h264pred_template.c 38.9 KB
Newer Older
1 2 3 4
/*
 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
 * Copyright (c) 2003-2011 Michael Niedermayer <michaelni@gmx.at>
 *
5
 * This file is part of FFmpeg.
6
 *
7
 * FFmpeg is free software; you can redistribute it and/or
8 9 10 11
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
12
 * FFmpeg is distributed in the hope that it will be useful,
13 14 15 16 17
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with FFmpeg; if not, write to the Free Software
19 20 21 22 23 24 25 26 27 28
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 * @file
 * H.264 / AVC / MPEG4 part10 prediction functions.
 * @author Michael Niedermayer <michaelni@gmx.at>
 */

#include "mathops.h"
29 30

#include "bit_depth_template.c"
31 32 33

static void FUNCC(pred4x4_vertical)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
34
    int stride = _stride>>(sizeof(pixel)-1);
35 36 37 38 39 40
    const pixel4 a= AV_RN4PA(src-stride);

    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
41 42 43 44
}

static void FUNCC(pred4x4_horizontal)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
45
    int stride = _stride>>(sizeof(pixel)-1);
46 47 48 49
    AV_WN4PA(src+0*stride, PIXEL_SPLAT_X4(src[-1+0*stride]));
    AV_WN4PA(src+1*stride, PIXEL_SPLAT_X4(src[-1+1*stride]));
    AV_WN4PA(src+2*stride, PIXEL_SPLAT_X4(src[-1+2*stride]));
    AV_WN4PA(src+3*stride, PIXEL_SPLAT_X4(src[-1+3*stride]));
50 51 52 53
}

static void FUNCC(pred4x4_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
54
    int stride = _stride>>(sizeof(pixel)-1);
55 56
    const int dc= (  src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
                   + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;
57
    const pixel4 a = PIXEL_SPLAT_X4(dc);
58

59 60 61 62
    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
63 64
}

65 66
static void FUNCC(pred4x4_left_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
67
    int stride = _stride>>(sizeof(pixel)-1);
68
    const int dc= (  src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;
69
    const pixel4 a = PIXEL_SPLAT_X4(dc);
70

71 72 73 74
    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
75 76
}

77 78
static void FUNCC(pred4x4_top_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
79
    int stride = _stride>>(sizeof(pixel)-1);
80
    const int dc= (  src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;
81
    const pixel4 a = PIXEL_SPLAT_X4(dc);
82

83 84 85 86
    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
87 88
}

89 90
static void FUNCC(pred4x4_128_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
91
    int stride = _stride>>(sizeof(pixel)-1);
92 93 94 95 96 97
    const pixel4 a = PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1));

    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
98 99
}

100 101
static void FUNCC(pred4x4_127_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
102
    int stride = _stride>>(sizeof(pixel)-1);
103 104 105 106 107 108
    const pixel4 a = PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))-1);

    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
109 110
}

111 112
static void FUNCC(pred4x4_129_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
113
    int stride = _stride>>(sizeof(pixel)-1);
114 115 116 117 118 119
    const pixel4 a = PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))+1);

    AV_WN4PA(src+0*stride, a);
    AV_WN4PA(src+1*stride, a);
    AV_WN4PA(src+2*stride, a);
    AV_WN4PA(src+3*stride, a);
120 121 122 123
}


#define LOAD_TOP_RIGHT_EDGE\
124 125 126 127
    const unsigned av_unused t4 = topright[0];\
    const unsigned av_unused t5 = topright[1];\
    const unsigned av_unused t6 = topright[2];\
    const unsigned av_unused t7 = topright[3];\
128 129

#define LOAD_DOWN_LEFT_EDGE\
130 131 132 133
    const unsigned av_unused l4 = src[-1+4*stride];\
    const unsigned av_unused l5 = src[-1+5*stride];\
    const unsigned av_unused l6 = src[-1+6*stride];\
    const unsigned av_unused l7 = src[-1+7*stride];\
134 135

#define LOAD_LEFT_EDGE\
136 137 138 139
    const unsigned av_unused l0 = src[-1+0*stride];\
    const unsigned av_unused l1 = src[-1+1*stride];\
    const unsigned av_unused l2 = src[-1+2*stride];\
    const unsigned av_unused l3 = src[-1+3*stride];\
140 141

#define LOAD_TOP_EDGE\
142 143 144 145
    const unsigned av_unused t0 = src[ 0-1*stride];\
    const unsigned av_unused t1 = src[ 1-1*stride];\
    const unsigned av_unused t2 = src[ 2-1*stride];\
    const unsigned av_unused t3 = src[ 3-1*stride];\
146

147 148
static void FUNCC(pred4x4_down_right)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
149
    int stride = _stride>>(sizeof(pixel)-1);
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
    const int lt= src[-1-1*stride];
    LOAD_TOP_EDGE
    LOAD_LEFT_EDGE

    src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;
    src[0+2*stride]=
    src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;
    src[0+1*stride]=
    src[1+2*stride]=
    src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;
    src[0+0*stride]=
    src[1+1*stride]=
    src[2+2*stride]=
    src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
    src[1+0*stride]=
    src[2+1*stride]=
    src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;
    src[2+0*stride]=
    src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
}

172 173 174
static void FUNCC(pred4x4_down_left)(uint8_t *_src, const uint8_t *_topright, int _stride){
    pixel *src = (pixel*)_src;
    const pixel *topright = (const pixel*)_topright;
175
    int stride = _stride>>(sizeof(pixel)-1);
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE
//    LOAD_LEFT_EDGE

    src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;
    src[1+0*stride]=
    src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;
    src[2+0*stride]=
    src[1+1*stride]=
    src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;
    src[3+0*stride]=
    src[2+1*stride]=
    src[1+2*stride]=
    src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;
    src[3+1*stride]=
    src[2+2*stride]=
    src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;
    src[3+2*stride]=
    src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;
    src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
}

198 199
static void FUNCC(pred4x4_vertical_right)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
200
    int stride = _stride>>(sizeof(pixel)-1);
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222
    const int lt= src[-1-1*stride];
    LOAD_TOP_EDGE
    LOAD_LEFT_EDGE

    src[0+0*stride]=
    src[1+2*stride]=(lt + t0 + 1)>>1;
    src[1+0*stride]=
    src[2+2*stride]=(t0 + t1 + 1)>>1;
    src[2+0*stride]=
    src[3+2*stride]=(t1 + t2 + 1)>>1;
    src[3+0*stride]=(t2 + t3 + 1)>>1;
    src[0+1*stride]=
    src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
    src[1+1*stride]=
    src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;
    src[2+1*stride]=
    src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;
    src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
    src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
}

223 224 225
static void FUNCC(pred4x4_vertical_left)(uint8_t *_src, const uint8_t *_topright, int _stride){
    pixel *src = (pixel*)_src;
    const pixel *topright = (const pixel*)_topright;
226
    int stride = _stride>>(sizeof(pixel)-1);
227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE

    src[0+0*stride]=(t0 + t1 + 1)>>1;
    src[1+0*stride]=
    src[0+2*stride]=(t1 + t2 + 1)>>1;
    src[2+0*stride]=
    src[1+2*stride]=(t2 + t3 + 1)>>1;
    src[3+0*stride]=
    src[2+2*stride]=(t3 + t4+ 1)>>1;
    src[3+2*stride]=(t4 + t5+ 1)>>1;
    src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[1+1*stride]=
    src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
    src[2+1*stride]=
    src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
    src[3+1*stride]=
    src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
    src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
}

248 249
static void FUNCC(pred4x4_horizontal_up)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
250
    int stride = _stride>>(sizeof(pixel)-1);
251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
    LOAD_LEFT_EDGE

    src[0+0*stride]=(l0 + l1 + 1)>>1;
    src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;
    src[2+0*stride]=
    src[0+1*stride]=(l1 + l2 + 1)>>1;
    src[3+0*stride]=
    src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;
    src[2+1*stride]=
    src[0+2*stride]=(l2 + l3 + 1)>>1;
    src[3+1*stride]=
    src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;
    src[3+2*stride]=
    src[1+3*stride]=
    src[0+3*stride]=
    src[2+2*stride]=
    src[2+3*stride]=
    src[3+3*stride]=l3;
}

271 272
static void FUNCC(pred4x4_horizontal_down)(uint8_t *_src, const uint8_t *topright, int _stride){
    pixel *src = (pixel*)_src;
273
    int stride = _stride>>(sizeof(pixel)-1);
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295
    const int lt= src[-1-1*stride];
    LOAD_TOP_EDGE
    LOAD_LEFT_EDGE

    src[0+0*stride]=
    src[2+1*stride]=(lt + l0 + 1)>>1;
    src[1+0*stride]=
    src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;
    src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;
    src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[0+1*stride]=
    src[2+2*stride]=(l0 + l1 + 1)>>1;
    src[1+1*stride]=
    src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
    src[0+2*stride]=
    src[2+3*stride]=(l1 + l2+ 1)>>1;
    src[1+2*stride]=
    src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
    src[0+3*stride]=(l2 + l3 + 1)>>1;
    src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
}

296
static void FUNCC(pred16x16_vertical)(uint8_t *_src, int _stride){
297
    int i;
298
    pixel *src = (pixel*)_src;
299
    int stride = _stride>>(sizeof(pixel)-1);
300 301 302 303
    const pixel4 a = AV_RN4PA(((pixel4*)(src-stride))+0);
    const pixel4 b = AV_RN4PA(((pixel4*)(src-stride))+1);
    const pixel4 c = AV_RN4PA(((pixel4*)(src-stride))+2);
    const pixel4 d = AV_RN4PA(((pixel4*)(src-stride))+3);
304 305

    for(i=0; i<16; i++){
306 307 308 309
        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
        AV_WN4PA(((pixel4*)(src+i*stride))+2, c);
        AV_WN4PA(((pixel4*)(src+i*stride))+3, d);
310 311 312
    }
}

313
static void FUNCC(pred16x16_horizontal)(uint8_t *_src, int stride){
314
    int i;
315
    pixel *src = (pixel*)_src;
316
    stride >>= sizeof(pixel)-1;
317 318

    for(i=0; i<16; i++){
319 320 321 322 323 324
        const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);

        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+2, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+3, a);
325 326 327
    }
}

328 329
#define PREDICT_16x16_DC(v)\
    for(i=0; i<16; i++){\
330 331 332 333
        AV_WN4PA(src+ 0, v);\
        AV_WN4PA(src+ 4, v);\
        AV_WN4PA(src+ 8, v);\
        AV_WN4PA(src+12, v);\
334 335 336 337
        src += stride;\
    }

static void FUNCC(pred16x16_dc)(uint8_t *_src, int stride){
338
    int i, dc=0;
339 340
    pixel *src = (pixel*)_src;
    pixel4 dcsplat;
341
    stride >>= sizeof(pixel)-1;
342 343 344 345 346 347 348 349 350

    for(i=0;i<16; i++){
        dc+= src[-1+i*stride];
    }

    for(i=0;i<16; i++){
        dc+= src[i-stride];
    }

351 352
    dcsplat = PIXEL_SPLAT_X4((dc+16)>>5);
    PREDICT_16x16_DC(dcsplat);
353 354
}

355
static void FUNCC(pred16x16_left_dc)(uint8_t *_src, int stride){
356
    int i, dc=0;
357 358
    pixel *src = (pixel*)_src;
    pixel4 dcsplat;
359
    stride >>= sizeof(pixel)-1;
360 361 362 363 364

    for(i=0;i<16; i++){
        dc+= src[-1+i*stride];
    }

365 366
    dcsplat = PIXEL_SPLAT_X4((dc+8)>>4);
    PREDICT_16x16_DC(dcsplat);
367 368
}

369
static void FUNCC(pred16x16_top_dc)(uint8_t *_src, int stride){
370
    int i, dc=0;
371 372
    pixel *src = (pixel*)_src;
    pixel4 dcsplat;
373
    stride >>= sizeof(pixel)-1;
374 375 376 377 378

    for(i=0;i<16; i++){
        dc+= src[i-stride];
    }

379 380
    dcsplat = PIXEL_SPLAT_X4((dc+8)>>4);
    PREDICT_16x16_DC(dcsplat);
381 382
}

383 384 385 386
#define PRED16x16_X(n, v) \
static void FUNCC(pred16x16_##n##_dc)(uint8_t *_src, int stride){\
    int i;\
    pixel *src = (pixel*)_src;\
387
    stride >>= sizeof(pixel)-1;\
388
    PREDICT_16x16_DC(PIXEL_SPLAT_X4(v));\
389 390
}

391 392 393
PRED16x16_X(127, (1<<(BIT_DEPTH-1))-1)
PRED16x16_X(128, (1<<(BIT_DEPTH-1))+0)
PRED16x16_X(129, (1<<(BIT_DEPTH-1))+1)
394

395
static inline void FUNCC(pred16x16_plane_compat)(uint8_t *p_src, int p_stride, const int svq3, const int rv40){
396 397
  int i, j, k;
  int a;
398
  INIT_CLIP
399 400
  pixel *src = (pixel*)p_src;
  int stride = p_stride>>(sizeof(pixel)-1);
401 402 403
  const pixel * const src0 = src +7-stride;
  const pixel *       src1 = src +8*stride-1;
  const pixel *       src2 = src1-2*stride;    // == src+6*stride-1;
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
  int H = src0[1] - src0[-1];
  int V = src1[0] - src2[ 0];
  for(k=2; k<=8; ++k) {
    src1 += stride; src2 -= stride;
    H += k*(src0[k] - src0[-k]);
    V += k*(src1[0] - src2[ 0]);
  }
  if(svq3){
    H = ( 5*(H/4) ) / 16;
    V = ( 5*(V/4) ) / 16;

    /* required for 100% accuracy */
    i = H; H = V; V = i;
  }else if(rv40){
    H = ( H + (H>>2) ) >> 4;
    V = ( V + (V>>2) ) >> 4;
  }else{
    H = ( 5*H+32 ) >> 6;
    V = ( 5*V+32 ) >> 6;
  }

  a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);
  for(j=16; j>0; --j) {
    int b = a;
    a += V;
    for(i=-16; i<0; i+=4) {
430 431 432 433
      src[16+i] = CLIP((b    ) >> 5);
      src[17+i] = CLIP((b+  H) >> 5);
      src[18+i] = CLIP((b+2*H) >> 5);
      src[19+i] = CLIP((b+3*H) >> 5);
434 435 436 437 438 439
      b += 4*H;
    }
    src += stride;
  }
}

440 441
static void FUNCC(pred16x16_plane)(uint8_t *src, int stride){
    FUNCC(pred16x16_plane_compat)(src, stride, 0, 0);
442 443
}

444
static void FUNCC(pred8x8_vertical)(uint8_t *_src, int _stride){
445
    int i;
446
    pixel *src = (pixel*)_src;
447
    int stride = _stride>>(sizeof(pixel)-1);
448 449
    const pixel4 a= AV_RN4PA(((pixel4*)(src-stride))+0);
    const pixel4 b= AV_RN4PA(((pixel4*)(src-stride))+1);
450 451

    for(i=0; i<8; i++){
452 453
        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
454 455 456
    }
}

457 458 459 460 461 462 463 464 465 466 467 468 469
static void FUNCC(pred8x16_vertical)(uint8_t *_src, int _stride){
    int i;
    pixel *src = (pixel*)_src;
    int stride = _stride>>(sizeof(pixel)-1);
    const pixel4 a= AV_RN4PA(((pixel4*)(src-stride))+0);
    const pixel4 b= AV_RN4PA(((pixel4*)(src-stride))+1);

    for(i=0; i<16; i++){
        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
    }
}

470
static void FUNCC(pred8x8_horizontal)(uint8_t *_src, int stride){
471
    int i;
472
    pixel *src = (pixel*)_src;
473
    stride >>= sizeof(pixel)-1;
474 475

    for(i=0; i<8; i++){
476 477 478
        const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
479 480 481
    }
}

482 483 484 485 486 487 488 489 490 491 492
static void FUNCC(pred8x16_horizontal)(uint8_t *_src, int stride){
    int i;
    pixel *src = (pixel*)_src;
    stride >>= sizeof(pixel)-1;
    for(i=0; i<16; i++){
        const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
    }
}

493 494 495
#define PRED8x8_X(n, v)\
static void FUNCC(pred8x8_##n##_dc)(uint8_t *_src, int stride){\
    int i;\
496
    const pixel4 a = PIXEL_SPLAT_X4(v);\
497
    pixel *src = (pixel*)_src;\
498
    stride >>= sizeof(pixel)-1;\
499
    for(i=0; i<8; i++){\
500 501
        AV_WN4PA(((pixel4*)(src+i*stride))+0, a);\
        AV_WN4PA(((pixel4*)(src+i*stride))+1, a);\
502
    }\
503 504
}

505 506 507
PRED8x8_X(127, (1<<(BIT_DEPTH-1))-1)
PRED8x8_X(128, (1<<(BIT_DEPTH-1))+0)
PRED8x8_X(129, (1<<(BIT_DEPTH-1))+1)
508

509 510 511 512 513
static void FUNCC(pred8x16_128_dc)(uint8_t *_src, int stride){
    FUNCC(pred8x8_128_dc)(_src, stride);
    FUNCC(pred8x8_128_dc)(_src+8*stride, stride);
}

514
static void FUNCC(pred8x8_left_dc)(uint8_t *_src, int stride){
515 516
    int i;
    int dc0, dc2;
517 518
    pixel4 dc0splat, dc2splat;
    pixel *src = (pixel*)_src;
519
    stride >>= sizeof(pixel)-1;
520 521 522 523 524 525

    dc0=dc2=0;
    for(i=0;i<4; i++){
        dc0+= src[-1+i*stride];
        dc2+= src[-1+(i+4)*stride];
    }
526 527
    dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);
    dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
528 529

    for(i=0; i<4; i++){
530 531
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc0splat);
532 533
    }
    for(i=4; i<8; i++){
534 535
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc2splat);
536 537 538
    }
}

539 540 541 542 543
static void FUNCC(pred8x16_left_dc)(uint8_t *_src, int stride){
    FUNCC(pred8x8_left_dc)(_src, stride);
    FUNCC(pred8x8_left_dc)(_src+8*stride, stride);
}

544
static void FUNCC(pred8x8_top_dc)(uint8_t *_src, int stride){
545 546
    int i;
    int dc0, dc1;
547 548
    pixel4 dc0splat, dc1splat;
    pixel *src = (pixel*)_src;
549
    stride >>= sizeof(pixel)-1;
550 551 552 553 554 555

    dc0=dc1=0;
    for(i=0;i<4; i++){
        dc0+= src[i-stride];
        dc1+= src[4+i-stride];
    }
556 557
    dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);
    dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
558 559

    for(i=0; i<4; i++){
560 561
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
562 563
    }
    for(i=4; i<8; i++){
564 565
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
566 567 568
    }
}

569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589
static void FUNCC(pred8x16_top_dc)(uint8_t *_src, int stride){
    int i;
    int dc0, dc1;
    pixel4 dc0splat, dc1splat;
    pixel *src = (pixel*)_src;
    stride >>= sizeof(pixel)-1;

    dc0=dc1=0;
    for(i=0;i<4; i++){
        dc0+= src[i-stride];
        dc1+= src[4+i-stride];
    }
    dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);
    dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);

    for(i=0; i<16; i++){
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
    }
}

590
static void FUNCC(pred8x8_dc)(uint8_t *_src, int stride){
591
    int i;
592 593 594
    int dc0, dc1, dc2;
    pixel4 dc0splat, dc1splat, dc2splat, dc3splat;
    pixel *src = (pixel*)_src;
595
    stride >>= sizeof(pixel)-1;
596 597 598 599 600 601 602

    dc0=dc1=dc2=0;
    for(i=0;i<4; i++){
        dc0+= src[-1+i*stride] + src[i-stride];
        dc1+= src[4+i-stride];
        dc2+= src[-1+(i+4)*stride];
    }
603 604 605 606
    dc0splat = PIXEL_SPLAT_X4((dc0 + 4)>>3);
    dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
    dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
    dc3splat = PIXEL_SPLAT_X4((dc1 + dc2 + 4)>>3);
607 608

    for(i=0; i<4; i++){
609 610
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
611 612
    }
    for(i=4; i<8; i++){
613 614
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc3splat);
615 616 617
    }
}

618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659
static void FUNCC(pred8x16_dc)(uint8_t *_src, int stride){
    int i;
    int dc0, dc1, dc2, dc3, dc4;
    pixel4 dc0splat, dc1splat, dc2splat, dc3splat, dc4splat, dc5splat, dc6splat, dc7splat;
    pixel *src = (pixel*)_src;
    stride >>= sizeof(pixel)-1;

    dc0=dc1=dc2=dc3=dc4=0;
    for(i=0;i<4; i++){
        dc0+= src[-1+i*stride] + src[i-stride];
        dc1+= src[4+i-stride];
        dc2+= src[-1+(i+4)*stride];
        dc3+= src[-1+(i+8)*stride];
        dc4+= src[-1+(i+12)*stride];
    }
    dc0splat = PIXEL_SPLAT_X4((dc0 + 4)>>3);
    dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
    dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
    dc3splat = PIXEL_SPLAT_X4((dc1 + dc2 + 4)>>3);
    dc4splat = PIXEL_SPLAT_X4((dc3 + 2)>>2);
    dc5splat = PIXEL_SPLAT_X4((dc1 + dc3 + 4)>>3);
    dc6splat = PIXEL_SPLAT_X4((dc4 + 2)>>2);
    dc7splat = PIXEL_SPLAT_X4((dc1 + dc4 + 4)>>3);

    for(i=0; i<4; i++){
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
    }
    for(i=4; i<8; i++){
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc3splat);
    }
    for(i=8; i<12; i++){
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc4splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc5splat);
    }
    for(i=12; i<16; i++){
        AV_WN4PA(((pixel4*)(src+i*stride))+0, dc6splat);
        AV_WN4PA(((pixel4*)(src+i*stride))+1, dc7splat);
    }
}

660
//the following 4 function should not be optimized!
661 662 663
static void FUNC(pred8x8_mad_cow_dc_l0t)(uint8_t *src, int stride){
    FUNCC(pred8x8_top_dc)(src, stride);
    FUNCC(pred4x4_dc)(src, NULL, stride);
664 665
}

666 667 668 669 670
static void FUNC(pred8x16_mad_cow_dc_l0t)(uint8_t *src, int stride){
    FUNCC(pred8x16_top_dc)(src, stride);
    FUNCC(pred4x4_dc)(src, NULL, stride);
}

671 672 673
static void FUNC(pred8x8_mad_cow_dc_0lt)(uint8_t *src, int stride){
    FUNCC(pred8x8_dc)(src, stride);
    FUNCC(pred4x4_top_dc)(src, NULL, stride);
674 675
}

676 677 678 679 680
static void FUNC(pred8x16_mad_cow_dc_0lt)(uint8_t *src, int stride){
    FUNCC(pred8x16_dc)(src, stride);
    FUNCC(pred4x4_top_dc)(src, NULL, stride);
}

681 682 683 684
static void FUNC(pred8x8_mad_cow_dc_l00)(uint8_t *src, int stride){
    FUNCC(pred8x8_left_dc)(src, stride);
    FUNCC(pred4x4_128_dc)(src + 4*stride                  , NULL, stride);
    FUNCC(pred4x4_128_dc)(src + 4*stride + 4*sizeof(pixel), NULL, stride);
685 686
}

687 688 689 690 691 692
static void FUNC(pred8x16_mad_cow_dc_l00)(uint8_t *src, int stride){
    FUNCC(pred8x16_left_dc)(src, stride);
    FUNCC(pred4x4_128_dc)(src + 4*stride                  , NULL, stride);
    FUNCC(pred4x4_128_dc)(src + 4*stride + 4*sizeof(pixel), NULL, stride);
}

693 694 695 696
static void FUNC(pred8x8_mad_cow_dc_0l0)(uint8_t *src, int stride){
    FUNCC(pred8x8_left_dc)(src, stride);
    FUNCC(pred4x4_128_dc)(src                  , NULL, stride);
    FUNCC(pred4x4_128_dc)(src + 4*sizeof(pixel), NULL, stride);
697 698
}

699 700 701 702 703 704
static void FUNC(pred8x16_mad_cow_dc_0l0)(uint8_t *src, int stride){
    FUNCC(pred8x16_left_dc)(src, stride);
    FUNCC(pred4x4_128_dc)(src                  , NULL, stride);
    FUNCC(pred4x4_128_dc)(src + 4*sizeof(pixel), NULL, stride);
}

705
static void FUNCC(pred8x8_plane)(uint8_t *_src, int _stride){
706 707
  int j, k;
  int a;
708 709
  INIT_CLIP
  pixel *src = (pixel*)_src;
710
  int stride = _stride>>(sizeof(pixel)-1);
711 712 713
  const pixel * const src0 = src +3-stride;
  const pixel *       src1 = src +4*stride-1;
  const pixel *       src2 = src1-2*stride;    // == src+2*stride-1;
714 715 716 717 718 719 720 721 722 723 724 725 726 727
  int H = src0[1] - src0[-1];
  int V = src1[0] - src2[ 0];
  for(k=2; k<=4; ++k) {
    src1 += stride; src2 -= stride;
    H += k*(src0[k] - src0[-k]);
    V += k*(src1[0] - src2[ 0]);
  }
  H = ( 17*H+16 ) >> 5;
  V = ( 17*V+16 ) >> 5;

  a = 16*(src1[0] + src2[8]+1) - 3*(V+H);
  for(j=8; j>0; --j) {
    int b = a;
    a += V;
728 729 730 731 732 733 734 735
    src[0] = CLIP((b    ) >> 5);
    src[1] = CLIP((b+  H) >> 5);
    src[2] = CLIP((b+2*H) >> 5);
    src[3] = CLIP((b+3*H) >> 5);
    src[4] = CLIP((b+4*H) >> 5);
    src[5] = CLIP((b+5*H) >> 5);
    src[6] = CLIP((b+6*H) >> 5);
    src[7] = CLIP((b+7*H) >> 5);
736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776
    src += stride;
  }
}

static void FUNCC(pred8x16_plane)(uint8_t *_src, int _stride){
  int j, k;
  int a;
  INIT_CLIP
  pixel *src = (pixel*)_src;
  int stride = _stride>>(sizeof(pixel)-1);
  const pixel * const src0 = src +3-stride;
  const pixel *       src1 = src +8*stride-1;
  const pixel *       src2 = src1-2*stride;    // == src+6*stride-1;
  int H = src0[1] - src0[-1];
  int V = src1[0] - src2[ 0];

  for (k = 2; k <= 4; ++k) {
      src1 += stride; src2 -= stride;
      H += k*(src0[k] - src0[-k]);
      V += k*(src1[0] - src2[ 0]);
  }
  for (; k <= 8; ++k) {
      src1 += stride; src2 -= stride;
      V += k*(src1[0] - src2[0]);
  }

  H = (17*H+16) >> 5;
  V = (5*V+32) >> 6;

  a = 16*(src1[0] + src2[8] + 1) - 7*V - 3*H;
  for(j=16; j>0; --j) {
    int b = a;
    a += V;
    src[0] = CLIP((b    ) >> 5);
    src[1] = CLIP((b+  H) >> 5);
    src[2] = CLIP((b+2*H) >> 5);
    src[3] = CLIP((b+3*H) >> 5);
    src[4] = CLIP((b+4*H) >> 5);
    src[5] = CLIP((b+5*H) >> 5);
    src[6] = CLIP((b+6*H) >> 5);
    src[7] = CLIP((b+7*H) >> 5);
777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813
    src += stride;
  }
}

#define SRC(x,y) src[(x)+(y)*stride]
#define PL(y) \
    const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
#define PREDICT_8x8_LOAD_LEFT \
    const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
                     + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
    PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
    const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2

#define PT(x) \
    const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
#define PREDICT_8x8_LOAD_TOP \
    const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
                     + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
    PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
    const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
                     + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2

#define PTR(x) \
    t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
#define PREDICT_8x8_LOAD_TOPRIGHT \
    int t8, t9, t10, t11, t12, t13, t14, t15; \
    if(has_topright) { \
        PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \
        t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \
    } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);

#define PREDICT_8x8_LOAD_TOPLEFT \
    const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2

#define PREDICT_8x8_DC(v) \
    int y; \
    for( y = 0; y < 8; y++ ) { \
814 815
        AV_WN4PA(((pixel4*)src)+0, v); \
        AV_WN4PA(((pixel4*)src)+1, v); \
816 817 818
        src += stride; \
    }

819
static void FUNCC(pred8x8l_128_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
820
{
821
    pixel *src = (pixel*)_src;
822
    int stride = _stride>>(sizeof(pixel)-1);
823 824

    PREDICT_8x8_DC(PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1)));
825
}
826
static void FUNCC(pred8x8l_left_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
827
{
828
    pixel *src = (pixel*)_src;
829
    int stride = _stride>>(sizeof(pixel)-1);
830

831
    PREDICT_8x8_LOAD_LEFT;
832
    const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3);
833 834
    PREDICT_8x8_DC(dc);
}
835
static void FUNCC(pred8x8l_top_dc)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
836
{
837 838
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
839

840
    PREDICT_8x8_LOAD_TOP;
841
    const pixel4 dc = PIXEL_SPLAT_X4((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3);
842 843
    PREDICT_8x8_DC(dc);
}
844
static void FUNCC(pred8x8l_dc)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
845
{
846 847
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
848

849 850
    PREDICT_8x8_LOAD_LEFT;
    PREDICT_8x8_LOAD_TOP;
851 852
    const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7
                                     +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4);
853 854
    PREDICT_8x8_DC(dc);
}
855
static void FUNCC(pred8x8l_horizontal)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
856
{
857 858
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
859
    pixel4 a;
860

861
    PREDICT_8x8_LOAD_LEFT;
862 863 864
#define ROW(y) a = PIXEL_SPLAT_X4(l##y); \
               AV_WN4PA(src+y*stride, a); \
               AV_WN4PA(src+y*stride+4, a);
865 866 867
    ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
#undef ROW
}
868
static void FUNCC(pred8x8l_vertical)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
869 870
{
    int y;
871
    pixel *src = (pixel*)_src;
872
    int stride = _stride>>(sizeof(pixel)-1);
873
    pixel4 a, b;
874

875 876 877 878 879 880 881 882 883
    PREDICT_8x8_LOAD_TOP;
    src[0] = t0;
    src[1] = t1;
    src[2] = t2;
    src[3] = t3;
    src[4] = t4;
    src[5] = t5;
    src[6] = t6;
    src[7] = t7;
884 885
    a = AV_RN4PA(((pixel4*)src)+0);
    b = AV_RN4PA(((pixel4*)src)+1);
886
    for( y = 1; y < 8; y++ ) {
887 888
        AV_WN4PA(((pixel4*)(src+y*stride))+0, a);
        AV_WN4PA(((pixel4*)(src+y*stride))+1, b);
889
    }
890
}
891
static void FUNCC(pred8x8l_down_left)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
892
{
893 894
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912
    PREDICT_8x8_LOAD_TOP;
    PREDICT_8x8_LOAD_TOPRIGHT;
    SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;
    SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;
    SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;
    SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;
    SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;
    SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;
    SRC(0,6)=SRC(1,5)=SRC(2,4)=SRC(3,3)=SRC(4,2)=SRC(5,1)=SRC(6,0)= (t6 + 2*t7 + t8 + 2) >> 2;
    SRC(0,7)=SRC(1,6)=SRC(2,5)=SRC(3,4)=SRC(4,3)=SRC(5,2)=SRC(6,1)=SRC(7,0)= (t7 + 2*t8 + t9 + 2) >> 2;
    SRC(1,7)=SRC(2,6)=SRC(3,5)=SRC(4,4)=SRC(5,3)=SRC(6,2)=SRC(7,1)= (t8 + 2*t9 + t10 + 2) >> 2;
    SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;
    SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;
    SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;
    SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;
    SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;
    SRC(7,7)= (t14 + 3*t15 + 2) >> 2;
}
913
static void FUNCC(pred8x8l_down_right)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
914
{
915 916
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935
    PREDICT_8x8_LOAD_TOP;
    PREDICT_8x8_LOAD_LEFT;
    PREDICT_8x8_LOAD_TOPLEFT;
    SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;
    SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;
    SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;
    SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;
    SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;
    SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;
    SRC(0,1)=SRC(1,2)=SRC(2,3)=SRC(3,4)=SRC(4,5)=SRC(5,6)=SRC(6,7)= (l1 + 2*l0 + lt + 2) >> 2;
    SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)=SRC(4,4)=SRC(5,5)=SRC(6,6)=SRC(7,7)= (l0 + 2*lt + t0 + 2) >> 2;
    SRC(1,0)=SRC(2,1)=SRC(3,2)=SRC(4,3)=SRC(5,4)=SRC(6,5)=SRC(7,6)= (lt + 2*t0 + t1 + 2) >> 2;
    SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;
    SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;
    SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;
    SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;
    SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;
    SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;
}
936
static void FUNCC(pred8x8l_vertical_right)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
937
{
938 939
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
    PREDICT_8x8_LOAD_TOP;
    PREDICT_8x8_LOAD_LEFT;
    PREDICT_8x8_LOAD_TOPLEFT;
    SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;
    SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;
    SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;
    SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;
    SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;
    SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;
    SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;
    SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;
    SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;
    SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;
    SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;
    SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;
    SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;
    SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;
    SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;
    SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;
    SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;
    SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;
    SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;
    SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;
    SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;
    SRC(7,0)= (t6 + t7 + 1) >> 1;
}
966
static void FUNCC(pred8x8l_horizontal_down)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
967
{
968 969
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995
    PREDICT_8x8_LOAD_TOP;
    PREDICT_8x8_LOAD_LEFT;
    PREDICT_8x8_LOAD_TOPLEFT;
    SRC(0,7)= (l6 + l7 + 1) >> 1;
    SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;
    SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;
    SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;
    SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;
    SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;
    SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;
    SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;
    SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;
    SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;
    SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;
    SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;
    SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;
    SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;
    SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;
    SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;
    SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;
    SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;
    SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;
    SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;
    SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;
    SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;
}
996
static void FUNCC(pred8x8l_vertical_left)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
997
{
998 999
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
    PREDICT_8x8_LOAD_TOP;
    PREDICT_8x8_LOAD_TOPRIGHT;
    SRC(0,0)= (t0 + t1 + 1) >> 1;
    SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;
    SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;
    SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;
    SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;
    SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;
    SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;
    SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;
    SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;
    SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;
    SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;
    SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;
    SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;
    SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;
    SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;
    SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;
    SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;
    SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;
    SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;
    SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;
    SRC(7,6)= (t10 + t11 + 1) >> 1;
    SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;
}
1025
static void FUNCC(pred8x8l_horizontal_up)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
1026
{
1027 1028
    pixel *src = (pixel*)p_src;
    int stride = p_stride>>(sizeof(pixel)-1);
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058
    PREDICT_8x8_LOAD_LEFT;
    SRC(0,0)= (l0 + l1 + 1) >> 1;
    SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;
    SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;
    SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;
    SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;
    SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;
    SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;
    SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;
    SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;
    SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;
    SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;
    SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;
    SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;
    SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;
    SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=
    SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=
    SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=
    SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;
}
#undef PREDICT_8x8_LOAD_LEFT
#undef PREDICT_8x8_LOAD_TOP
#undef PREDICT_8x8_LOAD_TOPLEFT
#undef PREDICT_8x8_LOAD_TOPRIGHT
#undef PREDICT_8x8_DC
#undef PTR
#undef PT
#undef PL
#undef SRC

1059
static void FUNCC(pred4x4_vertical_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
1060
    int i;
1061 1062 1063
    pixel *pix = (pixel*)p_pix;
    const dctcoef *block = (const dctcoef*)p_block;
    stride >>= sizeof(pixel)-1;
1064 1065
    pix -= stride;
    for(i=0; i<4; i++){
1066
        pixel v = pix[0];
1067 1068 1069 1070 1071 1072 1073 1074 1075
        pix[1*stride]= v += block[0];
        pix[2*stride]= v += block[4];
        pix[3*stride]= v += block[8];
        pix[4*stride]= v +  block[12];
        pix++;
        block++;
    }
}

1076
static void FUNCC(pred4x4_horizontal_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
1077
    int i;
1078 1079 1080
    pixel *pix = (pixel*)p_pix;
    const dctcoef *block = (const dctcoef*)p_block;
    stride >>= sizeof(pixel)-1;
1081
    for(i=0; i<4; i++){
1082
        pixel v = pix[-1];
1083 1084 1085 1086 1087 1088 1089 1090 1091
        pix[0]= v += block[0];
        pix[1]= v += block[1];
        pix[2]= v += block[2];
        pix[3]= v +  block[3];
        pix+= stride;
        block+= 4;
    }
}

1092
static void FUNCC(pred8x8l_vertical_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
1093
    int i;
1094 1095 1096
    pixel *pix = (pixel*)p_pix;
    const dctcoef *block = (const dctcoef*)p_block;
    stride >>= sizeof(pixel)-1;
1097 1098
    pix -= stride;
    for(i=0; i<8; i++){
1099
        pixel v = pix[0];
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112
        pix[1*stride]= v += block[0];
        pix[2*stride]= v += block[8];
        pix[3*stride]= v += block[16];
        pix[4*stride]= v += block[24];
        pix[5*stride]= v += block[32];
        pix[6*stride]= v += block[40];
        pix[7*stride]= v += block[48];
        pix[8*stride]= v +  block[56];
        pix++;
        block++;
    }
}

1113
static void FUNCC(pred8x8l_horizontal_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
1114
    int i;
1115 1116 1117
    pixel *pix = (pixel*)p_pix;
    const dctcoef *block = (const dctcoef*)p_block;
    stride >>= sizeof(pixel)-1;
1118
    for(i=0; i<8; i++){
1119
        pixel v = pix[-1];
1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132
        pix[0]= v += block[0];
        pix[1]= v += block[1];
        pix[2]= v += block[2];
        pix[3]= v += block[3];
        pix[4]= v += block[4];
        pix[5]= v += block[5];
        pix[6]= v += block[6];
        pix[7]= v +  block[7];
        pix+= stride;
        block+= 8;
    }
}

1133
static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1134 1135
    int i;
    for(i=0; i<16; i++)
1136
        FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
1137 1138
}

1139
static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1140 1141
    int i;
    for(i=0; i<16; i++)
1142
        FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
1143 1144
}

1145
static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1146 1147
    int i;
    for(i=0; i<4; i++)
1148
        FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
1149 1150
}

1151 1152 1153 1154 1155 1156 1157 1158
static void FUNCC(pred8x16_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
    int i;
    for(i=0; i<4; i++)
        FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
    for(i=4; i<8; i++)
        FUNCC(pred4x4_vertical_add)(pix + block_offset[i+4], block + i*16*sizeof(pixel), stride);
}

1159
static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1160 1161
    int i;
    for(i=0; i<4; i++)
1162
        FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
1163
}
1164 1165 1166 1167 1168 1169 1170 1171

static void FUNCC(pred8x16_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
    int i;
    for(i=0; i<4; i++)
        FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
    for(i=4; i<8; i++)
        FUNCC(pred4x4_horizontal_add)(pix + block_offset[i+4], block + i*16*sizeof(pixel), stride);
}