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