1 /*****************************************************************************
2 * deblock.c: deblocking
3 *****************************************************************************
4 * Copyright (C) 2003-2011 x264 project
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
8 * Fiona Glaser <fiona@x264.com>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
24 * This program is also available under a commercial proprietary license.
25 * For more information, contact us at licensing@x264.com.
26 *****************************************************************************/
30 /* Deblocking filter */
31 static const uint8_t i_alpha_table[52+12*3] =
33 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
34 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
35 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
36 0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
37 7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
38 25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
39 80, 90,101,113,127,144,162,182,203,226,
41 255,255,255,255,255,255,255,255,255,255,255,255,
43 static const uint8_t i_beta_table[52+12*3] =
45 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
46 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
47 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
48 0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
49 3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
50 8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
51 13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
53 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
55 static const int8_t i_tc0_table[52+12*3][4] =
57 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
58 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
59 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
60 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
61 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
62 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
63 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
64 {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
65 {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
66 {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
67 {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
68 {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
69 {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-1,13,17,25 },
70 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
71 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
73 #define alpha_table(x) i_alpha_table[(x)+24]
74 #define beta_table(x) i_beta_table[(x)+24]
75 #define tc0_table(x) i_tc0_table[(x)+24]
78 static inline void deblock_luma_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
80 for( int i = 0; i < 4; i++ )
87 for( int d = 0; d < 4; d++ )
89 int p2 = pix[-3*xstride];
90 int p1 = pix[-2*xstride];
91 int p0 = pix[-1*xstride];
92 int q0 = pix[ 0*xstride];
93 int q1 = pix[ 1*xstride];
94 int q2 = pix[ 2*xstride];
96 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
100 if( abs( p2 - p0 ) < beta )
103 pix[-2*xstride] = p1 + x264_clip3( (( p2 + ((p0 + q0 + 1) >> 1)) >> 1) - p1, -tc0[i], tc0[i] );
106 if( abs( q2 - q0 ) < beta )
109 pix[ 1*xstride] = q1 + x264_clip3( (( q2 + ((p0 + q0 + 1) >> 1)) >> 1) - q1, -tc0[i], tc0[i] );
113 delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
114 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
115 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
121 static void deblock_v_luma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
123 deblock_luma_c( pix, stride, 1, alpha, beta, tc0 );
125 static void deblock_h_luma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
127 deblock_luma_c( pix, 1, stride, alpha, beta, tc0 );
130 static inline void deblock_chroma_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
132 for( int i = 0; i < 4; i++ )
140 for( int d = 0; d < 2; d++, pix += ystride-2 )
141 for( int e = 0; e < 2; e++, pix++ )
143 int p1 = pix[-2*xstride];
144 int p0 = pix[-1*xstride];
145 int q0 = pix[ 0*xstride];
146 int q1 = pix[ 1*xstride];
148 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
150 int delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
151 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
152 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
157 static void deblock_v_chroma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
159 deblock_chroma_c( pix, stride, 2, alpha, beta, tc0 );
161 static void deblock_h_chroma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
163 deblock_chroma_c( pix, 2, stride, alpha, beta, tc0 );
166 static inline void deblock_luma_intra_c( pixel *pix, int xstride, int ystride, int alpha, int beta )
168 for( int d = 0; d < 16; d++ )
170 int p2 = pix[-3*xstride];
171 int p1 = pix[-2*xstride];
172 int p0 = pix[-1*xstride];
173 int q0 = pix[ 0*xstride];
174 int q1 = pix[ 1*xstride];
175 int q2 = pix[ 2*xstride];
177 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
179 if(abs( p0 - q0 ) < ((alpha >> 2) + 2) )
181 if( abs( p2 - p0 ) < beta ) /* p0', p1', p2' */
183 const int p3 = pix[-4*xstride];
184 pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
185 pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
186 pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
189 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
190 if( abs( q2 - q0 ) < beta ) /* q0', q1', q2' */
192 const int q3 = pix[3*xstride];
193 pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
194 pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
195 pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
198 pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
202 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
203 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
209 static void deblock_v_luma_intra_c( pixel *pix, int stride, int alpha, int beta )
211 deblock_luma_intra_c( pix, stride, 1, alpha, beta );
213 static void deblock_h_luma_intra_c( pixel *pix, int stride, int alpha, int beta )
215 deblock_luma_intra_c( pix, 1, stride, alpha, beta );
218 static inline void deblock_chroma_intra_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int dir )
220 for( int d = 0; d < (dir?16:8); d++, pix += ystride-2 )
221 for( int e = 0; e < (dir?1:2); e++, pix++ )
223 int p1 = pix[-2*xstride];
224 int p0 = pix[-1*xstride];
225 int q0 = pix[ 0*xstride];
226 int q1 = pix[ 1*xstride];
228 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
230 pix[-1*xstride] = (2*p1 + p0 + q1 + 2) >> 2; /* p0' */
231 pix[ 0*xstride] = (2*q1 + q0 + p1 + 2) >> 2; /* q0' */
235 static void deblock_v_chroma_intra_c( pixel *pix, int stride, int alpha, int beta )
237 deblock_chroma_intra_c( pix, stride, 2, alpha, beta, 1 );
239 static void deblock_h_chroma_intra_c( pixel *pix, int stride, int alpha, int beta )
241 deblock_chroma_intra_c( pix, 2, stride, alpha, beta, 0 );
244 static void deblock_strength_c( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
245 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4], int mvy_limit,
248 for( int dir = 0; dir < 2; dir++ )
250 int s1 = dir ? 1 : 8;
251 int s2 = dir ? 8 : 1;
252 for( int edge = 0; edge < 4; edge++ )
253 for( int i = 0, loc = X264_SCAN8_0+edge*s2; i < 4; i++, loc += s1 )
256 if( nnz[loc] || nnz[locn] )
257 bs[dir][edge][i] = 2;
258 else if( ref[0][loc] != ref[0][locn] ||
259 abs( mv[0][loc][0] - mv[0][locn][0] ) >= 4 ||
260 abs( mv[0][loc][1] - mv[0][locn][1] ) >= mvy_limit ||
261 (bframe && (ref[1][loc] != ref[1][locn] ||
262 abs( mv[1][loc][0] - mv[1][locn][0] ) >= 4 ||
263 abs( mv[1][loc][1] - mv[1][locn][1] ) >= mvy_limit )))
265 bs[dir][edge][i] = 1;
268 bs[dir][edge][i] = 0;
273 static inline void deblock_edge( x264_t *h, pixel *pix, int i_stride, uint8_t bS[4], int i_qp, int b_chroma, x264_deblock_inter_t pf_inter )
275 int index_a = i_qp-QP_BD_OFFSET + h->sh.i_alpha_c0_offset;
276 int index_b = i_qp-QP_BD_OFFSET + h->sh.i_beta_offset;
277 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
278 int beta = beta_table(index_b) << (BIT_DEPTH-8);
281 if( !M32(bS) || !alpha || !beta )
284 tc[0] = (tc0_table(index_a)[bS[0]] << (BIT_DEPTH-8)) + b_chroma;
285 tc[1] = (tc0_table(index_a)[bS[1]] << (BIT_DEPTH-8)) + b_chroma;
286 tc[2] = (tc0_table(index_a)[bS[2]] << (BIT_DEPTH-8)) + b_chroma;
287 tc[3] = (tc0_table(index_a)[bS[3]] << (BIT_DEPTH-8)) + b_chroma;
289 pf_inter( pix, i_stride, alpha, beta, tc );
292 static inline void deblock_edge_intra( x264_t *h, pixel *pix, int i_stride, uint8_t bS[4], int i_qp, int b_chroma, x264_deblock_intra_t pf_intra )
294 int index_a = i_qp-QP_BD_OFFSET + h->sh.i_alpha_c0_offset;
295 int index_b = i_qp-QP_BD_OFFSET + h->sh.i_beta_offset;
296 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
297 int beta = beta_table(index_b) << (BIT_DEPTH-8);
299 if( !alpha || !beta )
302 pf_intra( pix, i_stride, alpha, beta );
305 void x264_frame_deblock_row( x264_t *h, int mb_y )
307 int b_interlaced = h->sh.b_mbaff;
308 int qp_thresh = 15 - X264_MIN( h->sh.i_alpha_c0_offset, h->sh.i_beta_offset ) - X264_MAX( 0, h->param.analyse.i_chroma_qp_offset );
309 int stridey = h->fdec->i_stride[0];
310 int stride2y = stridey << b_interlaced;
311 int strideuv = h->fdec->i_stride[1];
312 int stride2uv = strideuv << b_interlaced;
314 for( int mb_x = 0; mb_x < h->mb.i_mb_width; mb_x += (~b_interlaced | mb_y)&1, mb_y ^= b_interlaced )
316 x264_prefetch_fenc( h, h->fdec, mb_x, mb_y );
317 x264_macroblock_cache_load_neighbours_deblock( h, mb_x, mb_y );
319 int mb_xy = h->mb.i_mb_xy;
320 int transform_8x8 = h->mb.mb_transform_size[h->mb.i_mb_xy];
321 int intra_cur = IS_INTRA( h->mb.type[mb_xy] );
322 uint8_t (*bs)[4][4] = h->deblock_strength[mb_y&1][mb_x];
324 pixel *pixy = h->fdec->plane[0] + 16*mb_y*stridey + 16*mb_x;
325 pixel *pixuv = h->fdec->plane[1] + 8*mb_y*strideuv + 16*mb_x;
326 if( mb_y & b_interlaced )
332 int qp = h->mb.qp[mb_xy];
333 int qpc = h->chroma_qp_table[qp];
334 int first_edge_only = h->mb.type[mb_xy] == P_SKIP || qp <= qp_thresh;
336 #define FILTER( intra, dir, edge, qp, chroma_qp )\
339 deblock_edge##intra( h, pixy + 4*edge*(dir?stride2y:1),\
340 stride2y, bs[dir][edge], qp, 0,\
341 h->loopf.deblock_luma##intra[dir] );\
343 deblock_edge##intra( h, pixuv + 2*edge*(dir?stride2uv:2),\
344 stride2uv, bs[dir][edge], chroma_qp, 1,\
345 h->loopf.deblock_chroma##intra[dir] );\
348 if( h->mb.i_neighbour & MB_LEFT )
350 int qpl = h->mb.qp[h->mb.i_mb_left_xy];
351 int qp_left = (qp + qpl + 1) >> 1;
352 int qpc_left = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpl] + 1) >> 1;
353 int intra_left = IS_INTRA( h->mb.type[h->mb.i_mb_left_xy] );
354 if( intra_cur || intra_left )
355 FILTER( _intra, 0, 0, qp_left, qpc_left );
357 FILTER( , 0, 0, qp_left, qpc_left );
360 if( !first_edge_only )
362 if( !transform_8x8 ) FILTER( , 0, 1, qp, qpc );
363 FILTER( , 0, 2, qp, qpc );
364 if( !transform_8x8 ) FILTER( , 0, 3, qp, qpc );
367 if( h->mb.i_neighbour & MB_TOP )
369 int qpt = h->mb.qp[h->mb.i_mb_top_xy];
370 int qp_top = (qp + qpt + 1) >> 1;
371 int qpc_top = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpt] + 1) >> 1;
372 int intra_top = IS_INTRA( h->mb.type[h->mb.i_mb_top_xy] );
373 if( ~b_interlaced & (intra_cur | intra_top) )
374 FILTER( _intra, 1, 0, qp_top, qpc_top );
378 M32( bs[1][0] ) = 0x03030303;
379 FILTER( , 1, 0, qp_top, qpc_top );
383 if( !first_edge_only )
385 if( !transform_8x8 ) FILTER( , 1, 1, qp, qpc );
386 FILTER( , 1, 2, qp, qpc );
387 if( !transform_8x8 ) FILTER( , 1, 3, qp, qpc );
394 /* For deblock-aware RD.
396 * deblock macroblock edges
397 * support analysis partitions smaller than 16x16
399 * handle duplicate refs correctly
400 * handle cavlc+8x8dct correctly
402 void x264_macroblock_deblock( x264_t *h )
404 int qp_thresh = 15 - X264_MIN( h->sh.i_alpha_c0_offset, h->sh.i_beta_offset ) - X264_MAX( 0, h->param.analyse.i_chroma_qp_offset );
406 if( qp <= qp_thresh || h->mb.i_type == P_SKIP )
409 uint8_t (*bs)[4][4] = h->deblock_strength[h->mb.i_mb_y&1][h->mb.i_mb_x];
410 if( IS_INTRA( h->mb.i_type ) )
411 memset( bs, 3, 2*4*4*sizeof(uint8_t) );
413 h->loopf.deblock_strength( h->mb.cache.non_zero_count, h->mb.cache.ref, h->mb.cache.mv,
414 bs, 4 >> h->sh.b_mbaff, h->sh.i_type == SLICE_TYPE_B );
416 int transform_8x8 = h->mb.b_transform_8x8;
417 pixel *fdec = h->mb.pic.p_fdec[0];
419 #define FILTER( dir, edge )\
422 deblock_edge( h, fdec + 4*edge*(dir?FDEC_STRIDE:1),\
423 FDEC_STRIDE, bs[dir][edge], qp, 0,\
424 h->loopf.deblock_luma[dir] );\
427 if( !transform_8x8 ) FILTER( 0, 1 );
429 if( !transform_8x8 ) FILTER( 0, 3 );
431 if( !transform_8x8 ) FILTER( 1, 1 );
433 if( !transform_8x8 ) FILTER( 1, 3 );
439 void x264_deblock_v_luma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
440 void x264_deblock_v_luma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
441 void x264_deblock_h_luma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
442 void x264_deblock_h_luma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
443 void x264_deblock_v_chroma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
444 void x264_deblock_v_chroma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
445 void x264_deblock_h_chroma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
446 void x264_deblock_h_chroma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
447 void x264_deblock_v_luma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
448 void x264_deblock_v_luma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
449 void x264_deblock_h_luma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
450 void x264_deblock_h_luma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
451 void x264_deblock_v_chroma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
452 void x264_deblock_v_chroma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
453 void x264_deblock_h_chroma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
454 void x264_deblock_h_chroma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
455 void x264_deblock_strength_mmxext( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
456 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
457 int mvy_limit, int bframe );
458 void x264_deblock_strength_sse2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
459 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
460 int mvy_limit, int bframe );
461 void x264_deblock_strength_ssse3 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
462 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
463 int mvy_limit, int bframe );
464 void x264_deblock_strength_avx ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
465 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
466 int mvy_limit, int bframe );
468 void x264_deblock_h_luma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
469 void x264_deblock_v8_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
470 void x264_deblock_v_chroma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
471 void x264_deblock_h_chroma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
472 void x264_deblock_h_luma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
473 void x264_deblock_v8_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
474 void x264_deblock_v_chroma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
475 void x264_deblock_h_chroma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
478 void x264_deblock_v_luma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
479 void x264_deblock_v_luma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
481 // FIXME this wrapper has a significant cpu cost
482 static void x264_deblock_v_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
484 x264_deblock_v8_luma_mmxext( pix, stride, alpha, beta, tc0 );
485 x264_deblock_v8_luma_mmxext( pix+8, stride, alpha, beta, tc0+2 );
487 static void x264_deblock_v_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta )
489 x264_deblock_v8_luma_intra_mmxext( pix, stride, alpha, beta );
490 x264_deblock_v8_luma_intra_mmxext( pix+8, stride, alpha, beta );
492 #endif // HIGH_BIT_DEPTH
497 void x264_deblock_v_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
498 void x264_deblock_h_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
502 void x264_deblock_v_luma_neon( uint8_t *, int, int, int, int8_t * );
503 void x264_deblock_h_luma_neon( uint8_t *, int, int, int, int8_t * );
504 void x264_deblock_v_chroma_neon( uint8_t *, int, int, int, int8_t * );
505 void x264_deblock_h_chroma_neon( uint8_t *, int, int, int, int8_t * );
508 void x264_deblock_init( int cpu, x264_deblock_function_t *pf )
510 pf->deblock_luma[1] = deblock_v_luma_c;
511 pf->deblock_luma[0] = deblock_h_luma_c;
512 pf->deblock_chroma[1] = deblock_v_chroma_c;
513 pf->deblock_chroma[0] = deblock_h_chroma_c;
514 pf->deblock_luma_intra[1] = deblock_v_luma_intra_c;
515 pf->deblock_luma_intra[0] = deblock_h_luma_intra_c;
516 pf->deblock_chroma_intra[1] = deblock_v_chroma_intra_c;
517 pf->deblock_chroma_intra[0] = deblock_h_chroma_intra_c;
518 pf->deblock_strength = deblock_strength_c;
521 if( cpu&X264_CPU_MMXEXT )
524 pf->deblock_luma[1] = x264_deblock_v_luma_mmxext;
525 pf->deblock_luma[0] = x264_deblock_h_luma_mmxext;
526 pf->deblock_chroma[1] = x264_deblock_v_chroma_mmxext;
527 pf->deblock_chroma[0] = x264_deblock_h_chroma_mmxext;
528 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_mmxext;
529 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_mmxext;
530 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_mmxext;
531 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_mmxext;
533 pf->deblock_strength = x264_deblock_strength_mmxext;
534 if( cpu&X264_CPU_SSE2 )
536 pf->deblock_strength = x264_deblock_strength_sse2;
537 if( !(cpu&X264_CPU_STACK_MOD4) )
539 pf->deblock_luma[1] = x264_deblock_v_luma_sse2;
540 pf->deblock_luma[0] = x264_deblock_h_luma_sse2;
541 pf->deblock_chroma[1] = x264_deblock_v_chroma_sse2;
542 pf->deblock_chroma[0] = x264_deblock_h_chroma_sse2;
543 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_sse2;
544 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_sse2;
545 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_sse2;
546 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_sse2;
549 if( cpu&X264_CPU_SSSE3 )
550 pf->deblock_strength = x264_deblock_strength_ssse3;
551 if( cpu&X264_CPU_AVX )
553 pf->deblock_strength = x264_deblock_strength_avx;
554 if( !(cpu&X264_CPU_STACK_MOD4) )
556 pf->deblock_luma[1] = x264_deblock_v_luma_avx;
557 pf->deblock_luma[0] = x264_deblock_h_luma_avx;
558 pf->deblock_chroma[1] = x264_deblock_v_chroma_avx;
559 pf->deblock_chroma[0] = x264_deblock_h_chroma_avx;
560 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_avx;
561 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_avx;
562 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_avx;
563 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_avx;
571 if( cpu&X264_CPU_ALTIVEC )
573 pf->deblock_luma[1] = x264_deblock_v_luma_altivec;
574 pf->deblock_luma[0] = x264_deblock_h_luma_altivec;
576 #endif // HAVE_ALTIVEC
579 if( cpu&X264_CPU_NEON )
581 pf->deblock_luma[1] = x264_deblock_v_luma_neon;
582 pf->deblock_luma[0] = x264_deblock_h_luma_neon;
583 // pf->deblock_chroma[1] = x264_deblock_v_chroma_neon;
584 // pf->deblock_chroma[0] = x264_deblock_h_chroma_neon;
587 #endif // !HIGH_BIT_DEPTH