1 /*****************************************************************************
2 * dct.c: h264 encoder library
3 *****************************************************************************
4 * Copyright (C) 2003 Laurent Aimar
5 * $Id: dct.c,v 1.1 2004/06/03 19:27:06 fenrir Exp $
7 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
22 *****************************************************************************/
36 # include "i386/dct.h"
40 static inline int clip_uint8( int a )
49 * XXX For all dct dc : input could be equal to output so ...
52 static void dct2x2dc( int16_t d[2][2] )
56 tmp[0][0] = d[0][0] + d[0][1];
57 tmp[1][0] = d[0][0] - d[0][1];
58 tmp[0][1] = d[1][0] + d[1][1];
59 tmp[1][1] = d[1][0] - d[1][1];
61 d[0][0] = tmp[0][0] + tmp[0][1];
62 d[0][1] = tmp[1][0] + tmp[1][1];
63 d[1][0] = tmp[0][0] - tmp[0][1];
64 d[1][1] = tmp[1][0] - tmp[1][1];
67 static void dct4x4dc( int16_t d[4][4] )
74 for( i = 0; i < 4; i++ )
76 s01 = d[i][0] + d[i][1];
77 d01 = d[i][0] - d[i][1];
78 s23 = d[i][2] + d[i][3];
79 d23 = d[i][2] - d[i][3];
81 tmp[0][i] = s01 + s23;
82 tmp[1][i] = s01 - s23;
83 tmp[2][i] = d01 - d23;
84 tmp[3][i] = d01 + d23;
87 for( i = 0; i < 4; i++ )
89 s01 = tmp[i][0] + tmp[i][1];
90 d01 = tmp[i][0] - tmp[i][1];
91 s23 = tmp[i][2] + tmp[i][3];
92 d23 = tmp[i][2] - tmp[i][3];
94 d[0][i] = ( s01 + s23 + 1 ) >> 1;
95 d[1][i] = ( s01 - s23 + 1 ) >> 1;
96 d[2][i] = ( d01 - d23 + 1 ) >> 1;
97 d[3][i] = ( d01 + d23 + 1 ) >> 1;
101 static void idct4x4dc( int16_t d[4][4] )
108 for( i = 0; i < 4; i++ )
110 s01 = d[0][i] + d[1][i];
111 d01 = d[0][i] - d[1][i];
112 s23 = d[2][i] + d[3][i];
113 d23 = d[2][i] - d[3][i];
115 tmp[0][i] = s01 + s23;
116 tmp[1][i] = s01 - s23;
117 tmp[2][i] = d01 - d23;
118 tmp[3][i] = d01 + d23;
121 for( i = 0; i < 4; i++ )
123 s01 = tmp[i][0] + tmp[i][1];
124 d01 = tmp[i][0] - tmp[i][1];
125 s23 = tmp[i][2] + tmp[i][3];
126 d23 = tmp[i][2] - tmp[i][3];
135 static void sub4x4_dct( int16_t dct[4][4], uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2 )
142 for( y = 0; y < 4; y++ )
144 for( x = 0; x < 4; x++ )
146 d[y][x] = pix1[x] - pix2[x];
152 for( i = 0; i < 4; i++ )
154 const int s03 = d[i][0] + d[i][3];
155 const int s12 = d[i][1] + d[i][2];
156 const int d03 = d[i][0] - d[i][3];
157 const int d12 = d[i][1] - d[i][2];
159 tmp[0][i] = s03 + s12;
160 tmp[1][i] = 2*d03 + d12;
161 tmp[2][i] = s03 - s12;
162 tmp[3][i] = d03 - 2*d12;
165 for( i = 0; i < 4; i++ )
167 const int s03 = tmp[i][0] + tmp[i][3];
168 const int s12 = tmp[i][1] + tmp[i][2];
169 const int d03 = tmp[i][0] - tmp[i][3];
170 const int d12 = tmp[i][1] - tmp[i][2];
172 dct[0][i] = s03 + s12;
173 dct[1][i] = 2*d03 + d12;
174 dct[2][i] = s03 - s12;
175 dct[3][i] = d03 - 2*d12;
179 static void sub8x8_dct( int16_t dct[4][4][4], uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2 )
181 sub4x4_dct( dct[0], &pix1[0], i_pix1, &pix2[0], i_pix2 );
182 sub4x4_dct( dct[1], &pix1[4], i_pix1, &pix2[4], i_pix2 );
183 sub4x4_dct( dct[2], &pix1[4*i_pix1+0], i_pix1, &pix2[4*i_pix2+0], i_pix2 );
184 sub4x4_dct( dct[3], &pix1[4*i_pix1+4], i_pix1, &pix2[4*i_pix2+4], i_pix2 );
187 static void sub16x16_dct( int16_t dct[16][4][4], uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2 )
189 sub8x8_dct( &dct[ 0], pix1, i_pix1, pix2, i_pix2 );
190 sub8x8_dct( &dct[ 4], &pix1[8], i_pix1, &pix2[8], i_pix2 );
191 sub8x8_dct( &dct[ 8], &pix1[8*i_pix1], i_pix1, &pix2[8*i_pix2], i_pix2 );
192 sub8x8_dct( &dct[12], &pix1[8*i_pix1+8], i_pix1, &pix2[8*i_pix2+8], i_pix2 );
196 static void add4x4_idct( uint8_t *p_dst, int i_dst, int16_t dct[4][4] )
203 for( i = 0; i < 4; i++ )
205 const int s02 = dct[i][0] + dct[i][2];
206 const int d02 = dct[i][0] - dct[i][2];
207 const int s13 = dct[i][1] + (dct[i][3]>>1);
208 const int d13 = (dct[i][1]>>1) - dct[i][3];
210 tmp[i][0] = s02 + s13;
211 tmp[i][1] = d02 + d13;
212 tmp[i][2] = d02 - d13;
213 tmp[i][3] = s02 - s13;
216 for( i = 0; i < 4; i++ )
218 const int s02 = tmp[0][i] + tmp[2][i];
219 const int d02 = tmp[0][i] - tmp[2][i];
220 const int s13 = tmp[1][i] + (tmp[3][i]>>1);
221 const int d13 = (tmp[1][i]>>1) - tmp[3][i];
223 d[0][i] = ( s02 + s13 + 32 ) >> 6;
224 d[1][i] = ( d02 + d13 + 32 ) >> 6;
225 d[2][i] = ( d02 - d13 + 32 ) >> 6;
226 d[3][i] = ( s02 - s13 + 32 ) >> 6;
230 for( y = 0; y < 4; y++ )
232 for( x = 0; x < 4; x++ )
234 p_dst[x] = clip_uint8( p_dst[x] + d[y][x] );
240 static void add8x8_idct( uint8_t *p_dst, int i_dst, int16_t dct[4][4][4] )
242 add4x4_idct( p_dst, i_dst, dct[0] );
243 add4x4_idct( &p_dst[4], i_dst, dct[1] );
244 add4x4_idct( &p_dst[4*i_dst+0], i_dst, dct[2] );
245 add4x4_idct( &p_dst[4*i_dst+4], i_dst, dct[3] );
248 static void add16x16_idct( uint8_t *p_dst, int i_dst, int16_t dct[16][4][4] )
250 add8x8_idct( &p_dst[0], i_dst, &dct[0] );
251 add8x8_idct( &p_dst[8], i_dst, &dct[4] );
252 add8x8_idct( &p_dst[8*i_dst], i_dst, &dct[8] );
253 add8x8_idct( &p_dst[8*i_dst+8], i_dst, &dct[12] );
258 /****************************************************************************
260 ****************************************************************************/
261 void x264_dct_init( int cpu, x264_dct_function_t *dctf )
263 dctf->sub4x4_dct = sub4x4_dct;
264 dctf->add4x4_idct = add4x4_idct;
266 dctf->sub8x8_dct = sub8x8_dct;
267 dctf->add8x8_idct = add8x8_idct;
269 dctf->sub16x16_dct = sub16x16_dct;
270 dctf->add16x16_idct = add16x16_idct;
272 dctf->dct4x4dc = dct4x4dc;
273 dctf->idct4x4dc = idct4x4dc;
275 dctf->dct2x2dc = dct2x2dc;
276 dctf->idct2x2dc = dct2x2dc;
279 if( cpu&X264_CPU_MMXEXT )
281 dctf->sub4x4_dct = x264_sub4x4_dct_mmxext;
282 dctf->sub8x8_dct = x264_sub8x8_dct_mmxext;
283 dctf->sub16x16_dct = x264_sub16x16_dct_mmxext;
285 dctf->add4x4_idct = x264_add4x4_idct_mmxext;
286 dctf->add8x8_idct = x264_add8x8_idct_mmxext;
287 dctf->add16x16_idct = x264_add16x16_idct_mmxext;
289 dctf->dct4x4dc = x264_dct4x4dc_mmxext;
290 dctf->idct4x4dc = x264_idct4x4dc_mmxext;