pixel.c 21.8 KB
Newer Older
Laurent Aimar's avatar
Laurent Aimar committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*****************************************************************************
 * pixel.c: h264 encoder
 *****************************************************************************
 * Copyright (C) 2003 Laurent Aimar
 * $Id: pixel.c,v 1.1 2004/06/03 19:27:07 fenrir Exp $
 *
 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
 *
 * 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
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA.
 *****************************************************************************/

Loren Merritt's avatar
Loren Merritt committed
24
#include "common.h"
Laurent Aimar's avatar
Laurent Aimar committed
25

26
#ifdef HAVE_MMX
27
#   include "x86/pixel.h"
Laurent Aimar's avatar
Laurent Aimar committed
28
#endif
29
#ifdef ARCH_PPC
Laurent Aimar's avatar
Laurent Aimar committed
30 31
#   include "ppc/pixel.h"
#endif
32 33 34
#ifdef ARCH_UltraSparc
#   include "sparc/pixel.h"
#endif
Laurent Aimar's avatar
Laurent Aimar committed
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58


/****************************************************************************
 * pixel_sad_WxH
 ****************************************************************************/
#define PIXEL_SAD_C( name, lx, ly ) \
static int name( uint8_t *pix1, int i_stride_pix1,  \
                 uint8_t *pix2, int i_stride_pix2 ) \
{                                                   \
    int i_sum = 0;                                  \
    int x, y;                                       \
    for( y = 0; y < ly; y++ )                       \
    {                                               \
        for( x = 0; x < lx; x++ )                   \
        {                                           \
            i_sum += abs( pix1[x] - pix2[x] );      \
        }                                           \
        pix1 += i_stride_pix1;                      \
        pix2 += i_stride_pix2;                      \
    }                                               \
    return i_sum;                                   \
}


59 60 61 62 63 64 65
PIXEL_SAD_C( x264_pixel_sad_16x16, 16, 16 )
PIXEL_SAD_C( x264_pixel_sad_16x8,  16,  8 )
PIXEL_SAD_C( x264_pixel_sad_8x16,   8, 16 )
PIXEL_SAD_C( x264_pixel_sad_8x8,    8,  8 )
PIXEL_SAD_C( x264_pixel_sad_8x4,    8,  4 )
PIXEL_SAD_C( x264_pixel_sad_4x8,    4,  8 )
PIXEL_SAD_C( x264_pixel_sad_4x4,    4,  4 )
Laurent Aimar's avatar
Laurent Aimar committed
66

67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89

/****************************************************************************
 * pixel_ssd_WxH
 ****************************************************************************/
#define PIXEL_SSD_C( name, lx, ly ) \
static int name( uint8_t *pix1, int i_stride_pix1,  \
                 uint8_t *pix2, int i_stride_pix2 ) \
{                                                   \
    int i_sum = 0;                                  \
    int x, y;                                       \
    for( y = 0; y < ly; y++ )                       \
    {                                               \
        for( x = 0; x < lx; x++ )                   \
        {                                           \
            int d = pix1[x] - pix2[x];              \
            i_sum += d*d;                           \
        }                                           \
        pix1 += i_stride_pix1;                      \
        pix2 += i_stride_pix2;                      \
    }                                               \
    return i_sum;                                   \
}

90 91 92 93 94 95 96
PIXEL_SSD_C( x264_pixel_ssd_16x16, 16, 16 )
PIXEL_SSD_C( x264_pixel_ssd_16x8,  16,  8 )
PIXEL_SSD_C( x264_pixel_ssd_8x16,   8, 16 )
PIXEL_SSD_C( x264_pixel_ssd_8x8,    8,  8 )
PIXEL_SSD_C( x264_pixel_ssd_8x4,    8,  4 )
PIXEL_SSD_C( x264_pixel_ssd_4x8,    4,  8 )
PIXEL_SSD_C( x264_pixel_ssd_4x4,    4,  4 )
97

Loren Merritt's avatar
Loren Merritt committed
98 99 100 101
int64_t x264_pixel_ssd_wxh( x264_pixel_function_t *pf, uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2, int i_width, int i_height )
{
    int64_t i_ssd = 0;
    int x, y;
Loren Merritt's avatar
Loren Merritt committed
102
    int align = !(((long)pix1 | (long)pix2 | i_pix1 | i_pix2) & 15);
Loren Merritt's avatar
Loren Merritt committed
103 104 105 106 107

#define SSD(size) i_ssd += pf->ssd[size]( pix1 + y*i_pix1 + x, i_pix1, \
                                          pix2 + y*i_pix2 + x, i_pix2 );
    for( y = 0; y < i_height-15; y += 16 )
    {
Loren Merritt's avatar
Loren Merritt committed
108 109 110 111 112
        x = 0;
        if( align )
            for( ; x < i_width-15; x += 16 )
                SSD(PIXEL_16x16);
        for( ; x < i_width-7; x += 8 )
Loren Merritt's avatar
Loren Merritt committed
113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
            SSD(PIXEL_8x16);
    }
    if( y < i_height-7 )
        for( x = 0; x < i_width-7; x += 8 )
            SSD(PIXEL_8x8);
#undef SSD

#define SSD1 { int d = pix1[y*i_pix1+x] - pix2[y*i_pix2+x]; i_ssd += d*d; }
    if( i_width % 8 != 0 )
    {
        for( y = 0; y < (i_height & ~7); y++ )
            for( x = i_width & ~7; x < i_width; x++ )
                SSD1;
    }
    if( i_height % 8 != 0 )
    {
        for( y = i_height & ~7; y < i_height; y++ )
            for( x = 0; x < i_width; x++ )
                SSD1;
    }
#undef SSD1

    return i_ssd;
}

138

139 140
static inline void pixel_sub_wxh( int16_t *diff, int i_size,
                                  uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2 )
Laurent Aimar's avatar
Laurent Aimar committed
141 142
{
    int y, x;
143
    for( y = 0; y < i_size; y++ )
Laurent Aimar's avatar
Laurent Aimar committed
144
    {
145
        for( x = 0; x < i_size; x++ )
Laurent Aimar's avatar
Laurent Aimar committed
146
        {
147
            diff[x + y*i_size] = pix1[x] - pix2[x];
Laurent Aimar's avatar
Laurent Aimar committed
148 149 150 151 152 153
        }
        pix1 += i_pix1;
        pix2 += i_pix2;
    }
}

154 155 156 157

/****************************************************************************
 * pixel_satd_WxH: sum of 4x4 Hadamard transformed differences
 ****************************************************************************/
Laurent Aimar's avatar
Laurent Aimar committed
158 159 160 161 162 163 164 165 166 167 168 169 170
static int pixel_satd_wxh( uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2, int i_width, int i_height )
{
    int16_t tmp[4][4];
    int16_t diff[4][4];
    int x, y;
    int i_satd = 0;

    for( y = 0; y < i_height; y += 4 )
    {
        for( x = 0; x < i_width; x += 4 )
        {
            int d;

171
            pixel_sub_wxh( (int16_t*)diff, 4, &pix1[x], i_pix1, &pix2[x], i_pix2 );
Laurent Aimar's avatar
Laurent Aimar committed
172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209

            for( d = 0; d < 4; d++ )
            {
                int s01, s23;
                int d01, d23;

                s01 = diff[d][0] + diff[d][1]; s23 = diff[d][2] + diff[d][3];
                d01 = diff[d][0] - diff[d][1]; d23 = diff[d][2] - diff[d][3];

                tmp[d][0] = s01 + s23;
                tmp[d][1] = s01 - s23;
                tmp[d][2] = d01 - d23;
                tmp[d][3] = d01 + d23;
            }
            for( d = 0; d < 4; d++ )
            {
                int s01, s23;
                int d01, d23;

                s01 = tmp[0][d] + tmp[1][d]; s23 = tmp[2][d] + tmp[3][d];
                d01 = tmp[0][d] - tmp[1][d]; d23 = tmp[2][d] - tmp[3][d];

                i_satd += abs( s01 + s23 ) + abs( s01 - s23 ) + abs( d01 - d23 ) + abs( d01 + d23 );
            }

        }
        pix1 += 4 * i_pix1;
        pix2 += 4 * i_pix2;
    }

    return i_satd / 2;
}
#define PIXEL_SATD_C( name, width, height ) \
static int name( uint8_t *pix1, int i_stride_pix1, \
                 uint8_t *pix2, int i_stride_pix2 ) \
{ \
    return pixel_satd_wxh( pix1, i_stride_pix1, pix2, i_stride_pix2, width, height ); \
}
210 211 212 213 214 215 216
PIXEL_SATD_C( x264_pixel_satd_16x16, 16, 16 )
PIXEL_SATD_C( x264_pixel_satd_16x8,  16, 8 )
PIXEL_SATD_C( x264_pixel_satd_8x16,  8, 16 )
PIXEL_SATD_C( x264_pixel_satd_8x8,   8, 8 )
PIXEL_SATD_C( x264_pixel_satd_8x4,   8, 4 )
PIXEL_SATD_C( x264_pixel_satd_4x8,   4, 8 )
PIXEL_SATD_C( x264_pixel_satd_4x4,   4, 4 )
Laurent Aimar's avatar
Laurent Aimar committed
217 218


219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
/****************************************************************************
 * pixel_sa8d_WxH: sum of 8x8 Hadamard transformed differences
 ****************************************************************************/
#define SA8D_1D {\
    const int a0 = SRC(0) + SRC(4);\
    const int a4 = SRC(0) - SRC(4);\
    const int a1 = SRC(1) + SRC(5);\
    const int a5 = SRC(1) - SRC(5);\
    const int a2 = SRC(2) + SRC(6);\
    const int a6 = SRC(2) - SRC(6);\
    const int a3 = SRC(3) + SRC(7);\
    const int a7 = SRC(3) - SRC(7);\
    const int b0 = a0 + a2;\
    const int b2 = a0 - a2;\
    const int b1 = a1 + a3;\
    const int b3 = a1 - a3;\
    const int b4 = a4 + a6;\
    const int b6 = a4 - a6;\
    const int b5 = a5 + a7;\
    const int b7 = a5 - a7;\
    DST(0, b0 + b1);\
    DST(1, b0 - b1);\
    DST(2, b2 + b3);\
    DST(3, b2 - b3);\
    DST(4, b4 + b5);\
    DST(5, b4 - b5);\
    DST(6, b6 + b7);\
    DST(7, b6 - b7);\
}

static inline int pixel_sa8d_wxh( uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2,
                                  int i_width, int i_height )
{
    int16_t diff[8][8];
    int i_satd = 0;
    int x, y;

    for( y = 0; y < i_height; y += 8 )
    {
        for( x = 0; x < i_width; x += 8 )
        {
            int i;
            pixel_sub_wxh( (int16_t*)diff, 8, pix1+x, i_pix1, pix2+x, i_pix2 );

#define SRC(x)     diff[i][x]
#define DST(x,rhs) diff[i][x] = (rhs)
            for( i = 0; i < 8; i++ )
                SA8D_1D
#undef SRC
#undef DST

#define SRC(x)     diff[x][i]
#define DST(x,rhs) i_satd += abs(rhs)
            for( i = 0; i < 8; i++ )
                SA8D_1D
#undef SRC
#undef DST
        }
        pix1 += 8 * i_pix1;
        pix2 += 8 * i_pix2;
    }

    return i_satd;
}

#define PIXEL_SA8D_C( width, height ) \
285 286
static int x264_pixel_sa8d_##width##x##height( uint8_t *pix1, int i_stride_pix1, \
                                               uint8_t *pix2, int i_stride_pix2 ) \
287 288 289 290 291 292 293 294
{ \
    return ( pixel_sa8d_wxh( pix1, i_stride_pix1, pix2, i_stride_pix2, width, height ) + 2 ) >> 2; \
}
PIXEL_SA8D_C( 16, 16 )
PIXEL_SA8D_C( 16, 8 )
PIXEL_SA8D_C( 8, 16 )
PIXEL_SA8D_C( 8, 8 )

Loren Merritt's avatar
Loren Merritt committed
295 296 297
/****************************************************************************
 * pixel_sad_x4
 ****************************************************************************/
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326
#define SAD_X( size ) \
static void x264_pixel_sad_x3_##size( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2, int i_stride, int scores[3] )\
{\
    scores[0] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix0, i_stride );\
    scores[1] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix1, i_stride );\
    scores[2] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix2, i_stride );\
}\
static void x264_pixel_sad_x4_##size( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2, uint8_t *pix3, int i_stride, int scores[4] )\
{\
    scores[0] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix0, i_stride );\
    scores[1] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix1, i_stride );\
    scores[2] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix2, i_stride );\
    scores[3] = x264_pixel_sad_##size( fenc, FENC_STRIDE, pix3, i_stride );\
}

SAD_X( 16x16 )
SAD_X( 16x8 )
SAD_X( 8x16 )
SAD_X( 8x8 )
SAD_X( 8x4 )
SAD_X( 4x8 )
SAD_X( 4x4 )

#ifdef ARCH_UltraSparc
SAD_X( 16x16_vis )
SAD_X( 16x8_vis )
SAD_X( 8x16_vis )
SAD_X( 8x8_vis )
#endif
327

328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
/****************************************************************************
 * pixel_satd_x4
 * no faster than single satd, but needed for satd to be a drop-in replacement for sad
 ****************************************************************************/

#define SATD_X( size, cpu ) \
static void x264_pixel_satd_x3_##size##cpu( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2, int i_stride, int scores[3] )\
{\
    scores[0] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix0, i_stride );\
    scores[1] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix1, i_stride );\
    scores[2] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix2, i_stride );\
}\
static void x264_pixel_satd_x4_##size##cpu( uint8_t *fenc, uint8_t *pix0, uint8_t *pix1, uint8_t *pix2, uint8_t *pix3, int i_stride, int scores[4] )\
{\
    scores[0] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix0, i_stride );\
    scores[1] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix1, i_stride );\
    scores[2] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix2, i_stride );\
    scores[3] = x264_pixel_satd_##size##cpu( fenc, FENC_STRIDE, pix3, i_stride );\
}
#define SATD_X_DECL5( cpu )\
SATD_X( 16x16, cpu )\
SATD_X( 16x8, cpu )\
SATD_X( 8x16, cpu )\
SATD_X( 8x8, cpu )\
SATD_X( 8x4, cpu )
#define SATD_X_DECL7( cpu )\
SATD_X_DECL5( cpu )\
SATD_X( 4x8, cpu )\
SATD_X( 4x4, cpu )

SATD_X_DECL7()
#ifdef HAVE_MMX
SATD_X_DECL7( _mmxext )
SATD_X_DECL5( _sse2 )
Fiona Glaser's avatar
Fiona Glaser committed
362
SATD_X_DECL7( _ssse3 )
363 364
#endif

Loren Merritt's avatar
Loren Merritt committed
365 366 367
/****************************************************************************
 * structural similarity metric
 ****************************************************************************/
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
static void ssim_4x4x2_core( const uint8_t *pix1, int stride1,
                             const uint8_t *pix2, int stride2,
                             int sums[2][4])
{
    int x, y, z;
    for(z=0; z<2; z++)
    {
        uint32_t s1=0, s2=0, ss=0, s12=0;
        for(y=0; y<4; y++)
            for(x=0; x<4; x++)
            {
                int a = pix1[x+y*stride1];
                int b = pix2[x+y*stride2];
                s1  += a;
                s2  += b;
                ss  += a*a;
                ss  += b*b;
                s12 += a*b;
            }
        sums[z][0] = s1;
        sums[z][1] = s2;
        sums[z][2] = ss;
        sums[z][3] = s12;
        pix1 += 4;
        pix2 += 4;
    }
}

static float ssim_end1( int s1, int s2, int ss, int s12 )
{
    static const int ssim_c1 = (int)(.01*.01*255*255*64 + .5);
    static const int ssim_c2 = (int)(.03*.03*255*255*64*63 + .5);
    int vars = ss*64 - s1*s1 - s2*s2;
    int covar = s12*64 - s1*s2;
    return (float)(2*s1*s2 + ssim_c1) * (float)(2*covar + ssim_c2)\
           / ((float)(s1*s1 + s2*s2 + ssim_c1) * (float)(vars + ssim_c2));
}

static float ssim_end4( int sum0[5][4], int sum1[5][4], int width )
{
    int i;
    float ssim = 0.0;
    for( i = 0; i < width; i++ )
        ssim += ssim_end1( sum0[i][0] + sum0[i+1][0] + sum1[i][0] + sum1[i+1][0],
                           sum0[i][1] + sum0[i+1][1] + sum1[i][1] + sum1[i+1][1],
                           sum0[i][2] + sum0[i+1][2] + sum1[i][2] + sum1[i+1][2],
                           sum0[i][3] + sum0[i+1][3] + sum1[i][3] + sum1[i+1][3] );
    return ssim;
}

float x264_pixel_ssim_wxh( x264_pixel_function_t *pf,
                           uint8_t *pix1, int stride1,
                           uint8_t *pix2, int stride2,
                           int width, int height )
{
    int x, y, z;
    float ssim = 0.0;
425 426
    int (*sum0)[4] = x264_malloc(4 * (width/4+3) * sizeof(int));
    int (*sum1)[4] = x264_malloc(4 * (width/4+3) * sizeof(int));
427 428 429 430 431 432 433 434 435 436 437 438 439 440
    width >>= 2;
    height >>= 2;
    z = 0;
    for( y = 1; y < height; y++ )
    {
        for( ; z <= y; z++ )
        {
            XCHG( void*, sum0, sum1 );
            for( x = 0; x < width; x+=2 )
                pf->ssim_4x4x2_core( &pix1[4*(x+z*stride1)], stride1, &pix2[4*(x+z*stride2)], stride2, &sum0[x] );
        }
        for( x = 0; x < width-1; x += 4 )
            ssim += pf->ssim_end4( sum0+x, sum1+x, X264_MIN(4,width-x-1) );
    }
441 442
    x264_free(sum0);
    x264_free(sum1);
443 444 445 446
    return ssim / ((width-1) * (height-1));
}


Loren Merritt's avatar
Loren Merritt committed
447 448 449
/****************************************************************************
 * successive elimination
 ****************************************************************************/
450 451
static int x264_pixel_ads4( int enc_dc[4], uint16_t *sums, int delta,
                            uint16_t *cost_mvx, int16_t *mvs, int width, int thresh )
Loren Merritt's avatar
Loren Merritt committed
452
{
453
    int nmv=0, i;
Loren Merritt's avatar
Loren Merritt committed
454
    for( i=0; i<width; i++, sums++ )
455 456 457 458 459 460 461 462 463 464
    {
        int ads = abs( enc_dc[0] - sums[0] )
                + abs( enc_dc[1] - sums[8] )
                + abs( enc_dc[2] - sums[delta] )
                + abs( enc_dc[3] - sums[delta+8] )
                + cost_mvx[i];
        if( ads < thresh )
            mvs[nmv++] = i;
    }
    return nmv;
Loren Merritt's avatar
Loren Merritt committed
465 466
}

467 468
static int x264_pixel_ads2( int enc_dc[2], uint16_t *sums, int delta,
                            uint16_t *cost_mvx, int16_t *mvs, int width, int thresh )
Loren Merritt's avatar
Loren Merritt committed
469
{
470
    int nmv=0, i;
Loren Merritt's avatar
Loren Merritt committed
471
    for( i=0; i<width; i++, sums++ )
472 473 474 475 476 477 478 479
    {
        int ads = abs( enc_dc[0] - sums[0] )
                + abs( enc_dc[1] - sums[delta] )
                + cost_mvx[i];
        if( ads < thresh )
            mvs[nmv++] = i;
    }
    return nmv;
Loren Merritt's avatar
Loren Merritt committed
480 481
}

482 483
static int x264_pixel_ads1( int enc_dc[1], uint16_t *sums, int delta,
                            uint16_t *cost_mvx, int16_t *mvs, int width, int thresh )
Loren Merritt's avatar
Loren Merritt committed
484
{
485
    int nmv=0, i;
Loren Merritt's avatar
Loren Merritt committed
486
    for( i=0; i<width; i++, sums++ )
487 488 489 490 491 492 493
    {
        int ads = abs( enc_dc[0] - sums[0] )
                + cost_mvx[i];
        if( ads < thresh )
            mvs[nmv++] = i;
    }
    return nmv;
Loren Merritt's avatar
Loren Merritt committed
494 495 496
}


Laurent Aimar's avatar
Laurent Aimar committed
497 498 499 500 501
/****************************************************************************
 * x264_pixel_init:
 ****************************************************************************/
void x264_pixel_init( int cpu, x264_pixel_function_t *pixf )
{
502 503
    memset( pixf, 0, sizeof(*pixf) );

504
#define INIT2( name, cpu ) \
505
    pixf->name[PIXEL_16x16] = x264_pixel_##name##_16x16##cpu;\
506 507 508
    pixf->name[PIXEL_16x8]  = x264_pixel_##name##_16x8##cpu;
#define INIT4( name, cpu ) \
    INIT2( name, cpu ) \
509
    pixf->name[PIXEL_8x16]  = x264_pixel_##name##_8x16##cpu;\
510 511 512 513 514 515
    pixf->name[PIXEL_8x8]   = x264_pixel_##name##_8x8##cpu;
#define INIT5( name, cpu ) \
    INIT4( name, cpu ) \
    pixf->name[PIXEL_8x4]   = x264_pixel_##name##_8x4##cpu;
#define INIT7( name, cpu ) \
    INIT5( name, cpu ) \
516 517 518
    pixf->name[PIXEL_4x8]   = x264_pixel_##name##_4x8##cpu;\
    pixf->name[PIXEL_4x4]   = x264_pixel_##name##_4x4##cpu;

519 520 521 522 523
#define INIT_ADS( cpu ) \
    pixf->ads[PIXEL_16x16] = x264_pixel_ads4##cpu;\
    pixf->ads[PIXEL_16x8] = x264_pixel_ads2##cpu;\
    pixf->ads[PIXEL_8x8] = x264_pixel_ads1##cpu;

524 525 526 527 528
    INIT7( sad, );
    INIT7( sad_x3, );
    INIT7( sad_x4, );
    INIT7( ssd, );
    INIT7( satd, );
529 530
    INIT7( satd_x3, );
    INIT7( satd_x4, );
531
    INIT4( sa8d, );
532
    INIT_ADS( );
533

534 535
    pixf->ssim_4x4x2_core = ssim_4x4x2_core;
    pixf->ssim_end4 = ssim_end4;
536

537
#ifdef HAVE_MMX
538 539
    if( cpu&X264_CPU_MMX )
    {
540
        INIT7( ssd, _mmx );
541 542
    }

Laurent Aimar's avatar
Laurent Aimar committed
543 544
    if( cpu&X264_CPU_MMXEXT )
    {
545 546 547 548
        INIT7( sad, _mmxext );
        INIT7( sad_x3, _mmxext );
        INIT7( sad_x4, _mmxext );
        INIT7( satd, _mmxext );
549 550
        INIT7( satd_x3, _mmxext );
        INIT7( satd_x4, _mmxext );
551
        INIT_ADS( _mmxext );
Loren Merritt's avatar
Loren Merritt committed
552

553 554 555
#ifdef ARCH_X86
        pixf->sa8d[PIXEL_16x16] = x264_pixel_sa8d_16x16_mmxext;
        pixf->sa8d[PIXEL_8x8]   = x264_pixel_sa8d_8x8_mmxext;
556
        pixf->intra_sa8d_x3_8x8 = x264_intra_sa8d_x3_8x8_mmxext;
557
        pixf->ssim_4x4x2_core  = x264_pixel_ssim_4x4x2_core_mmxext;
558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584

        if( cpu&X264_CPU_CACHELINE_SPLIT )
        {
            if( cpu&X264_CPU_CACHELINE_32 )
            {
                INIT5( sad, _cache32_mmxext );
                INIT4( sad_x3, _cache32_mmxext );
                INIT4( sad_x4, _cache32_mmxext );
            }
            else
            {
                INIT5( sad, _cache64_mmxext );
                INIT4( sad_x3, _cache64_mmxext );
                INIT4( sad_x4, _cache64_mmxext );
            }
        }
#else
        if( cpu&X264_CPU_CACHELINE_SPLIT )
        {
            pixf->sad[PIXEL_8x16] = x264_pixel_sad_8x16_cache64_mmxext;
            pixf->sad[PIXEL_8x8]  = x264_pixel_sad_8x8_cache64_mmxext;
            pixf->sad[PIXEL_8x4]  = x264_pixel_sad_8x4_cache64_mmxext;
            pixf->sad_x3[PIXEL_8x16] = x264_pixel_sad_x3_8x16_cache64_mmxext;
            pixf->sad_x3[PIXEL_8x8]  = x264_pixel_sad_x3_8x8_cache64_mmxext;
            pixf->sad_x4[PIXEL_8x16] = x264_pixel_sad_x4_8x16_cache64_mmxext;
            pixf->sad_x4[PIXEL_8x8]  = x264_pixel_sad_x4_8x8_cache64_mmxext;
        }
585
#endif
586 587 588
        pixf->intra_satd_x3_16x16 = x264_intra_satd_x3_16x16_mmxext;
        pixf->intra_satd_x3_8x8c  = x264_intra_satd_x3_8x8c_mmxext;
        pixf->intra_satd_x3_4x4   = x264_intra_satd_x3_4x4_mmxext;
Laurent Aimar's avatar
Laurent Aimar committed
589
    }
590 591 592 593

    // disable on AMD processors since it is slower
    if( (cpu&X264_CPU_SSE2) && !(cpu&X264_CPU_3DNOW) )
    {
594 595 596 597
        INIT2( sad, _sse2 );
        INIT2( sad_x3, _sse2 );
        INIT2( sad_x4, _sse2 );
        INIT5( satd, _sse2 );
598 599
        INIT5( satd_x3, _sse2 );
        INIT5( satd_x4, _sse2 );
600
        INIT_ADS( _sse2 );
601 602 603 604 605 606 607 608 609

#ifdef ARCH_X86
        if( cpu&X264_CPU_CACHELINE_SPLIT )
        {
            INIT2( sad, _cache64_sse2 );
            INIT2( sad_x3, _cache64_sse2 );
            INIT2( sad_x4, _cache64_sse2 );
        }
#endif
610 611 612 613
    }
    // these are faster on both Intel and AMD
    if( cpu&X264_CPU_SSE2 )
    {
Loren Merritt's avatar
Loren Merritt committed
614
        INIT5( ssd, _sse2 );
615 616
        pixf->ssim_4x4x2_core  = x264_pixel_ssim_4x4x2_core_sse2;
        pixf->ssim_end4        = x264_pixel_ssim_end4_sse2;
617 618
        pixf->sa8d[PIXEL_16x16] = x264_pixel_sa8d_16x16_sse2;
        pixf->sa8d[PIXEL_8x8]   = x264_pixel_sa8d_8x8_sse2;
619
#ifdef ARCH_X86_64
620
        pixf->intra_sa8d_x3_8x8 = x264_intra_sa8d_x3_8x8_sse2;
621 622 623
#endif
    }

624 625 626 627 628 629 630
    if( (cpu&X264_CPU_SSE3) && (cpu&X264_CPU_CACHELINE_SPLIT) )
    {
        INIT2( sad, _sse3 );
        INIT2( sad_x3, _sse3 );
        INIT2( sad_x4, _sse3 );
    }

631 632
    if( cpu&X264_CPU_SSSE3 )
    {
Fiona Glaser's avatar
Fiona Glaser committed
633 634 635
        INIT7( satd, _ssse3 );
        INIT7( satd_x3, _ssse3 );
        INIT7( satd_x4, _ssse3 );
636
        INIT_ADS( _ssse3 );
637 638
        pixf->sa8d[PIXEL_16x16]= x264_pixel_sa8d_16x16_ssse3;
        pixf->sa8d[PIXEL_8x8]  = x264_pixel_sa8d_8x8_ssse3;
639 640 641 642 643
        pixf->intra_satd_x3_16x16 = x264_intra_satd_x3_16x16_ssse3;
        pixf->intra_satd_x3_8x8c  = x264_intra_satd_x3_8x8c_ssse3;
        pixf->intra_satd_x3_4x4   = x264_intra_satd_x3_4x4_ssse3;
#ifdef ARCH_X86_64
        pixf->intra_sa8d_x3_8x8 = x264_intra_sa8d_x3_8x8_ssse3;
Loren Merritt's avatar
Loren Merritt committed
644
#endif
645 646 647 648 649 650
        if( cpu&X264_CPU_CACHELINE_SPLIT )
        {
            INIT2( sad, _cache64_ssse3 );
            INIT2( sad_x3, _cache64_ssse3 );
            INIT2( sad_x4, _cache64_ssse3 );
        }
651
    }
652
#endif //HAVE_MMX
653

654
#ifdef ARCH_PPC
Laurent Aimar's avatar
Laurent Aimar committed
655 656 657 658 659
    if( cpu&X264_CPU_ALTIVEC )
    {
        x264_pixel_altivec_init( pixf );
    }
#endif
660
#ifdef ARCH_UltraSparc
661 662 663
    INIT4( sad, _vis );
    INIT4( sad_x3, _vis );
    INIT4( sad_x4, _vis );
664
#endif
Loren Merritt's avatar
Loren Merritt committed
665 666 667 668 669

    pixf->ads[PIXEL_8x16] =
    pixf->ads[PIXEL_8x4] =
    pixf->ads[PIXEL_4x8] = pixf->ads[PIXEL_16x8];
    pixf->ads[PIXEL_4x4] = pixf->ads[PIXEL_8x8];
Laurent Aimar's avatar
Laurent Aimar committed
670 671
}