libavcodec/h264qpel_template.c

   1 /*
   2  * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
   3  * Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at>
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * Libav is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with Libav; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 #include "libavutil/common.h"
  23 #include "bit_depth_template.c"
  24 #include "hpel_template.c"
  25
  26 static inline void FUNC(copy_block2)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
  27 {
  28     int i;
  29     for(i=0; i<h; i++)
  30     {
  31         AV_WN2P(dst   , AV_RN2P(src   ));
  32         dst+=dstStride;
  33         src+=srcStride;
  34     }
  35 }
  36
  37 static inline void FUNC(copy_block4)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
  38 {
  39     int i;
  40     for(i=0; i<h; i++)
  41     {
  42         AV_WN4P(dst   , AV_RN4P(src   ));
  43         dst+=dstStride;
  44         src+=srcStride;
  45     }
  46 }
  47
  48 static inline void FUNC(copy_block8)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
  49 {
  50     int i;
  51     for(i=0; i<h; i++)
  52     {
  53         AV_WN4P(dst                , AV_RN4P(src                ));
  54         AV_WN4P(dst+4*sizeof(pixel), AV_RN4P(src+4*sizeof(pixel)));
  55         dst+=dstStride;
  56         src+=srcStride;
  57     }
  58 }
  59
  60 static inline void FUNC(copy_block16)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
  61 {
  62     int i;
  63     for(i=0; i<h; i++)
  64     {
  65         AV_WN4P(dst                 , AV_RN4P(src                 ));
  66         AV_WN4P(dst+ 4*sizeof(pixel), AV_RN4P(src+ 4*sizeof(pixel)));
  67         AV_WN4P(dst+ 8*sizeof(pixel), AV_RN4P(src+ 8*sizeof(pixel)));
  68         AV_WN4P(dst+12*sizeof(pixel), AV_RN4P(src+12*sizeof(pixel)));
  69         dst+=dstStride;
  70         src+=srcStride;
  71     }
  72 }
  73
  74 #define H264_LOWPASS(OPNAME, OP, OP2) \
  75 static av_unused void FUNC(OPNAME ## h264_qpel2_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
  76     const int h=2;\
  77     INIT_CLIP\
  78     int i;\
  79     pixel *dst = (pixel*)_dst;\
  80     pixel *src = (pixel*)_src;\
  81     dstStride /= sizeof(pixel);\
  82     srcStride /= sizeof(pixel);\
  83     for(i=0; i<h; i++)\
  84     {\
  85         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
  86         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
  87         dst+=dstStride;\
  88         src+=srcStride;\
  89     }\
  90 }\
  91 \
  92 static av_unused void FUNC(OPNAME ## h264_qpel2_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
  93     const int w=2;\
  94     INIT_CLIP\
  95     int i;\
  96     pixel *dst = (pixel*)_dst;\
  97     pixel *src = (pixel*)_src;\
  98     dstStride /= sizeof(pixel);\
  99     srcStride /= sizeof(pixel);\
 100     for(i=0; i<w; i++)\
 101     {\
 102         const int srcB= src[-2*srcStride];\
 103         const int srcA= src[-1*srcStride];\
 104         const int src0= src[0 *srcStride];\
 105         const int src1= src[1 *srcStride];\
 106         const int src2= src[2 *srcStride];\
 107         const int src3= src[3 *srcStride];\
 108         const int src4= src[4 *srcStride];\
 109         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
 110         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
 111         dst++;\
 112         src++;\
 113     }\
 114 }\
 115 \
 116 static av_unused void FUNC(OPNAME ## h264_qpel2_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\
 117     const int h=2;\
 118     const int w=2;\
 119     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
 120     INIT_CLIP\
 121     int i;\
 122     pixel *dst = (pixel*)_dst;\
 123     pixel *src = (pixel*)_src;\
 124     dstStride /= sizeof(pixel);\
 125     srcStride /= sizeof(pixel);\
 126     src -= 2*srcStride;\
 127     for(i=0; i<h+5; i++)\
 128     {\
 129         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
 130         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
 131         tmp+=tmpStride;\
 132         src+=srcStride;\
 133     }\
 134     tmp -= tmpStride*(h+5-2);\
 135     for(i=0; i<w; i++)\
 136     {\
 137         const int tmpB= tmp[-2*tmpStride] - pad;\
 138         const int tmpA= tmp[-1*tmpStride] - pad;\
 139         const int tmp0= tmp[0 *tmpStride] - pad;\
 140         const int tmp1= tmp[1 *tmpStride] - pad;\
 141         const int tmp2= tmp[2 *tmpStride] - pad;\
 142         const int tmp3= tmp[3 *tmpStride] - pad;\
 143         const int tmp4= tmp[4 *tmpStride] - pad;\
 144         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
 145         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
 146         dst++;\
 147         tmp++;\
 148     }\
 149 }\
 150 static void FUNC(OPNAME ## h264_qpel4_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
 151     const int h=4;\
 152     INIT_CLIP\
 153     int i;\
 154     pixel *dst = (pixel*)_dst;\
 155     pixel *src = (pixel*)_src;\
 156     dstStride /= sizeof(pixel);\
 157     srcStride /= sizeof(pixel);\
 158     for(i=0; i<h; i++)\
 159     {\
 160         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
 161         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
 162         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]));\
 163         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]));\
 164         dst+=dstStride;\
 165         src+=srcStride;\
 166     }\
 167 }\
 168 \
 169 static void FUNC(OPNAME ## h264_qpel4_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
 170     const int w=4;\
 171     INIT_CLIP\
 172     int i;\
 173     pixel *dst = (pixel*)_dst;\
 174     pixel *src = (pixel*)_src;\
 175     dstStride /= sizeof(pixel);\
 176     srcStride /= sizeof(pixel);\
 177     for(i=0; i<w; i++)\
 178     {\
 179         const int srcB= src[-2*srcStride];\
 180         const int srcA= src[-1*srcStride];\
 181         const int src0= src[0 *srcStride];\
 182         const int src1= src[1 *srcStride];\
 183         const int src2= src[2 *srcStride];\
 184         const int src3= src[3 *srcStride];\
 185         const int src4= src[4 *srcStride];\
 186         const int src5= src[5 *srcStride];\
 187         const int src6= src[6 *srcStride];\
 188         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
 189         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
 190         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
 191         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
 192         dst++;\
 193         src++;\
 194     }\
 195 }\
 196 \
 197 static void FUNC(OPNAME ## h264_qpel4_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\
 198     const int h=4;\
 199     const int w=4;\
 200     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
 201     INIT_CLIP\
 202     int i;\
 203     pixel *dst = (pixel*)_dst;\
 204     pixel *src = (pixel*)_src;\
 205     dstStride /= sizeof(pixel);\
 206     srcStride /= sizeof(pixel);\
 207     src -= 2*srcStride;\
 208     for(i=0; i<h+5; i++)\
 209     {\
 210         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
 211         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
 212         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]) + pad;\
 213         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]) + pad;\
 214         tmp+=tmpStride;\
 215         src+=srcStride;\
 216     }\
 217     tmp -= tmpStride*(h+5-2);\
 218     for(i=0; i<w; i++)\
 219     {\
 220         const int tmpB= tmp[-2*tmpStride] - pad;\
 221         const int tmpA= tmp[-1*tmpStride] - pad;\
 222         const int tmp0= tmp[0 *tmpStride] - pad;\
 223         const int tmp1= tmp[1 *tmpStride] - pad;\
 224         const int tmp2= tmp[2 *tmpStride] - pad;\
 225         const int tmp3= tmp[3 *tmpStride] - pad;\
 226         const int tmp4= tmp[4 *tmpStride] - pad;\
 227         const int tmp5= tmp[5 *tmpStride] - pad;\
 228         const int tmp6= tmp[6 *tmpStride] - pad;\
 229         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
 230         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
 231         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
 232         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
 233         dst++;\
 234         tmp++;\
 235     }\
 236 }\
 237 \
 238 static void FUNC(OPNAME ## h264_qpel8_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
 239     const int h=8;\
 240     INIT_CLIP\
 241     int i;\
 242     pixel *dst = (pixel*)_dst;\
 243     pixel *src = (pixel*)_src;\
 244     dstStride /= sizeof(pixel);\
 245     srcStride /= sizeof(pixel);\
 246     for(i=0; i<h; i++)\
 247     {\
 248         OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]));\
 249         OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]));\
 250         OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]));\
 251         OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]));\
 252         OP(dst[4], (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]));\
 253         OP(dst[5], (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]));\
 254         OP(dst[6], (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]));\
 255         OP(dst[7], (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]));\
 256         dst+=dstStride;\
 257         src+=srcStride;\
 258     }\
 259 }\
 260 \
 261 static void FUNC(OPNAME ## h264_qpel8_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
 262     const int w=8;\
 263     INIT_CLIP\
 264     int i;\
 265     pixel *dst = (pixel*)_dst;\
 266     pixel *src = (pixel*)_src;\
 267     dstStride /= sizeof(pixel);\
 268     srcStride /= sizeof(pixel);\
 269     for(i=0; i<w; i++)\
 270     {\
 271         const int srcB= src[-2*srcStride];\
 272         const int srcA= src[-1*srcStride];\
 273         const int src0= src[0 *srcStride];\
 274         const int src1= src[1 *srcStride];\
 275         const int src2= src[2 *srcStride];\
 276         const int src3= src[3 *srcStride];\
 277         const int src4= src[4 *srcStride];\
 278         const int src5= src[5 *srcStride];\
 279         const int src6= src[6 *srcStride];\
 280         const int src7= src[7 *srcStride];\
 281         const int src8= src[8 *srcStride];\
 282         const int src9= src[9 *srcStride];\
 283         const int src10=src[10*srcStride];\
 284         OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
 285         OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
 286         OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
 287         OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
 288         OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
 289         OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
 290         OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
 291         OP(dst[7*dstStride], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
 292         dst++;\
 293         src++;\
 294     }\
 295 }\
 296 \
 297 static void FUNC(OPNAME ## h264_qpel8_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\
 298     const int h=8;\
 299     const int w=8;\
 300     const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
 301     INIT_CLIP\
 302     int i;\
 303     pixel *dst = (pixel*)_dst;\
 304     pixel *src = (pixel*)_src;\
 305     dstStride /= sizeof(pixel);\
 306     srcStride /= sizeof(pixel);\
 307     src -= 2*srcStride;\
 308     for(i=0; i<h+5; i++)\
 309     {\
 310         tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]) + pad;\
 311         tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]) + pad;\
 312         tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]) + pad;\
 313         tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]) + pad;\
 314         tmp[4]= (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]) + pad;\
 315         tmp[5]= (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]) + pad;\
 316         tmp[6]= (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]) + pad;\
 317         tmp[7]= (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]) + pad;\
 318         tmp+=tmpStride;\
 319         src+=srcStride;\
 320     }\
 321     tmp -= tmpStride*(h+5-2);\
 322     for(i=0; i<w; i++)\
 323     {\
 324         const int tmpB= tmp[-2*tmpStride] - pad;\
 325         const int tmpA= tmp[-1*tmpStride] - pad;\
 326         const int tmp0= tmp[0 *tmpStride] - pad;\
 327         const int tmp1= tmp[1 *tmpStride] - pad;\
 328         const int tmp2= tmp[2 *tmpStride] - pad;\
 329         const int tmp3= tmp[3 *tmpStride] - pad;\
 330         const int tmp4= tmp[4 *tmpStride] - pad;\
 331         const int tmp5= tmp[5 *tmpStride] - pad;\
 332         const int tmp6= tmp[6 *tmpStride] - pad;\
 333         const int tmp7= tmp[7 *tmpStride] - pad;\
 334         const int tmp8= tmp[8 *tmpStride] - pad;\
 335         const int tmp9= tmp[9 *tmpStride] - pad;\
 336         const int tmp10=tmp[10*tmpStride] - pad;\
 337         OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
 338         OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
 339         OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
 340         OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
 341         OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\
 342         OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\
 343         OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\
 344         OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\
 345         dst++;\
 346         tmp++;\
 347     }\
 348 }\
 349 \
 350 static void FUNC(OPNAME ## h264_qpel16_v_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
 351     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\
 352     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
 353     src += 8*srcStride;\
 354     dst += 8*dstStride;\
 355     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst                , src                , dstStride, srcStride);\
 356     FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
 357 }\
 358 \
 359 static void FUNC(OPNAME ## h264_qpel16_h_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
 360     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\
 361     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
 362     src += 8*srcStride;\
 363     dst += 8*dstStride;\
 364     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst                , src                , dstStride, srcStride);\
 365     FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
 366 }\
 367 \
 368 static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
 369     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\
 370     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
 371     src += 8*srcStride;\
 372     dst += 8*dstStride;\
 373     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst                , tmp  , src                , dstStride, tmpStride, srcStride);\
 374     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
 375 }\
 376
 377 #define H264_MC(OPNAME, SIZE) \
 378 static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\
 379     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\
 380 }\
 381 \
 382 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\
 383     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
 384     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
 385     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\
 386 }\
 387 \
 388 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\
 389     FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\
 390 }\
 391 \
 392 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\
 393     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
 394     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
 395     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\
 396 }\
 397 \
 398 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\
 399     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 400     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 401     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
 402     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
 403     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 404     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 405 }\
 406 \
 407 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\
 408     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 409     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 410     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
 411     FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\
 412 }\
 413 \
 414 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\
 415     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 416     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 417     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
 418     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
 419     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 420     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 421 }\
 422 \
 423 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\
 424     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 425     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 426     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
 427     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
 428     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
 429     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
 430     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 431     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 432 }\
 433 \
 434 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\
 435     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 436     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 437     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
 438     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
 439     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
 440     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
 441     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 442     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 443 }\
 444 \
 445 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\
 446     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 447     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 448     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
 449     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
 450     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
 451     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
 452     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 453     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 454 }\
 455 \
 456 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\
 457     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 458     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 459     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
 460     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
 461     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
 462     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
 463     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 464     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 465 }\
 466 \
 467 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\
 468     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
 469     FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\
 470 }\
 471 \
 472 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\
 473     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
 474     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
 475     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
 476     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
 477     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
 478     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 479 }\
 480 \
 481 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\
 482     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
 483     uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
 484     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
 485     FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
 486     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
 487     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 488 }\
 489 \
 490 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\
 491     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 492     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 493     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
 494     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
 495     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
 496     FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel),  stride, SIZE + 5);\
 497     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 498     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
 499     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 500 }\
 501 \
 502 static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\
 503     uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
 504     uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
 505     int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
 506     uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
 507     uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
 508     FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel),  stride, SIZE + 5);\
 509     FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
 510     FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
 511     FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
 512 }\
 513
 514 #define op_avg(a, b)  a = (((a)+CLIP(((b) + 16)>>5)+1)>>1)
 515 //#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)
 516 #define op_put(a, b)  a = CLIP(((b) + 16)>>5)
 517 #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)
 518 #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)
 519
 520 H264_LOWPASS(put_       , op_put, op2_put)
 521 H264_LOWPASS(avg_       , op_avg, op2_avg)
 522 H264_MC(put_, 2)
 523 H264_MC(put_, 4)
 524 H264_MC(put_, 8)
 525 H264_MC(put_, 16)
 526 H264_MC(avg_, 4)
 527 H264_MC(avg_, 8)
 528 H264_MC(avg_, 16)
 529
 530 #undef op_avg
 531 #undef op_put
 532 #undef op2_avg
 533 #undef op2_put
 534
 535 #if BIT_DEPTH == 8
 536 #   define put_h264_qpel8_mc00_8_c  ff_put_pixels8x8_8_c
 537 #   define avg_h264_qpel8_mc00_8_c  ff_avg_pixels8x8_8_c
 538 #   define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c
 539 #   define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c
 540 #elif BIT_DEPTH == 9
 541 #   define put_h264_qpel8_mc00_9_c  ff_put_pixels8x8_9_c
 542 #   define avg_h264_qpel8_mc00_9_c  ff_avg_pixels8x8_9_c
 543 #   define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c
 544 #   define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c
 545 #elif BIT_DEPTH == 10
 546 #   define put_h264_qpel8_mc00_10_c  ff_put_pixels8x8_10_c
 547 #   define avg_h264_qpel8_mc00_10_c  ff_avg_pixels8x8_10_c
 548 #   define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c
 549 #   define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c
 550 #endif