00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00028 #include "mathops.h"
00029 #include "high_bit_depth.h"
00030
00031 static void FUNCC(pred4x4_vertical)(uint8_t *_src, const uint8_t *topright, int _stride){
00032 pixel *src = (pixel*)_src;
00033 int stride = _stride>>(sizeof(pixel)-1);
00034 const pixel4 a= AV_RN4PA(src-stride);
00035
00036 AV_WN4PA(src+0*stride, a);
00037 AV_WN4PA(src+1*stride, a);
00038 AV_WN4PA(src+2*stride, a);
00039 AV_WN4PA(src+3*stride, a);
00040 }
00041
00042 static void FUNCC(pred4x4_horizontal)(uint8_t *_src, const uint8_t *topright, int _stride){
00043 pixel *src = (pixel*)_src;
00044 int stride = _stride>>(sizeof(pixel)-1);
00045 AV_WN4PA(src+0*stride, PIXEL_SPLAT_X4(src[-1+0*stride]));
00046 AV_WN4PA(src+1*stride, PIXEL_SPLAT_X4(src[-1+1*stride]));
00047 AV_WN4PA(src+2*stride, PIXEL_SPLAT_X4(src[-1+2*stride]));
00048 AV_WN4PA(src+3*stride, PIXEL_SPLAT_X4(src[-1+3*stride]));
00049 }
00050
00051 static void FUNCC(pred4x4_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
00052 pixel *src = (pixel*)_src;
00053 int stride = _stride>>(sizeof(pixel)-1);
00054 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
00055 + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;
00056 const pixel4 a = PIXEL_SPLAT_X4(dc);
00057
00058 AV_WN4PA(src+0*stride, a);
00059 AV_WN4PA(src+1*stride, a);
00060 AV_WN4PA(src+2*stride, a);
00061 AV_WN4PA(src+3*stride, a);
00062 }
00063
00064 static void FUNCC(pred4x4_left_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
00065 pixel *src = (pixel*)_src;
00066 int stride = _stride>>(sizeof(pixel)-1);
00067 const int dc= ( src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;
00068 const pixel4 a = PIXEL_SPLAT_X4(dc);
00069
00070 AV_WN4PA(src+0*stride, a);
00071 AV_WN4PA(src+1*stride, a);
00072 AV_WN4PA(src+2*stride, a);
00073 AV_WN4PA(src+3*stride, a);
00074 }
00075
00076 static void FUNCC(pred4x4_top_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
00077 pixel *src = (pixel*)_src;
00078 int stride = _stride>>(sizeof(pixel)-1);
00079 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;
00080 const pixel4 a = PIXEL_SPLAT_X4(dc);
00081
00082 AV_WN4PA(src+0*stride, a);
00083 AV_WN4PA(src+1*stride, a);
00084 AV_WN4PA(src+2*stride, a);
00085 AV_WN4PA(src+3*stride, a);
00086 }
00087
00088 static void FUNCC(pred4x4_128_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
00089 pixel *src = (pixel*)_src;
00090 int stride = _stride>>(sizeof(pixel)-1);
00091 const pixel4 a = PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1));
00092
00093 AV_WN4PA(src+0*stride, a);
00094 AV_WN4PA(src+1*stride, a);
00095 AV_WN4PA(src+2*stride, a);
00096 AV_WN4PA(src+3*stride, a);
00097 }
00098
00099 static void FUNCC(pred4x4_127_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
00100 pixel *src = (pixel*)_src;
00101 int stride = _stride>>(sizeof(pixel)-1);
00102 const pixel4 a = PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))-1);
00103
00104 AV_WN4PA(src+0*stride, a);
00105 AV_WN4PA(src+1*stride, a);
00106 AV_WN4PA(src+2*stride, a);
00107 AV_WN4PA(src+3*stride, a);
00108 }
00109
00110 static void FUNCC(pred4x4_129_dc)(uint8_t *_src, const uint8_t *topright, int _stride){
00111 pixel *src = (pixel*)_src;
00112 int stride = _stride>>(sizeof(pixel)-1);
00113 const pixel4 a = PIXEL_SPLAT_X4((1<<(BIT_DEPTH-1))+1);
00114
00115 AV_WN4PA(src+0*stride, a);
00116 AV_WN4PA(src+1*stride, a);
00117 AV_WN4PA(src+2*stride, a);
00118 AV_WN4PA(src+3*stride, a);
00119 }
00120
00121
00122 #define LOAD_TOP_RIGHT_EDGE\
00123 const unsigned av_unused t4 = topright[0];\
00124 const unsigned av_unused t5 = topright[1];\
00125 const unsigned av_unused t6 = topright[2];\
00126 const unsigned av_unused t7 = topright[3];\
00127
00128 #define LOAD_DOWN_LEFT_EDGE\
00129 const unsigned av_unused l4 = src[-1+4*stride];\
00130 const unsigned av_unused l5 = src[-1+5*stride];\
00131 const unsigned av_unused l6 = src[-1+6*stride];\
00132 const unsigned av_unused l7 = src[-1+7*stride];\
00133
00134 #define LOAD_LEFT_EDGE\
00135 const unsigned av_unused l0 = src[-1+0*stride];\
00136 const unsigned av_unused l1 = src[-1+1*stride];\
00137 const unsigned av_unused l2 = src[-1+2*stride];\
00138 const unsigned av_unused l3 = src[-1+3*stride];\
00139
00140 #define LOAD_TOP_EDGE\
00141 const unsigned av_unused t0 = src[ 0-1*stride];\
00142 const unsigned av_unused t1 = src[ 1-1*stride];\
00143 const unsigned av_unused t2 = src[ 2-1*stride];\
00144 const unsigned av_unused t3 = src[ 3-1*stride];\
00145
00146 static void FUNCC(pred4x4_down_right)(uint8_t *_src, const uint8_t *topright, int _stride){
00147 pixel *src = (pixel*)_src;
00148 int stride = _stride>>(sizeof(pixel)-1);
00149 const int lt= src[-1-1*stride];
00150 LOAD_TOP_EDGE
00151 LOAD_LEFT_EDGE
00152
00153 src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;
00154 src[0+2*stride]=
00155 src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;
00156 src[0+1*stride]=
00157 src[1+2*stride]=
00158 src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;
00159 src[0+0*stride]=
00160 src[1+1*stride]=
00161 src[2+2*stride]=
00162 src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
00163 src[1+0*stride]=
00164 src[2+1*stride]=
00165 src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;
00166 src[2+0*stride]=
00167 src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
00168 src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
00169 }
00170
00171 static void FUNCC(pred4x4_down_left)(uint8_t *_src, const uint8_t *_topright, int _stride){
00172 pixel *src = (pixel*)_src;
00173 const pixel *topright = (const pixel*)_topright;
00174 int stride = _stride>>(sizeof(pixel)-1);
00175 LOAD_TOP_EDGE
00176 LOAD_TOP_RIGHT_EDGE
00177
00178
00179 src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;
00180 src[1+0*stride]=
00181 src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;
00182 src[2+0*stride]=
00183 src[1+1*stride]=
00184 src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;
00185 src[3+0*stride]=
00186 src[2+1*stride]=
00187 src[1+2*stride]=
00188 src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;
00189 src[3+1*stride]=
00190 src[2+2*stride]=
00191 src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;
00192 src[3+2*stride]=
00193 src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;
00194 src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
00195 }
00196
00197 static void FUNCC(pred4x4_vertical_right)(uint8_t *_src, const uint8_t *topright, int _stride){
00198 pixel *src = (pixel*)_src;
00199 int stride = _stride>>(sizeof(pixel)-1);
00200 const int lt= src[-1-1*stride];
00201 LOAD_TOP_EDGE
00202 LOAD_LEFT_EDGE
00203
00204 src[0+0*stride]=
00205 src[1+2*stride]=(lt + t0 + 1)>>1;
00206 src[1+0*stride]=
00207 src[2+2*stride]=(t0 + t1 + 1)>>1;
00208 src[2+0*stride]=
00209 src[3+2*stride]=(t1 + t2 + 1)>>1;
00210 src[3+0*stride]=(t2 + t3 + 1)>>1;
00211 src[0+1*stride]=
00212 src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
00213 src[1+1*stride]=
00214 src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;
00215 src[2+1*stride]=
00216 src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;
00217 src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;
00218 src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
00219 src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
00220 }
00221
00222 static void FUNCC(pred4x4_vertical_left)(uint8_t *_src, const uint8_t *_topright, int _stride){
00223 pixel *src = (pixel*)_src;
00224 const pixel *topright = (const pixel*)_topright;
00225 int stride = _stride>>(sizeof(pixel)-1);
00226 LOAD_TOP_EDGE
00227 LOAD_TOP_RIGHT_EDGE
00228
00229 src[0+0*stride]=(t0 + t1 + 1)>>1;
00230 src[1+0*stride]=
00231 src[0+2*stride]=(t1 + t2 + 1)>>1;
00232 src[2+0*stride]=
00233 src[1+2*stride]=(t2 + t3 + 1)>>1;
00234 src[3+0*stride]=
00235 src[2+2*stride]=(t3 + t4+ 1)>>1;
00236 src[3+2*stride]=(t4 + t5+ 1)>>1;
00237 src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
00238 src[1+1*stride]=
00239 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
00240 src[2+1*stride]=
00241 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
00242 src[3+1*stride]=
00243 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
00244 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
00245 }
00246
00247 static void FUNCC(pred4x4_horizontal_up)(uint8_t *_src, const uint8_t *topright, int _stride){
00248 pixel *src = (pixel*)_src;
00249 int stride = _stride>>(sizeof(pixel)-1);
00250 LOAD_LEFT_EDGE
00251
00252 src[0+0*stride]=(l0 + l1 + 1)>>1;
00253 src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;
00254 src[2+0*stride]=
00255 src[0+1*stride]=(l1 + l2 + 1)>>1;
00256 src[3+0*stride]=
00257 src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;
00258 src[2+1*stride]=
00259 src[0+2*stride]=(l2 + l3 + 1)>>1;
00260 src[3+1*stride]=
00261 src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;
00262 src[3+2*stride]=
00263 src[1+3*stride]=
00264 src[0+3*stride]=
00265 src[2+2*stride]=
00266 src[2+3*stride]=
00267 src[3+3*stride]=l3;
00268 }
00269
00270 static void FUNCC(pred4x4_horizontal_down)(uint8_t *_src, const uint8_t *topright, int _stride){
00271 pixel *src = (pixel*)_src;
00272 int stride = _stride>>(sizeof(pixel)-1);
00273 const int lt= src[-1-1*stride];
00274 LOAD_TOP_EDGE
00275 LOAD_LEFT_EDGE
00276
00277 src[0+0*stride]=
00278 src[2+1*stride]=(lt + l0 + 1)>>1;
00279 src[1+0*stride]=
00280 src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;
00281 src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;
00282 src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;
00283 src[0+1*stride]=
00284 src[2+2*stride]=(l0 + l1 + 1)>>1;
00285 src[1+1*stride]=
00286 src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
00287 src[0+2*stride]=
00288 src[2+3*stride]=(l1 + l2+ 1)>>1;
00289 src[1+2*stride]=
00290 src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
00291 src[0+3*stride]=(l2 + l3 + 1)>>1;
00292 src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
00293 }
00294
00295 static void FUNCC(pred16x16_vertical)(uint8_t *_src, int _stride){
00296 int i;
00297 pixel *src = (pixel*)_src;
00298 int stride = _stride>>(sizeof(pixel)-1);
00299 const pixel4 a = AV_RN4PA(((pixel4*)(src-stride))+0);
00300 const pixel4 b = AV_RN4PA(((pixel4*)(src-stride))+1);
00301 const pixel4 c = AV_RN4PA(((pixel4*)(src-stride))+2);
00302 const pixel4 d = AV_RN4PA(((pixel4*)(src-stride))+3);
00303
00304 for(i=0; i<16; i++){
00305 AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
00306 AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
00307 AV_WN4PA(((pixel4*)(src+i*stride))+2, c);
00308 AV_WN4PA(((pixel4*)(src+i*stride))+3, d);
00309 }
00310 }
00311
00312 static void FUNCC(pred16x16_horizontal)(uint8_t *_src, int stride){
00313 int i;
00314 pixel *src = (pixel*)_src;
00315 stride >>= sizeof(pixel)-1;
00316
00317 for(i=0; i<16; i++){
00318 const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
00319
00320 AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
00321 AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
00322 AV_WN4PA(((pixel4*)(src+i*stride))+2, a);
00323 AV_WN4PA(((pixel4*)(src+i*stride))+3, a);
00324 }
00325 }
00326
00327 #define PREDICT_16x16_DC(v)\
00328 for(i=0; i<16; i++){\
00329 AV_WN4PA(src+ 0, v);\
00330 AV_WN4PA(src+ 4, v);\
00331 AV_WN4PA(src+ 8, v);\
00332 AV_WN4PA(src+12, v);\
00333 src += stride;\
00334 }
00335
00336 static void FUNCC(pred16x16_dc)(uint8_t *_src, int stride){
00337 int i, dc=0;
00338 pixel *src = (pixel*)_src;
00339 pixel4 dcsplat;
00340 stride >>= sizeof(pixel)-1;
00341
00342 for(i=0;i<16; i++){
00343 dc+= src[-1+i*stride];
00344 }
00345
00346 for(i=0;i<16; i++){
00347 dc+= src[i-stride];
00348 }
00349
00350 dcsplat = PIXEL_SPLAT_X4((dc+16)>>5);
00351 PREDICT_16x16_DC(dcsplat);
00352 }
00353
00354 static void FUNCC(pred16x16_left_dc)(uint8_t *_src, int stride){
00355 int i, dc=0;
00356 pixel *src = (pixel*)_src;
00357 pixel4 dcsplat;
00358 stride >>= sizeof(pixel)-1;
00359
00360 for(i=0;i<16; i++){
00361 dc+= src[-1+i*stride];
00362 }
00363
00364 dcsplat = PIXEL_SPLAT_X4((dc+8)>>4);
00365 PREDICT_16x16_DC(dcsplat);
00366 }
00367
00368 static void FUNCC(pred16x16_top_dc)(uint8_t *_src, int stride){
00369 int i, dc=0;
00370 pixel *src = (pixel*)_src;
00371 pixel4 dcsplat;
00372 stride >>= sizeof(pixel)-1;
00373
00374 for(i=0;i<16; i++){
00375 dc+= src[i-stride];
00376 }
00377
00378 dcsplat = PIXEL_SPLAT_X4((dc+8)>>4);
00379 PREDICT_16x16_DC(dcsplat);
00380 }
00381
00382 #define PRED16x16_X(n, v) \
00383 static void FUNCC(pred16x16_##n##_dc)(uint8_t *_src, int stride){\
00384 int i;\
00385 pixel *src = (pixel*)_src;\
00386 stride >>= sizeof(pixel)-1;\
00387 PREDICT_16x16_DC(PIXEL_SPLAT_X4(v));\
00388 }
00389
00390 PRED16x16_X(127, (1<<(BIT_DEPTH-1))-1);
00391 PRED16x16_X(128, (1<<(BIT_DEPTH-1))+0);
00392 PRED16x16_X(129, (1<<(BIT_DEPTH-1))+1);
00393
00394 static inline void FUNCC(pred16x16_plane_compat)(uint8_t *p_src, int p_stride, const int svq3, const int rv40){
00395 int i, j, k;
00396 int a;
00397 INIT_CLIP
00398 pixel *src = (pixel*)p_src;
00399 int stride = p_stride>>(sizeof(pixel)-1);
00400 const pixel * const src0 = src +7-stride;
00401 const pixel * src1 = src +8*stride-1;
00402 const pixel * src2 = src1-2*stride;
00403 int H = src0[1] - src0[-1];
00404 int V = src1[0] - src2[ 0];
00405 for(k=2; k<=8; ++k) {
00406 src1 += stride; src2 -= stride;
00407 H += k*(src0[k] - src0[-k]);
00408 V += k*(src1[0] - src2[ 0]);
00409 }
00410 if(svq3){
00411 H = ( 5*(H/4) ) / 16;
00412 V = ( 5*(V/4) ) / 16;
00413
00414
00415 i = H; H = V; V = i;
00416 }else if(rv40){
00417 H = ( H + (H>>2) ) >> 4;
00418 V = ( V + (V>>2) ) >> 4;
00419 }else{
00420 H = ( 5*H+32 ) >> 6;
00421 V = ( 5*V+32 ) >> 6;
00422 }
00423
00424 a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);
00425 for(j=16; j>0; --j) {
00426 int b = a;
00427 a += V;
00428 for(i=-16; i<0; i+=4) {
00429 src[16+i] = CLIP((b ) >> 5);
00430 src[17+i] = CLIP((b+ H) >> 5);
00431 src[18+i] = CLIP((b+2*H) >> 5);
00432 src[19+i] = CLIP((b+3*H) >> 5);
00433 b += 4*H;
00434 }
00435 src += stride;
00436 }
00437 }
00438
00439 static void FUNCC(pred16x16_plane)(uint8_t *src, int stride){
00440 FUNCC(pred16x16_plane_compat)(src, stride, 0, 0);
00441 }
00442
00443 static void FUNCC(pred8x8_vertical)(uint8_t *_src, int _stride){
00444 int i;
00445 pixel *src = (pixel*)_src;
00446 int stride = _stride>>(sizeof(pixel)-1);
00447 const pixel4 a= AV_RN4PA(((pixel4*)(src-stride))+0);
00448 const pixel4 b= AV_RN4PA(((pixel4*)(src-stride))+1);
00449
00450 for(i=0; i<8; i++){
00451 AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
00452 AV_WN4PA(((pixel4*)(src+i*stride))+1, b);
00453 }
00454 }
00455
00456 static void FUNCC(pred8x8_horizontal)(uint8_t *_src, int stride){
00457 int i;
00458 pixel *src = (pixel*)_src;
00459 stride >>= sizeof(pixel)-1;
00460
00461 for(i=0; i<8; i++){
00462 const pixel4 a = PIXEL_SPLAT_X4(src[-1+i*stride]);
00463 AV_WN4PA(((pixel4*)(src+i*stride))+0, a);
00464 AV_WN4PA(((pixel4*)(src+i*stride))+1, a);
00465 }
00466 }
00467
00468 #define PRED8x8_X(n, v)\
00469 static void FUNCC(pred8x8_##n##_dc)(uint8_t *_src, int stride){\
00470 int i;\
00471 const pixel4 a = PIXEL_SPLAT_X4(v);\
00472 pixel *src = (pixel*)_src;\
00473 stride >>= sizeof(pixel)-1;\
00474 for(i=0; i<8; i++){\
00475 AV_WN4PA(((pixel4*)(src+i*stride))+0, a);\
00476 AV_WN4PA(((pixel4*)(src+i*stride))+1, a);\
00477 }\
00478 }
00479
00480 PRED8x8_X(127, (1<<(BIT_DEPTH-1))-1);
00481 PRED8x8_X(128, (1<<(BIT_DEPTH-1))+0);
00482 PRED8x8_X(129, (1<<(BIT_DEPTH-1))+1);
00483
00484 static void FUNCC(pred8x8_left_dc)(uint8_t *_src, int stride){
00485 int i;
00486 int dc0, dc2;
00487 pixel4 dc0splat, dc2splat;
00488 pixel *src = (pixel*)_src;
00489 stride >>= sizeof(pixel)-1;
00490
00491 dc0=dc2=0;
00492 for(i=0;i<4; i++){
00493 dc0+= src[-1+i*stride];
00494 dc2+= src[-1+(i+4)*stride];
00495 }
00496 dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);
00497 dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
00498
00499 for(i=0; i<4; i++){
00500 AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
00501 AV_WN4PA(((pixel4*)(src+i*stride))+1, dc0splat);
00502 }
00503 for(i=4; i<8; i++){
00504 AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
00505 AV_WN4PA(((pixel4*)(src+i*stride))+1, dc2splat);
00506 }
00507 }
00508
00509 static void FUNCC(pred8x8_top_dc)(uint8_t *_src, int stride){
00510 int i;
00511 int dc0, dc1;
00512 pixel4 dc0splat, dc1splat;
00513 pixel *src = (pixel*)_src;
00514 stride >>= sizeof(pixel)-1;
00515
00516 dc0=dc1=0;
00517 for(i=0;i<4; i++){
00518 dc0+= src[i-stride];
00519 dc1+= src[4+i-stride];
00520 }
00521 dc0splat = PIXEL_SPLAT_X4((dc0 + 2)>>2);
00522 dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
00523
00524 for(i=0; i<4; i++){
00525 AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
00526 AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
00527 }
00528 for(i=4; i<8; i++){
00529 AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
00530 AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
00531 }
00532 }
00533
00534 static void FUNCC(pred8x8_dc)(uint8_t *_src, int stride){
00535 int i;
00536 int dc0, dc1, dc2;
00537 pixel4 dc0splat, dc1splat, dc2splat, dc3splat;
00538 pixel *src = (pixel*)_src;
00539 stride >>= sizeof(pixel)-1;
00540
00541 dc0=dc1=dc2=0;
00542 for(i=0;i<4; i++){
00543 dc0+= src[-1+i*stride] + src[i-stride];
00544 dc1+= src[4+i-stride];
00545 dc2+= src[-1+(i+4)*stride];
00546 }
00547 dc0splat = PIXEL_SPLAT_X4((dc0 + 4)>>3);
00548 dc1splat = PIXEL_SPLAT_X4((dc1 + 2)>>2);
00549 dc2splat = PIXEL_SPLAT_X4((dc2 + 2)>>2);
00550 dc3splat = PIXEL_SPLAT_X4((dc1 + dc2 + 4)>>3);
00551
00552 for(i=0; i<4; i++){
00553 AV_WN4PA(((pixel4*)(src+i*stride))+0, dc0splat);
00554 AV_WN4PA(((pixel4*)(src+i*stride))+1, dc1splat);
00555 }
00556 for(i=4; i<8; i++){
00557 AV_WN4PA(((pixel4*)(src+i*stride))+0, dc2splat);
00558 AV_WN4PA(((pixel4*)(src+i*stride))+1, dc3splat);
00559 }
00560 }
00561
00562
00563 static void FUNC(pred8x8_mad_cow_dc_l0t)(uint8_t *src, int stride){
00564 FUNCC(pred8x8_top_dc)(src, stride);
00565 FUNCC(pred4x4_dc)(src, NULL, stride);
00566 }
00567
00568 static void FUNC(pred8x8_mad_cow_dc_0lt)(uint8_t *src, int stride){
00569 FUNCC(pred8x8_dc)(src, stride);
00570 FUNCC(pred4x4_top_dc)(src, NULL, stride);
00571 }
00572
00573 static void FUNC(pred8x8_mad_cow_dc_l00)(uint8_t *src, int stride){
00574 FUNCC(pred8x8_left_dc)(src, stride);
00575 FUNCC(pred4x4_128_dc)(src + 4*stride , NULL, stride);
00576 FUNCC(pred4x4_128_dc)(src + 4*stride + 4*sizeof(pixel), NULL, stride);
00577 }
00578
00579 static void FUNC(pred8x8_mad_cow_dc_0l0)(uint8_t *src, int stride){
00580 FUNCC(pred8x8_left_dc)(src, stride);
00581 FUNCC(pred4x4_128_dc)(src , NULL, stride);
00582 FUNCC(pred4x4_128_dc)(src + 4*sizeof(pixel), NULL, stride);
00583 }
00584
00585 static void FUNCC(pred8x8_plane)(uint8_t *_src, int _stride){
00586 int j, k;
00587 int a;
00588 INIT_CLIP
00589 pixel *src = (pixel*)_src;
00590 int stride = _stride>>(sizeof(pixel)-1);
00591 const pixel * const src0 = src +3-stride;
00592 const pixel * src1 = src +4*stride-1;
00593 const pixel * src2 = src1-2*stride;
00594 int H = src0[1] - src0[-1];
00595 int V = src1[0] - src2[ 0];
00596 for(k=2; k<=4; ++k) {
00597 src1 += stride; src2 -= stride;
00598 H += k*(src0[k] - src0[-k]);
00599 V += k*(src1[0] - src2[ 0]);
00600 }
00601 H = ( 17*H+16 ) >> 5;
00602 V = ( 17*V+16 ) >> 5;
00603
00604 a = 16*(src1[0] + src2[8]+1) - 3*(V+H);
00605 for(j=8; j>0; --j) {
00606 int b = a;
00607 a += V;
00608 src[0] = CLIP((b ) >> 5);
00609 src[1] = CLIP((b+ H) >> 5);
00610 src[2] = CLIP((b+2*H) >> 5);
00611 src[3] = CLIP((b+3*H) >> 5);
00612 src[4] = CLIP((b+4*H) >> 5);
00613 src[5] = CLIP((b+5*H) >> 5);
00614 src[6] = CLIP((b+6*H) >> 5);
00615 src[7] = CLIP((b+7*H) >> 5);
00616 src += stride;
00617 }
00618 }
00619
00620 #define SRC(x,y) src[(x)+(y)*stride]
00621 #define PL(y) \
00622 const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
00623 #define PREDICT_8x8_LOAD_LEFT \
00624 const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
00625 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
00626 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
00627 const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2
00628
00629 #define PT(x) \
00630 const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
00631 #define PREDICT_8x8_LOAD_TOP \
00632 const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
00633 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
00634 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
00635 const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
00636 + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2
00637
00638 #define PTR(x) \
00639 t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
00640 #define PREDICT_8x8_LOAD_TOPRIGHT \
00641 int t8, t9, t10, t11, t12, t13, t14, t15; \
00642 if(has_topright) { \
00643 PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \
00644 t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \
00645 } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);
00646
00647 #define PREDICT_8x8_LOAD_TOPLEFT \
00648 const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2
00649
00650 #define PREDICT_8x8_DC(v) \
00651 int y; \
00652 for( y = 0; y < 8; y++ ) { \
00653 AV_WN4PA(((pixel4*)src)+0, v); \
00654 AV_WN4PA(((pixel4*)src)+1, v); \
00655 src += stride; \
00656 }
00657
00658 static void FUNCC(pred8x8l_128_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
00659 {
00660 pixel *src = (pixel*)_src;
00661 int stride = _stride>>(sizeof(pixel)-1);
00662
00663 PREDICT_8x8_DC(PIXEL_SPLAT_X4(1<<(BIT_DEPTH-1)));
00664 }
00665 static void FUNCC(pred8x8l_left_dc)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
00666 {
00667 pixel *src = (pixel*)_src;
00668 int stride = _stride>>(sizeof(pixel)-1);
00669
00670 PREDICT_8x8_LOAD_LEFT;
00671 const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3);
00672 PREDICT_8x8_DC(dc);
00673 }
00674 static void FUNCC(pred8x8l_top_dc)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00675 {
00676 pixel *src = (pixel*)p_src;
00677 int stride = p_stride>>(sizeof(pixel)-1);
00678
00679 PREDICT_8x8_LOAD_TOP;
00680 const pixel4 dc = PIXEL_SPLAT_X4((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3);
00681 PREDICT_8x8_DC(dc);
00682 }
00683 static void FUNCC(pred8x8l_dc)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00684 {
00685 pixel *src = (pixel*)p_src;
00686 int stride = p_stride>>(sizeof(pixel)-1);
00687
00688 PREDICT_8x8_LOAD_LEFT;
00689 PREDICT_8x8_LOAD_TOP;
00690 const pixel4 dc = PIXEL_SPLAT_X4((l0+l1+l2+l3+l4+l5+l6+l7
00691 +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4);
00692 PREDICT_8x8_DC(dc);
00693 }
00694 static void FUNCC(pred8x8l_horizontal)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00695 {
00696 pixel *src = (pixel*)p_src;
00697 int stride = p_stride>>(sizeof(pixel)-1);
00698 pixel4 a;
00699
00700 PREDICT_8x8_LOAD_LEFT;
00701 #define ROW(y) a = PIXEL_SPLAT_X4(l##y); \
00702 AV_WN4PA(src+y*stride, a); \
00703 AV_WN4PA(src+y*stride+4, a);
00704 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
00705 #undef ROW
00706 }
00707 static void FUNCC(pred8x8l_vertical)(uint8_t *_src, int has_topleft, int has_topright, int _stride)
00708 {
00709 int y;
00710 pixel *src = (pixel*)_src;
00711 int stride = _stride>>(sizeof(pixel)-1);
00712 pixel4 a, b;
00713
00714 PREDICT_8x8_LOAD_TOP;
00715 src[0] = t0;
00716 src[1] = t1;
00717 src[2] = t2;
00718 src[3] = t3;
00719 src[4] = t4;
00720 src[5] = t5;
00721 src[6] = t6;
00722 src[7] = t7;
00723 a = AV_RN4PA(((pixel4*)src)+0);
00724 b = AV_RN4PA(((pixel4*)src)+1);
00725 for( y = 1; y < 8; y++ ) {
00726 AV_WN4PA(((pixel4*)(src+y*stride))+0, a);
00727 AV_WN4PA(((pixel4*)(src+y*stride))+1, b);
00728 }
00729 }
00730 static void FUNCC(pred8x8l_down_left)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00731 {
00732 pixel *src = (pixel*)p_src;
00733 int stride = p_stride>>(sizeof(pixel)-1);
00734 PREDICT_8x8_LOAD_TOP;
00735 PREDICT_8x8_LOAD_TOPRIGHT;
00736 SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;
00737 SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;
00738 SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;
00739 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;
00740 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;
00741 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;
00742 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;
00743 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;
00744 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;
00745 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;
00746 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;
00747 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;
00748 SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;
00749 SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;
00750 SRC(7,7)= (t14 + 3*t15 + 2) >> 2;
00751 }
00752 static void FUNCC(pred8x8l_down_right)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00753 {
00754 pixel *src = (pixel*)p_src;
00755 int stride = p_stride>>(sizeof(pixel)-1);
00756 PREDICT_8x8_LOAD_TOP;
00757 PREDICT_8x8_LOAD_LEFT;
00758 PREDICT_8x8_LOAD_TOPLEFT;
00759 SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;
00760 SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;
00761 SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;
00762 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;
00763 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;
00764 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;
00765 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;
00766 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;
00767 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;
00768 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;
00769 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;
00770 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;
00771 SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;
00772 SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;
00773 SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;
00774 }
00775 static void FUNCC(pred8x8l_vertical_right)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00776 {
00777 pixel *src = (pixel*)p_src;
00778 int stride = p_stride>>(sizeof(pixel)-1);
00779 PREDICT_8x8_LOAD_TOP;
00780 PREDICT_8x8_LOAD_LEFT;
00781 PREDICT_8x8_LOAD_TOPLEFT;
00782 SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;
00783 SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;
00784 SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;
00785 SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;
00786 SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;
00787 SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;
00788 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;
00789 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;
00790 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;
00791 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;
00792 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;
00793 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;
00794 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;
00795 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;
00796 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;
00797 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;
00798 SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;
00799 SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;
00800 SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;
00801 SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;
00802 SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;
00803 SRC(7,0)= (t6 + t7 + 1) >> 1;
00804 }
00805 static void FUNCC(pred8x8l_horizontal_down)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00806 {
00807 pixel *src = (pixel*)p_src;
00808 int stride = p_stride>>(sizeof(pixel)-1);
00809 PREDICT_8x8_LOAD_TOP;
00810 PREDICT_8x8_LOAD_LEFT;
00811 PREDICT_8x8_LOAD_TOPLEFT;
00812 SRC(0,7)= (l6 + l7 + 1) >> 1;
00813 SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;
00814 SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;
00815 SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;
00816 SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;
00817 SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;
00818 SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;
00819 SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;
00820 SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;
00821 SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;
00822 SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;
00823 SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;
00824 SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;
00825 SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;
00826 SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;
00827 SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;
00828 SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;
00829 SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;
00830 SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;
00831 SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;
00832 SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;
00833 SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;
00834 }
00835 static void FUNCC(pred8x8l_vertical_left)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00836 {
00837 pixel *src = (pixel*)p_src;
00838 int stride = p_stride>>(sizeof(pixel)-1);
00839 PREDICT_8x8_LOAD_TOP;
00840 PREDICT_8x8_LOAD_TOPRIGHT;
00841 SRC(0,0)= (t0 + t1 + 1) >> 1;
00842 SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;
00843 SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;
00844 SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;
00845 SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;
00846 SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;
00847 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;
00848 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;
00849 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;
00850 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;
00851 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;
00852 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;
00853 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;
00854 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;
00855 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;
00856 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;
00857 SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;
00858 SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;
00859 SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;
00860 SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;
00861 SRC(7,6)= (t10 + t11 + 1) >> 1;
00862 SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;
00863 }
00864 static void FUNCC(pred8x8l_horizontal_up)(uint8_t *p_src, int has_topleft, int has_topright, int p_stride)
00865 {
00866 pixel *src = (pixel*)p_src;
00867 int stride = p_stride>>(sizeof(pixel)-1);
00868 PREDICT_8x8_LOAD_LEFT;
00869 SRC(0,0)= (l0 + l1 + 1) >> 1;
00870 SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;
00871 SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;
00872 SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;
00873 SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;
00874 SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;
00875 SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;
00876 SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;
00877 SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;
00878 SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;
00879 SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;
00880 SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;
00881 SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;
00882 SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;
00883 SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=
00884 SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=
00885 SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=
00886 SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;
00887 }
00888 #undef PREDICT_8x8_LOAD_LEFT
00889 #undef PREDICT_8x8_LOAD_TOP
00890 #undef PREDICT_8x8_LOAD_TOPLEFT
00891 #undef PREDICT_8x8_LOAD_TOPRIGHT
00892 #undef PREDICT_8x8_DC
00893 #undef PTR
00894 #undef PT
00895 #undef PL
00896 #undef SRC
00897
00898 static void FUNCC(pred4x4_vertical_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
00899 int i;
00900 pixel *pix = (pixel*)p_pix;
00901 const dctcoef *block = (const dctcoef*)p_block;
00902 stride >>= sizeof(pixel)-1;
00903 pix -= stride;
00904 for(i=0; i<4; i++){
00905 pixel v = pix[0];
00906 pix[1*stride]= v += block[0];
00907 pix[2*stride]= v += block[4];
00908 pix[3*stride]= v += block[8];
00909 pix[4*stride]= v + block[12];
00910 pix++;
00911 block++;
00912 }
00913 }
00914
00915 static void FUNCC(pred4x4_horizontal_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
00916 int i;
00917 pixel *pix = (pixel*)p_pix;
00918 const dctcoef *block = (const dctcoef*)p_block;
00919 stride >>= sizeof(pixel)-1;
00920 for(i=0; i<4; i++){
00921 pixel v = pix[-1];
00922 pix[0]= v += block[0];
00923 pix[1]= v += block[1];
00924 pix[2]= v += block[2];
00925 pix[3]= v + block[3];
00926 pix+= stride;
00927 block+= 4;
00928 }
00929 }
00930
00931 static void FUNCC(pred8x8l_vertical_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
00932 int i;
00933 pixel *pix = (pixel*)p_pix;
00934 const dctcoef *block = (const dctcoef*)p_block;
00935 stride >>= sizeof(pixel)-1;
00936 pix -= stride;
00937 for(i=0; i<8; i++){
00938 pixel v = pix[0];
00939 pix[1*stride]= v += block[0];
00940 pix[2*stride]= v += block[8];
00941 pix[3*stride]= v += block[16];
00942 pix[4*stride]= v += block[24];
00943 pix[5*stride]= v += block[32];
00944 pix[6*stride]= v += block[40];
00945 pix[7*stride]= v += block[48];
00946 pix[8*stride]= v + block[56];
00947 pix++;
00948 block++;
00949 }
00950 }
00951
00952 static void FUNCC(pred8x8l_horizontal_add)(uint8_t *p_pix, const DCTELEM *p_block, int stride){
00953 int i;
00954 pixel *pix = (pixel*)p_pix;
00955 const dctcoef *block = (const dctcoef*)p_block;
00956 stride >>= sizeof(pixel)-1;
00957 for(i=0; i<8; i++){
00958 pixel v = pix[-1];
00959 pix[0]= v += block[0];
00960 pix[1]= v += block[1];
00961 pix[2]= v += block[2];
00962 pix[3]= v += block[3];
00963 pix[4]= v += block[4];
00964 pix[5]= v += block[5];
00965 pix[6]= v += block[6];
00966 pix[7]= v + block[7];
00967 pix+= stride;
00968 block+= 8;
00969 }
00970 }
00971
00972 static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
00973 int i;
00974 for(i=0; i<16; i++)
00975 FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
00976 }
00977
00978 static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
00979 int i;
00980 for(i=0; i<16; i++)
00981 FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
00982 }
00983
00984 static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
00985 int i;
00986 for(i=0; i<4; i++)
00987 FUNCC(pred4x4_vertical_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
00988 }
00989
00990 static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
00991 int i;
00992 for(i=0; i<4; i++)
00993 FUNCC(pred4x4_horizontal_add)(pix + block_offset[i], block + i*16*sizeof(pixel), stride);
00994 }
