4 * Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at>
6 * This file is part of Libav.
8 * Libav is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
13 * Libav is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with Libav; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
29 based upon some outcommented c code from mpeg2dec (idct_mmx.c
30 written by Aaron Holtzman <aholtzma@ess.engr.uvic.ca>)
33 #include "simple_idct.h"
35 #include "bit_depth_template.c"
52 #define W1 22725 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
53 #define W2 21407 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
54 #define W3 19266 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
55 #define W4 16383 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
56 #define W5 12873 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
57 #define W6 8867 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
58 #define W7 4520 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5
64 #define MUL(a, b) MUL16(a, b)
65 #define MAC(a, b, c) MAC16(a, b, c)
81 #define MUL(a, b) ((a) * (b))
82 #define MAC(a, b, c) ((a) += (b) * (c))
86 #error "Unsupported bitdepth"
90 static inline void FUNC(idctRowCondDC)(int16_t *row, int extra_shift)
92 int a0, a1, a2, a3, b0, b1, b2, b3;
95 #define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN)
96 if (((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) {
98 if (DC_SHIFT - extra_shift > 0) {
99 temp = (row[0] << (DC_SHIFT - extra_shift)) & 0xffff;
101 temp = (row[0] >> (extra_shift - DC_SHIFT)) & 0xffff;
105 ((uint64_t *)row)[0] = temp;
106 ((uint64_t *)row)[1] = temp;
110 if (!(((uint32_t*)row)[1] |
111 ((uint32_t*)row)[2] |
112 ((uint32_t*)row)[3] |
115 if (DC_SHIFT - extra_shift > 0) {
116 temp = (row[0] << (DC_SHIFT - extra_shift)) & 0xffff;
118 temp = (row[0] >> (extra_shift - DC_SHIFT)) & 0xffff;
121 ((uint32_t*)row)[0]=((uint32_t*)row)[1] =
122 ((uint32_t*)row)[2]=((uint32_t*)row)[3] = temp;
127 a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1));
137 b0 = MUL(W1, row[1]);
139 b1 = MUL(W3, row[1]);
140 MAC(b1, -W7, row[3]);
141 b2 = MUL(W5, row[1]);
142 MAC(b2, -W1, row[3]);
143 b3 = MUL(W7, row[1]);
144 MAC(b3, -W5, row[3]);
146 if (AV_RN64A(row + 4)) {
147 a0 += W4*row[4] + W6*row[6];
148 a1 += - W4*row[4] - W2*row[6];
149 a2 += - W4*row[4] + W2*row[6];
150 a3 += W4*row[4] - W6*row[6];
155 MAC(b1, -W1, row[5]);
156 MAC(b1, -W5, row[7]);
162 MAC(b3, -W1, row[7]);
165 row[0] = (a0 + b0) >> (ROW_SHIFT + extra_shift);
166 row[7] = (a0 - b0) >> (ROW_SHIFT + extra_shift);
167 row[1] = (a1 + b1) >> (ROW_SHIFT + extra_shift);
168 row[6] = (a1 - b1) >> (ROW_SHIFT + extra_shift);
169 row[2] = (a2 + b2) >> (ROW_SHIFT + extra_shift);
170 row[5] = (a2 - b2) >> (ROW_SHIFT + extra_shift);
171 row[3] = (a3 + b3) >> (ROW_SHIFT + extra_shift);
172 row[4] = (a3 - b3) >> (ROW_SHIFT + extra_shift);
175 #define IDCT_COLS do { \
176 a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); \
183 a2 += -W6*col[8*2]; \
184 a3 += -W2*col[8*2]; \
186 b0 = MUL(W1, col[8*1]); \
187 b1 = MUL(W3, col[8*1]); \
188 b2 = MUL(W5, col[8*1]); \
189 b3 = MUL(W7, col[8*1]); \
191 MAC(b0, W3, col[8*3]); \
192 MAC(b1, -W7, col[8*3]); \
193 MAC(b2, -W1, col[8*3]); \
194 MAC(b3, -W5, col[8*3]); \
198 a1 += -W4*col[8*4]; \
199 a2 += -W4*col[8*4]; \
204 MAC(b0, W5, col[8*5]); \
205 MAC(b1, -W1, col[8*5]); \
206 MAC(b2, W7, col[8*5]); \
207 MAC(b3, W3, col[8*5]); \
212 a1 += -W2*col[8*6]; \
214 a3 += -W6*col[8*6]; \
218 MAC(b0, W7, col[8*7]); \
219 MAC(b1, -W5, col[8*7]); \
220 MAC(b2, W3, col[8*7]); \
221 MAC(b3, -W1, col[8*7]); \
225 static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size,
228 int a0, a1, a2, a3, b0, b1, b2, b3;
232 dest[0] = av_clip_pixel((a0 + b0) >> COL_SHIFT);
234 dest[0] = av_clip_pixel((a1 + b1) >> COL_SHIFT);
236 dest[0] = av_clip_pixel((a2 + b2) >> COL_SHIFT);
238 dest[0] = av_clip_pixel((a3 + b3) >> COL_SHIFT);
240 dest[0] = av_clip_pixel((a3 - b3) >> COL_SHIFT);
242 dest[0] = av_clip_pixel((a2 - b2) >> COL_SHIFT);
244 dest[0] = av_clip_pixel((a1 - b1) >> COL_SHIFT);
246 dest[0] = av_clip_pixel((a0 - b0) >> COL_SHIFT);
249 static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size,
252 int a0, a1, a2, a3, b0, b1, b2, b3;
256 dest[0] = av_clip_pixel(dest[0] + ((a0 + b0) >> COL_SHIFT));
258 dest[0] = av_clip_pixel(dest[0] + ((a1 + b1) >> COL_SHIFT));
260 dest[0] = av_clip_pixel(dest[0] + ((a2 + b2) >> COL_SHIFT));
262 dest[0] = av_clip_pixel(dest[0] + ((a3 + b3) >> COL_SHIFT));
264 dest[0] = av_clip_pixel(dest[0] + ((a3 - b3) >> COL_SHIFT));
266 dest[0] = av_clip_pixel(dest[0] + ((a2 - b2) >> COL_SHIFT));
268 dest[0] = av_clip_pixel(dest[0] + ((a1 - b1) >> COL_SHIFT));
270 dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT));
273 static inline void FUNC(idctSparseCol)(int16_t *col)
275 int a0, a1, a2, a3, b0, b1, b2, b3;
279 col[0 ] = ((a0 + b0) >> COL_SHIFT);
280 col[8 ] = ((a1 + b1) >> COL_SHIFT);
281 col[16] = ((a2 + b2) >> COL_SHIFT);
282 col[24] = ((a3 + b3) >> COL_SHIFT);
283 col[32] = ((a3 - b3) >> COL_SHIFT);
284 col[40] = ((a2 - b2) >> COL_SHIFT);
285 col[48] = ((a1 - b1) >> COL_SHIFT);
286 col[56] = ((a0 - b0) >> COL_SHIFT);
289 void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, int16_t *block)
291 pixel *dest = (pixel *)dest_;
294 line_size /= sizeof(pixel);
296 for (i = 0; i < 8; i++)
297 FUNC(idctRowCondDC)(block + i*8, 0);
299 for (i = 0; i < 8; i++)
300 FUNC(idctSparseColPut)(dest + i, line_size, block + i);
303 void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, int16_t *block)
305 pixel *dest = (pixel *)dest_;
308 line_size /= sizeof(pixel);
310 for (i = 0; i < 8; i++)
311 FUNC(idctRowCondDC)(block + i*8, 0);
313 for (i = 0; i < 8; i++)
314 FUNC(idctSparseColAdd)(dest + i, line_size, block + i);
317 void FUNC(ff_simple_idct)(int16_t *block)
321 for (i = 0; i < 8; i++)
322 FUNC(idctRowCondDC)(block + i*8, 0);
324 for (i = 0; i < 8; i++)
325 FUNC(idctSparseCol)(block + i);