pixel.c 13.5 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.
 *****************************************************************************/

24 25 26 27 28
#ifdef HAVE_STDINT_H
#include <stdint.h>
#else
#include <inttypes.h>
#endif
Laurent Aimar's avatar
Laurent Aimar committed
29 30
#include <stdlib.h>
#include <string.h>
31
#include <stdarg.h>
Laurent Aimar's avatar
Laurent Aimar committed
32

33
#include "x264.h"
Laurent Aimar's avatar
Laurent Aimar committed
34
#include "pixel.h"
35
#include "clip1.h"
Laurent Aimar's avatar
Laurent Aimar committed
36 37 38 39

#ifdef HAVE_MMXEXT
#   include "i386/pixel.h"
#endif
40
#ifdef ARCH_PPC
Laurent Aimar's avatar
Laurent Aimar committed
41 42
#   include "ppc/pixel.h"
#endif
43 44 45
#ifdef ARCH_UltraSparc
#   include "sparc/pixel.h"
#endif
Laurent Aimar's avatar
Laurent Aimar committed
46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77


/****************************************************************************
 * 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;                                   \
}


PIXEL_SAD_C( pixel_sad_16x16, 16, 16 )
PIXEL_SAD_C( pixel_sad_16x8,  16,  8 )
PIXEL_SAD_C( pixel_sad_8x16,   8, 16 )
PIXEL_SAD_C( pixel_sad_8x8,    8,  8 )
PIXEL_SAD_C( pixel_sad_8x4,    8,  4 )
PIXEL_SAD_C( pixel_sad_4x8,    4,  8 )
PIXEL_SAD_C( pixel_sad_4x4,    4,  4 )

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108

/****************************************************************************
 * 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;                                   \
}

PIXEL_SSD_C( pixel_ssd_16x16, 16, 16 )
PIXEL_SSD_C( pixel_ssd_16x8,  16,  8 )
PIXEL_SSD_C( pixel_ssd_8x16,   8, 16 )
PIXEL_SSD_C( pixel_ssd_8x8,    8,  8 )
PIXEL_SSD_C( pixel_ssd_8x4,    8,  4 )
PIXEL_SSD_C( pixel_ssd_4x8,    4,  8 )
PIXEL_SSD_C( pixel_ssd_4x4,    4,  4 )

Loren Merritt's avatar
Loren Merritt committed
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
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;

#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 )
    {
        for( x = 0; x < i_width-15; x += 16 )
            SSD(PIXEL_16x16);
        if( x < i_width-7 )
            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;
}

146

147 148
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
149 150
{
    int y, x;
151
    for( y = 0; y < i_size; y++ )
Laurent Aimar's avatar
Laurent Aimar committed
152
    {
153
        for( x = 0; x < i_size; x++ )
Laurent Aimar's avatar
Laurent Aimar committed
154
        {
155
            diff[x + y*i_size] = pix1[x] - pix2[x];
Laurent Aimar's avatar
Laurent Aimar committed
156 157 158 159 160 161
        }
        pix1 += i_pix1;
        pix2 += i_pix2;
    }
}

162 163 164 165

/****************************************************************************
 * pixel_satd_WxH: sum of 4x4 Hadamard transformed differences
 ****************************************************************************/
Laurent Aimar's avatar
Laurent Aimar committed
166 167 168 169 170 171 172 173 174 175 176 177 178
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;

179
            pixel_sub_wxh( (int16_t*)diff, 4, &pix1[x], i_pix1, &pix2[x], i_pix2 );
Laurent Aimar's avatar
Laurent Aimar committed
180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226

            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 ); \
}
PIXEL_SATD_C( pixel_satd_16x16, 16, 16 )
PIXEL_SATD_C( pixel_satd_16x8,  16, 8 )
PIXEL_SATD_C( pixel_satd_8x16,  8, 16 )
PIXEL_SATD_C( pixel_satd_8x8,   8, 8 )
PIXEL_SATD_C( pixel_satd_8x4,   8, 4 )
PIXEL_SATD_C( pixel_satd_4x8,   4, 8 )
PIXEL_SATD_C( pixel_satd_4x4,   4, 4 )


227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303
/****************************************************************************
 * 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 ) \
static int pixel_sa8d_##width##x##height( uint8_t *pix1, int i_stride_pix1, \
                 uint8_t *pix2, int i_stride_pix2 ) \
{ \
    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 )


Laurent Aimar's avatar
Laurent Aimar committed
304 305 306 307 308 309 310 311 312 313 314 315 316
/****************************************************************************
 * x264_pixel_init:
 ****************************************************************************/
void x264_pixel_init( int cpu, x264_pixel_function_t *pixf )
{
    pixf->sad[PIXEL_16x16] = pixel_sad_16x16;
    pixf->sad[PIXEL_16x8]  = pixel_sad_16x8;
    pixf->sad[PIXEL_8x16]  = pixel_sad_8x16;
    pixf->sad[PIXEL_8x8]   = pixel_sad_8x8;
    pixf->sad[PIXEL_8x4]   = pixel_sad_8x4;
    pixf->sad[PIXEL_4x8]   = pixel_sad_4x8;
    pixf->sad[PIXEL_4x4]   = pixel_sad_4x4;

317 318 319 320 321 322 323 324
    pixf->ssd[PIXEL_16x16] = pixel_ssd_16x16;
    pixf->ssd[PIXEL_16x8]  = pixel_ssd_16x8;
    pixf->ssd[PIXEL_8x16]  = pixel_ssd_8x16;
    pixf->ssd[PIXEL_8x8]   = pixel_ssd_8x8;
    pixf->ssd[PIXEL_8x4]   = pixel_ssd_8x4;
    pixf->ssd[PIXEL_4x8]   = pixel_ssd_4x8;
    pixf->ssd[PIXEL_4x4]   = pixel_ssd_4x4;

Laurent Aimar's avatar
Laurent Aimar committed
325 326 327 328 329 330 331 332
    pixf->satd[PIXEL_16x16]= pixel_satd_16x16;
    pixf->satd[PIXEL_16x8] = pixel_satd_16x8;
    pixf->satd[PIXEL_8x16] = pixel_satd_8x16;
    pixf->satd[PIXEL_8x8]  = pixel_satd_8x8;
    pixf->satd[PIXEL_8x4]  = pixel_satd_8x4;
    pixf->satd[PIXEL_4x8]  = pixel_satd_4x8;
    pixf->satd[PIXEL_4x4]  = pixel_satd_4x4;

333 334 335 336 337
    pixf->sa8d[PIXEL_16x16]= pixel_sa8d_16x16;
    pixf->sa8d[PIXEL_16x8] = pixel_sa8d_16x8;
    pixf->sa8d[PIXEL_8x16] = pixel_sa8d_8x16;
    pixf->sa8d[PIXEL_8x8]  = pixel_sa8d_8x8;

Laurent Aimar's avatar
Laurent Aimar committed
338 339 340 341 342 343 344 345 346 347 348
#ifdef HAVE_MMXEXT
    if( cpu&X264_CPU_MMXEXT )
    {
        pixf->sad[PIXEL_16x16] = x264_pixel_sad_16x16_mmxext;
        pixf->sad[PIXEL_16x8 ] = x264_pixel_sad_16x8_mmxext;
        pixf->sad[PIXEL_8x16 ] = x264_pixel_sad_8x16_mmxext;
        pixf->sad[PIXEL_8x8  ] = x264_pixel_sad_8x8_mmxext;
        pixf->sad[PIXEL_8x4  ] = x264_pixel_sad_8x4_mmxext;
        pixf->sad[PIXEL_4x8  ] = x264_pixel_sad_4x8_mmxext;
        pixf->sad[PIXEL_4x4]   = x264_pixel_sad_4x4_mmxext;

349 350 351 352
        pixf->sad_pde[PIXEL_16x16] = x264_pixel_sad_pde_16x16_mmxext;
        pixf->sad_pde[PIXEL_16x8 ] = x264_pixel_sad_pde_16x8_mmxext;
        pixf->sad_pde[PIXEL_8x16 ] = x264_pixel_sad_pde_8x16_mmxext;

353 354 355 356 357 358 359 360
        pixf->ssd[PIXEL_16x16] = x264_pixel_ssd_16x16_mmxext;
        pixf->ssd[PIXEL_16x8]  = x264_pixel_ssd_16x8_mmxext;
        pixf->ssd[PIXEL_8x16]  = x264_pixel_ssd_8x16_mmxext;
        pixf->ssd[PIXEL_8x8]   = x264_pixel_ssd_8x8_mmxext;
        pixf->ssd[PIXEL_8x4]   = x264_pixel_ssd_8x4_mmxext;
        pixf->ssd[PIXEL_4x8]   = x264_pixel_ssd_4x8_mmxext;
        pixf->ssd[PIXEL_4x4]   = x264_pixel_ssd_4x4_mmxext;
  
Laurent Aimar's avatar
Laurent Aimar committed
361 362 363 364 365 366 367 368 369
        pixf->satd[PIXEL_16x16]= x264_pixel_satd_16x16_mmxext;
        pixf->satd[PIXEL_16x8] = x264_pixel_satd_16x8_mmxext;
        pixf->satd[PIXEL_8x16] = x264_pixel_satd_8x16_mmxext;
        pixf->satd[PIXEL_8x8]  = x264_pixel_satd_8x8_mmxext;
        pixf->satd[PIXEL_8x4]  = x264_pixel_satd_8x4_mmxext;
        pixf->satd[PIXEL_4x8]  = x264_pixel_satd_4x8_mmxext;
        pixf->satd[PIXEL_4x4]  = x264_pixel_satd_4x4_mmxext;
    }
#endif
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391

#ifdef HAVE_SSE2
    // disable on AMD processors since it is slower
    if( (cpu&X264_CPU_SSE2) && !(cpu&X264_CPU_3DNOW) )
    {
        pixf->sad[PIXEL_16x16] = x264_pixel_sad_16x16_sse2;
        pixf->sad[PIXEL_16x8 ] = x264_pixel_sad_16x8_sse2;

        pixf->satd[PIXEL_16x16]= x264_pixel_satd_16x16_sse2;
        pixf->satd[PIXEL_16x8] = x264_pixel_satd_16x8_sse2;
        pixf->satd[PIXEL_8x16] = x264_pixel_satd_8x16_sse2;
        pixf->satd[PIXEL_8x8]  = x264_pixel_satd_8x8_sse2;
        pixf->satd[PIXEL_8x4]  = x264_pixel_satd_8x4_sse2;
    }
    // these are faster on both Intel and AMD
    if( cpu&X264_CPU_SSE2 )
    {
        pixf->ssd[PIXEL_16x16] = x264_pixel_ssd_16x16_sse2;
        pixf->ssd[PIXEL_16x8]  = x264_pixel_ssd_16x8_sse2;
    }
#endif

392
#ifdef ARCH_PPC
Laurent Aimar's avatar
Laurent Aimar committed
393 394 395 396 397
    if( cpu&X264_CPU_ALTIVEC )
    {
        x264_pixel_altivec_init( pixf );
    }
#endif
398
#ifdef ARCH_UltraSparc
Laurent Aimar's avatar
Laurent Aimar committed
399 400 401 402
      pixf->sad[PIXEL_8x8]   = x264_pixel_sad_8x8_vis;
      pixf->sad[PIXEL_8x16]  = x264_pixel_sad_8x16_vis;
      pixf->sad[PIXEL_16x8]  = x264_pixel_sad_16x8_vis;
      pixf->sad[PIXEL_16x16] = x264_pixel_sad_16x16_vis;
403
#endif
Laurent Aimar's avatar
Laurent Aimar committed
404 405
}