %%xmm4, (%1)    # Store low YUYV                                \n\
punpckhbw %%xmm1, %%xmm3  #                     u3 Y7 v3 Y6 u2 Y5 v2 Y4   \n\
movdqu    %%xmm3, 16(%1)  # Store high YUYV                               \n\
290 291
"

292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
#define SSE2_YUV420_UYVY_ALIGNED "                                        \n\
movq        (%4), %%xmm1  # Load 8 Cb           00 00 00 00 u3 u2 u1 u0   \n\
movq        (%5), %%xmm2  # Load 8 Cr           00 00 00 00 v3 v2 v1 v0   \n\
movdqa      (%2), %%xmm0  # Load 16 Y           y7 y6 y5 y4 y3 y2 y1 y0   \n\
movdqa      (%3), %%xmm3  # Load 16 Y           Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0   \n\
punpcklbw %%xmm2, %%xmm1  #                     v3 u3 v2 u2 v1 u1 v0 u0   \n\
movdqa    %%xmm1, %%xmm2  #                     v3 u3 v2 u2 v1 u1 v0 u0   \n\
punpcklbw %%xmm0, %%xmm2  #                     y3 v1 y2 u1 y1 v0 y0 u0   \n\
movntdq   %%xmm2, (%0)    # Store low UYVY                                \n\
movdqa    %%xmm1, %%xmm2  #                     u3 v3 u2 v2 u1 v1 u0 v0   \n\
punpckhbw %%xmm0, %%xmm2  #                     y3 v1 y2 u1 y1 v0 y0 u0   \n\
movntdq   %%xmm2, 16(%0)  # Store high UYVY                               \n\
movdqa    %%xmm1, %%xmm4  #                     u3 v3 u2 v2 u1 v1 u0 v0   \n\
punpcklbw %%xmm3, %%xmm4  #                     Y3 v1 Y2 u1 Y1 v0 Y0 u0   \n\
movntdq   %%xmm4, (%1)    # Store low UYVY                                \n\
punpckhbw %%xmm3, %%xmm1  #                     Y7 v3 Y6 u3 Y5 v2 Y4 u2   \n\
movntdq   %%xmm1, 16(%1)  # Store high UYVY                               \n\
309 310
"

311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329
#define SSE2_YUV420_UYVY_UNALIGNED "                                      \n\
movq        (%4), %%xmm1  # Load 8 Cb           00 00 00 00 u3 u2 u1 u0   \n\
movq        (%5), %%xmm2  # Load 8 Cr           00 00 00 00 v3 v2 v1 v0   \n\
movdqu      (%2), %%xmm0  # Load 16 Y           y7 y6 y5 y4 y3 y2 y1 y0   \n\
movdqu      (%3), %%xmm3  # Load 16 Y           Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0   \n\
prefetchnta (%0)          # Tell CPU not to cache output UYVY data        \n\
prefetchnta (%1)          # Tell CPU not to cache output UYVY data        \n\
punpcklbw %%xmm2, %%xmm1  #                     v3 u3 v2 u2 v1 u1 v0 u0   \n\
movdqu    %%xmm1, %%xmm2  #                     v3 u3 v2 u2 v1 u1 v0 u0   \n\
punpcklbw %%xmm0, %%xmm2  #                     y3 v1 y2 u1 y1 v0 y0 u0   \n\
movdqu    %%xmm2, (%0)    # Store low UYVY                                \n\
movdqu    %%xmm1, %%xmm2  #                     u3 v3 u2 v2 u1 v1 u0 v0   \n\
punpckhbw %%xmm0, %%xmm2  #                     y3 v1 y2 u1 y1 v0 y0 u0   \n\
movdqu    %%xmm2, 16(%0)  # Store high UYVY                               \n\
movdqu    %%xmm1, %%xmm4  #                     u3 v3 u2 v2 u1 v1 u0 v0   \n\
punpcklbw %%xmm3, %%xmm4  #                     Y3 v1 Y2 u1 Y1 v0 Y0 u0   \n\
movdqu    %%xmm4, (%1)    # Store low UYVY                                \n\
punpckhbw %%xmm3, %%xmm1  #                     Y7 v3 Y6 u3 Y5 v2 Y4 u2   \n\
movdqu    %%xmm1, 16(%1)  # Store high UYVY                               \n\
330 331
"

332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460
#elif defined(HAVE_SSE2_INTRINSICS)

/* SSE2 intrinsics */

#include <emmintrin.h>

#define SSE2_CALL(SSE2_INSTRUCTIONS)            \
    do {                                        \
        __m128i xmm0, xmm1, xmm2, xmm3, xmm4;   \
        SSE2_INSTRUCTIONS                       \
        p_line1 += 32; p_line2 += 32;           \
        p_y1 += 16; p_y2 += 16;                 \
        p_u += 8; p_v += 8;                     \
    } while(0)

#define SSE2_END  _mm_sfence()

#define SSE2_YUV420_YUYV_ALIGNED                    \
    xmm1 = _mm_loadl_epi64((__m128i *)p_u);         \
    xmm2 = _mm_loadl_epi64((__m128i *)p_v);         \
    xmm0 = _mm_load_si128((__m128i *)p_y1);         \
    xmm3 = _mm_load_si128((__m128i *)p_y2);         \
    xmm1 = _mm_unpacklo_epi8(xmm1, xmm2);           \
    xmm2 = xmm0;                                    \
    xmm2 = _mm_unpacklo_epi8(xmm2, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line1), xmm2);    \
    xmm0 = _mm_unpackhi_epi8(xmm0, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line1+16), xmm0); \
    xmm4 = xmm3;                                    \
    xmm4 = _mm_unpacklo_epi8(xmm4, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line2), xmm4);    \
    xmm3 = _mm_unpackhi_epi8(xmm3, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line1+16), xmm3);

#define SSE2_YUV420_YUYV_UNALIGNED                  \
    xmm1 = _mm_loadl_epi64((__m128i *)p_u);         \
    xmm2 = _mm_loadl_epi64((__m128i *)p_v);         \
    xmm0 = _mm_load_si128((__m128i *)p_y1);         \
    xmm3 = _mm_load_si128((__m128i *)p_y2);         \
    _mm_prefetch(p_line1, _MM_HINT_NTA);            \
    _mm_prefetch(p_line2, _MM_HINT_NTA);            \
    xmm1 = _mm_unpacklo_epi8(xmm1, xmm2);           \
    xmm2 = xmm0;                                    \
    xmm2 = _mm_unpacklo_epi8(xmm2, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line1), xmm2);    \
    xmm0 = _mm_unpackhi_epi8(xmm0, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line1+16), xmm0); \
    xmm4 = xmm3;                                    \
    xmm4 = _mm_unpacklo_epi8(xmm4, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line2), xmm4);    \
    xmm3 = _mm_unpackhi_epi8(xmm3, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line1+16), xmm3);

#define SSE2_YUV420_YVYU_ALIGNED                    \
    xmm1 = _mm_loadl_epi64((__m128i *)p_v);         \
    xmm2 = _mm_loadl_epi64((__m128i *)p_u);         \
    xmm0 = _mm_load_si128((__m128i *)p_y1);         \
    xmm3 = _mm_load_si128((__m128i *)p_y2);         \
    xmm1 = _mm_unpacklo_epi8(xmm1, xmm2);           \
    xmm2 = xmm0;                                    \
    xmm2 = _mm_unpacklo_epi8(xmm2, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line1), xmm2);    \
    xmm0 = _mm_unpackhi_epi8(xmm0, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line1+16), xmm0); \
    xmm4 = xmm3;                                    \
    xmm4 = _mm_unpacklo_epi8(xmm4, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line2), xmm4);    \
    xmm3 = _mm_unpackhi_epi8(xmm3, xmm1);           \
    _mm_stream_si128((__m128i*)(p_line1+16), xmm3);

#define SSE2_YUV420_YVYU_UNALIGNED                  \
    xmm1 = _mm_loadl_epi64((__m128i *)p_v);         \
    xmm2 = _mm_loadl_epi64((__m128i *)p_u);         \
    xmm0 = _mm_load_si128((__m128i *)p_y1);         \
    xmm3 = _mm_load_si128((__m128i *)p_y2);         \
    _mm_prefetch(p_line1, _MM_HINT_NTA);            \
    _mm_prefetch(p_line2, _MM_HINT_NTA);            \
    xmm1 = _mm_unpacklo_epi8(xmm1, xmm2);           \
    xmm2 = xmm0;                                    \
    xmm2 = _mm_unpacklo_epi8(xmm2, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line1), xmm2);    \
    xmm0 = _mm_unpackhi_epi8(xmm0, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line1+16), xmm0); \
    xmm4 = xmm3;                                    \
    xmm4 = _mm_unpacklo_epi8(xmm4, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line2), xmm4);    \
    xmm3 = _mm_unpackhi_epi8(xmm3, xmm1);           \
    _mm_storeu_si128((__m128i*)(p_line1+16), xmm3);

#define SSE2_YUV420_UYVY_ALIGNED                    \
    xmm1 = _mm_loadl_epi64((__m128i *)p_u);         \
    xmm2 = _mm_loadl_epi64((__m128i *)p_v);         \
    xmm0 = _mm_load_si128((__m128i *)p_y1);         \
    xmm3 = _mm_load_si128((__m128i *)p_y2);         \
    xmm1 = _mm_unpacklo_epi8(xmm1, xmm2);           \
    xmm2 = xmm1;                                    \
    xmm2 = _mm_unpacklo_epi8(xmm2, xmm0);           \
    _mm_stream_si128((__m128i*)(p_line1), xmm2);    \
    xmm2 = xmm1;                                    \
    xmm2 = _mm_unpackhi_epi8(xmm2, xmm0);           \
    _mm_stream_si128((__m128i*)(p_line1+16), xmm2); \
    xmm4 = xmm1;                                    \
    xmm4 = _mm_unpacklo_epi8(xmm4, xmm3);           \
    _mm_stream_si128((__m128i*)(p_line2), xmm4);    \
    xmm1 = _mm_unpackhi_epi8(xmm1, xmm3);           \
    _mm_stream_si128((__m128i*)(p_line1+16), xmm1);

#define SSE2_YUV420_UYVY_UNALIGNED                  \
    xmm1 = _mm_loadl_epi64((__m128i *)p_u);         \
    xmm2 = _mm_loadl_epi64((__m128i *)p_v);         \
    xmm0 = _mm_load_si128((__m128i *)p_y1);         \
    xmm3 = _mm_load_si128((__m128i *)p_y2);         \
    _mm_prefetch(p_line1, _MM_HINT_NTA);            \
    _mm_prefetch(p_line2, _MM_HINT_NTA);            \
    xmm1 = _mm_unpacklo_epi8(xmm1, xmm2);           \
    xmm2 = xmm1;                                    \
    xmm2 = _mm_unpacklo_epi8(xmm2, xmm0);           \
    _mm_storeu_si128((__m128i*)(p_line1), xmm2);    \
    xmm2 = xmm1;                                    \
    xmm2 = _mm_unpackhi_epi8(xmm2, xmm0);           \
    _mm_storeu_si128((__m128i*)(p_line1+16), xmm2); \
    xmm4 = xmm1;                                    \
    xmm4 = _mm_unpacklo_epi8(xmm4, xmm3);           \
    _mm_storeu_si128((__m128i*)(p_line2), xmm4);    \
    xmm1 = _mm_unpackhi_epi8(xmm1, xmm3);           \
    _mm_storeu_si128((__m128i*)(p_line1+16), xmm1);

#endif

461 462 463
#endif

/* Used in both accelerated and C modules */
464 465 466 467 468 469 470 471

#define C_YUV420_YVYU( )                                                    \
    *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
    *(p_line1)++ =            *(p_line2)++ = *(p_v)++;                      \
    *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
    *(p_line1)++ =            *(p_line2)++ = *(p_u)++;                      \

#define C_YUV420_Y211( )                                                    \
472 473 474 475 476 477
    *(p_line1)++ = *(p_y1); p_y1 += 2;                                      \
    *(p_line2)++ = *(p_y2); p_y2 += 2;                                      \
    *(p_line1)++ = *(p_line2)++ = *(p_u) - 0x80; p_u += 2;                  \
    *(p_line1)++ = *(p_y1); p_y1 += 2;                                      \
    *(p_line2)++ = *(p_y2); p_y2 += 2;                                      \
    *(p_line1)++ = *(p_line2)++ = *(p_v) - 0x80; p_v += 2;                  \
478 479


480 481 482 483 484 485
#define C_YUV420_YUYV( )                                                    \
    *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
    *(p_line1)++ =            *(p_line2)++ = *(p_u)++;                      \
    *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
    *(p_line1)++ =            *(p_line2)++ = *(p_v)++;                      \

486 487 488 489 490 491
#define C_YUV420_UYVY( )                                                    \
    *(p_line1)++ =            *(p_line2)++ = *(p_u)++;                      \
    *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \
    *(p_line1)++ =            *(p_line2)++ = *(p_v)++;                      \
    *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++;                     \