26 #   define FUNC(n) AV_JOIN(n ## _simple_, BITS) 
   27 #   define PIXEL_SHIFT (BITS >> 4) 
   29 #   define FUNC(n) n ## _complex 
   30 #   define PIXEL_SHIFT h->pixel_shift 
   43     const int mb_x    = h->mb_x;
 
   44     const int mb_y    = h->mb_y;
 
   45     const int mb_xy   = h->mb_xy;
 
   46     const int mb_type = h->cur_pic.mb_type[
mb_xy];
 
   47     uint8_t *dest_y, *dest_cb, *dest_cr;
 
   51     const int transform_bypass = !
SIMPLE && (h->qscale == 0 && h->sps.transform_bypass);
 
   55     const int block_h   = 16 >> h->chroma_y_shift;
 
   58     dest_y  = h->cur_pic.f.data[0] + ((mb_x << 
PIXEL_SHIFT)     + mb_y * h->linesize)  * 16;
 
   59     dest_cb = h->cur_pic.f.data[1] +  (mb_x << 
PIXEL_SHIFT) * 8 + mb_y * h->uvlinesize * block_h;
 
   60     dest_cr = h->cur_pic.f.data[2] +  (mb_x << 
PIXEL_SHIFT) * 8 + mb_y * h->uvlinesize * block_h;
 
   62     h->vdsp.prefetch(dest_y  + (h->mb_x & 3) * 4 * h->linesize   + (64 << 
PIXEL_SHIFT), h->linesize,       4);
 
   63     h->vdsp.prefetch(dest_cb + (h->mb_x & 7)     * h->uvlinesize + (64 << 
PIXEL_SHIFT), dest_cr - dest_cb, 2);
 
   65     h->list_counts[
mb_xy] = h->list_count;
 
   68         linesize     = h->mb_linesize = h->linesize * 2;
 
   69         uvlinesize   = h->mb_uvlinesize = h->uvlinesize * 2;
 
   70         block_offset = &h->block_offset[48];
 
   72             dest_y  -= h->linesize * 15;
 
   73             dest_cb -= h->uvlinesize * (block_h - 1);
 
   74             dest_cr -= h->uvlinesize * (block_h - 1);
 
   78             for (list = 0; list < h->list_count; list++) {
 
   82                     int8_t *ref = &h->ref_cache[list][
scan8[0]];
 
   85                     for (i = 0; i < 16; i += 4) {
 
   86                         int ref = h->ref_cache[list][
scan8[i]];
 
   89                                            8, (16 + ref) ^ (h->mb_y & 1), 1);
 
   95         linesize   = h->mb_linesize   = h->linesize;
 
   96         uvlinesize = h->mb_uvlinesize = h->uvlinesize;
 
  101         const int bit_depth = h->sps.bit_depth_luma;
 
  108             for (i = 0; i < 16; i++) {
 
  109                 uint16_t *tmp_y = (uint16_t *)(dest_y + i * linesize);
 
  110                 for (j = 0; j < 16; j++)
 
  111                     tmp_y[j] = 
get_bits(&gb, bit_depth);
 
  114                 if (!h->sps.chroma_format_idc) {
 
  115                     for (i = 0; i < block_h; i++) {
 
  116                         uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
 
  117                         uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
 
  118                         for (j = 0; j < 8; j++) {
 
  119                             tmp_cb[j] = tmp_cr[j] = 1 << (bit_depth - 1);
 
  123                     for (i = 0; i < block_h; i++) {
 
  124                         uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
 
  125                         for (j = 0; j < 8; j++)
 
  126                             tmp_cb[j] = 
get_bits(&gb, bit_depth);
 
  128                     for (i = 0; i < block_h; i++) {
 
  129                         uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
 
  130                         for (j = 0; j < 8; j++)
 
  131                             tmp_cr[j] = 
get_bits(&gb, bit_depth);
 
  136             for (i = 0; i < 16; i++)
 
  137                 memcpy(dest_y + i * linesize, h->intra_pcm_ptr + i * 16, 16);
 
  139                 if (!h->sps.chroma_format_idc) {
 
  140                     for (i = 0; i < 8; i++) {
 
  141                         memset(dest_cb + i * uvlinesize, 1 << (bit_depth - 1), 8);
 
  142                         memset(dest_cr + i * uvlinesize, 1 << (bit_depth - 1), 8);
 
  145                     const uint8_t *src_cb = h->intra_pcm_ptr + 256;
 
  146                     const uint8_t *src_cr = h->intra_pcm_ptr + 256 + block_h * 8;
 
  147                     for (i = 0; i < block_h; i++) {
 
  148                         memcpy(dest_cb + i * uvlinesize, src_cb + i * 8, 8);
 
  149                         memcpy(dest_cr + i * uvlinesize, src_cr + i * 8, 8);
 
  156             if (h->deblocking_filter)
 
  161                 h->hpc.pred8x8[h->chroma_pred_mode](dest_cb, 
uvlinesize);
 
  162                 h->hpc.pred8x8[h->chroma_pred_mode](dest_cr, 
uvlinesize);
 
  167                                       block_offset, linesize, dest_y, 0);
 
  169             if (h->deblocking_filter)
 
  172         } 
else if (is_h264) {
 
  174                 FUNC(hl_motion_422)(h, dest_y, dest_cb, dest_cr,
 
  175                               h->me.qpel_put, h->h264chroma.put_h264_chroma_pixels_tab,
 
  176                               h->me.qpel_avg, h->h264chroma.avg_h264_chroma_pixels_tab,
 
  177                               h->h264dsp.weight_h264_pixels_tab,
 
  178                               h->h264dsp.biweight_h264_pixels_tab);
 
  180                 FUNC(hl_motion_420)(h, dest_y, dest_cb, dest_cr,
 
  181                               h->me.qpel_put, h->h264chroma.put_h264_chroma_pixels_tab,
 
  182                               h->me.qpel_avg, h->h264chroma.avg_h264_chroma_pixels_tab,
 
  183                               h->h264dsp.weight_h264_pixels_tab,
 
  184                               h->h264dsp.biweight_h264_pixels_tab);
 
  193             uint8_t *dest[2] = { dest_cb, dest_cr };
 
  194             if (transform_bypass) {
 
  195                 if (
IS_INTRA(mb_type) && h->sps.profile_idc == 244 &&
 
  198                     h->hpc.pred8x8_add[h->chroma_pred_mode](dest[0],
 
  202                     h->hpc.pred8x8_add[h->chroma_pred_mode](dest[1],
 
  207                     idct_add = h->h264dsp.h264_add_pixels4_clear;
 
  208                     for (j = 1; j < 3; j++) {
 
  209                         for (i = j * 16; i < j * 16 + 4; i++)
 
  210                             if (h->non_zero_count_cache[
scan8[i]] ||
 
  212                                 idct_add(dest[j - 1] + block_offset[i],
 
  216                             for (i = j * 16 + 4; i < j * 16 + 8; i++)
 
  217                                 if (h->non_zero_count_cache[
scan8[i + 4]] ||
 
  219                                     idct_add(dest[j - 1] + block_offset[i + 4],
 
  229                         qp[0] = h->chroma_qp[0] + 3;
 
  230                         qp[1] = h->chroma_qp[1] + 3;
 
  232                         qp[0] = h->chroma_qp[0];
 
  233                         qp[1] = h->chroma_qp[1];
 
  236                         h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16 * 16 * 1 << 
PIXEL_SHIFT),
 
  237                                                                h->dequant4_coeff[
IS_INTRA(mb_type) ? 1 : 4][qp[0]][0]);
 
  239                         h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16 * 16 * 2 << 
PIXEL_SHIFT),
 
  240                                                                h->dequant4_coeff[
IS_INTRA(mb_type) ? 2 : 5][qp[1]][0]);
 
  241                     h->h264dsp.h264_idct_add8(dest, block_offset,
 
  243                                               h->non_zero_count_cache);
 
  244                 } 
else if (CONFIG_SVQ3_DECODER) {
 
  245                     h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16 * 16 * 1,
 
  246                                                            h->dequant4_coeff[
IS_INTRA(mb_type) ? 1 : 4][h->chroma_qp[0]][0]);
 
  247                     h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16 * 16 * 2,
 
  248                                                            h->dequant4_coeff[
IS_INTRA(mb_type) ? 2 : 5][h->chroma_qp[1]][0]);
 
  249                     for (j = 1; j < 3; j++) {
 
  250                         for (i = j * 16; i < j * 16 + 4; i++)
 
  251                             if (h->non_zero_count_cache[
scan8[i]] || h->mb[i * 16]) {
 
  252                                 uint8_t *
const ptr = dest[j - 1] + block_offset[i];
 
  264 #if !SIMPLE || BITS == 8 
  272     const int mb_x    = h->mb_x;
 
  273     const int mb_y    = h->mb_y;
 
  274     const int mb_xy   = h->mb_xy;
 
  275     const int mb_type = h->cur_pic.mb_type[
mb_xy];
 
  280     const int transform_bypass = !
SIMPLE && (h->qscale == 0 && h->sps.transform_bypass);
 
  283     for (p = 0; p < plane_count; p++) {
 
  284         dest[p] = h->cur_pic.f.data[p] +
 
  286         h->vdsp.prefetch(dest[p] + (h->mb_x & 3) * 4 * h->linesize + (64 << 
PIXEL_SHIFT),
 
  290     h->list_counts[
mb_xy] = h->list_count;
 
  293         linesize     = h->mb_linesize = h->mb_uvlinesize = h->linesize * 2;
 
  294         block_offset = &h->block_offset[48];
 
  296             for (p = 0; p < 3; p++)
 
  297                 dest[p] -= h->linesize * 15;
 
  300             for (list = 0; list < h->list_count; list++) {
 
  304                     int8_t *ref = &h->ref_cache[list][
scan8[0]];
 
  307                     for (i = 0; i < 16; i += 4) {
 
  308                         int ref = h->ref_cache[list][
scan8[i]];
 
  311                                            8, (16 + ref) ^ (h->mb_y & 1), 1);
 
  317         linesize = h->mb_linesize = h->mb_uvlinesize = h->linesize;
 
  322             const int bit_depth = h->sps.bit_depth_luma;
 
  326             for (p = 0; p < plane_count; p++)
 
  327                 for (i = 0; i < 16; i++) {
 
  328                     uint16_t *tmp = (uint16_t *)(dest[p] + i * linesize);
 
  329                     for (j = 0; j < 16; j++)
 
  333             for (p = 0; p < plane_count; p++)
 
  334                 for (i = 0; i < 16; i++)
 
  335                     memcpy(dest[p] + i * linesize,
 
  336                            h->intra_pcm_ptr + p * 256 + i * 16, 16);
 
  340             if (h->deblocking_filter)
 
  344             for (p = 0; p < plane_count; p++)
 
  347                                           block_offset, linesize, dest[p], p);
 
  349             if (h->deblocking_filter)
 
  353             FUNC(hl_motion_444)(h, dest[0], dest[1], dest[2],
 
  354                       h->me.qpel_put, h->h264chroma.put_h264_chroma_pixels_tab,
 
  355                       h->me.qpel_avg, h->h264chroma.avg_h264_chroma_pixels_tab,
 
  356                       h->h264dsp.weight_h264_pixels_tab,
 
  357                       h->h264dsp.biweight_h264_pixels_tab);
 
  360         for (p = 0; p < plane_count; p++)