| 6148 |
serge |
1 |
/* |
|
|
2 |
* Simple IDCT |
|
|
3 |
* |
|
|
4 |
* Copyright (c) 2001 Michael Niedermayer |
|
|
5 |
* |
|
|
6 |
* This file is part of FFmpeg. |
|
|
7 |
* |
|
|
8 |
* FFmpeg 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. |
|
|
12 |
* |
|
|
13 |
* FFmpeg 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. |
|
|
17 |
* |
|
|
18 |
* You should have received a copy of the GNU Lesser General Public |
|
|
19 |
* License along with FFmpeg; if not, write to the Free Software |
|
|
20 |
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
|
21 |
*/ |
|
|
22 |
|
|
|
23 |
/** |
|
|
24 |
* @file |
|
|
25 |
* simpleidct in C. |
|
|
26 |
*/ |
|
|
27 |
|
|
|
28 |
/* |
|
|
29 |
based upon some outcommented c code from mpeg2dec (idct_mmx.c |
|
|
30 |
written by Aaron Holtzman ) |
|
|
31 |
*/ |
|
|
32 |
|
|
|
33 |
#include "bit_depth_template.c" |
|
|
34 |
|
|
|
35 |
#undef W1 |
|
|
36 |
#undef W2 |
|
|
37 |
#undef W3 |
|
|
38 |
#undef W4 |
|
|
39 |
#undef W5 |
|
|
40 |
#undef W6 |
|
|
41 |
#undef W7 |
|
|
42 |
#undef ROW_SHIFT |
|
|
43 |
#undef COL_SHIFT |
|
|
44 |
#undef DC_SHIFT |
|
|
45 |
#undef MUL |
|
|
46 |
#undef MAC |
|
|
47 |
|
|
|
48 |
#if BIT_DEPTH == 8 |
|
|
49 |
|
|
|
50 |
#define W1 22725 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
51 |
#define W2 21407 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
52 |
#define W3 19266 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
53 |
#define W4 16383 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
54 |
#define W5 12873 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
55 |
#define W6 8867 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
56 |
#define W7 4520 //cos(i*M_PI/16)*sqrt(2)*(1<<14) + 0.5 |
|
|
57 |
|
|
|
58 |
#define ROW_SHIFT 11 |
|
|
59 |
#define COL_SHIFT 20 |
|
|
60 |
#define DC_SHIFT 3 |
|
|
61 |
|
|
|
62 |
#define MUL(a, b) MUL16(a, b) |
|
|
63 |
#define MAC(a, b, c) MAC16(a, b, c) |
|
|
64 |
|
|
|
65 |
#elif BIT_DEPTH == 10 || BIT_DEPTH == 12 |
|
|
66 |
|
|
|
67 |
#if BIT_DEPTH == 10 |
|
|
68 |
#define W1 90901 |
|
|
69 |
#define W2 85627 |
|
|
70 |
#define W3 77062 |
|
|
71 |
#define W4 65535 |
|
|
72 |
#define W5 51491 |
|
|
73 |
#define W6 35468 |
|
|
74 |
#define W7 18081 |
|
|
75 |
|
|
|
76 |
#define ROW_SHIFT 15 |
|
|
77 |
#define COL_SHIFT 20 |
|
|
78 |
#define DC_SHIFT 1 |
|
|
79 |
#else |
|
|
80 |
#define W1 45451 |
|
|
81 |
#define W2 42813 |
|
|
82 |
#define W3 38531 |
|
|
83 |
#define W4 32767 |
|
|
84 |
#define W5 25746 |
|
|
85 |
#define W6 17734 |
|
|
86 |
#define W7 9041 |
|
|
87 |
|
|
|
88 |
#define ROW_SHIFT 16 |
|
|
89 |
#define COL_SHIFT 17 |
|
|
90 |
#define DC_SHIFT -1 |
|
|
91 |
#endif |
|
|
92 |
|
|
|
93 |
#define MUL(a, b) ((a) * (b)) |
|
|
94 |
#define MAC(a, b, c) ((a) += (b) * (c)) |
|
|
95 |
|
|
|
96 |
#else |
|
|
97 |
|
|
|
98 |
#error "Unsupported bitdepth" |
|
|
99 |
|
|
|
100 |
#endif |
|
|
101 |
|
|
|
102 |
static inline void FUNC(idctRowCondDC)(int16_t *row, int extra_shift) |
|
|
103 |
{ |
|
|
104 |
int a0, a1, a2, a3, b0, b1, b2, b3; |
|
|
105 |
|
|
|
106 |
#if HAVE_FAST_64BIT |
|
|
107 |
#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN) |
|
|
108 |
if (((((uint64_t *)row)[0] & ~ROW0_MASK) | ((uint64_t *)row)[1]) == 0) { |
|
|
109 |
uint64_t temp; |
|
|
110 |
if (DC_SHIFT - extra_shift > 0) { |
|
|
111 |
temp = (row[0] << (DC_SHIFT - extra_shift)) & 0xffff; |
|
|
112 |
} else { |
|
|
113 |
temp = (row[0] >> (extra_shift - DC_SHIFT)) & 0xffff; |
|
|
114 |
} |
|
|
115 |
temp += temp << 16; |
|
|
116 |
temp += temp << 32; |
|
|
117 |
((uint64_t *)row)[0] = temp; |
|
|
118 |
((uint64_t *)row)[1] = temp; |
|
|
119 |
return; |
|
|
120 |
} |
|
|
121 |
#else |
|
|
122 |
if (!(((uint32_t*)row)[1] | |
|
|
123 |
((uint32_t*)row)[2] | |
|
|
124 |
((uint32_t*)row)[3] | |
|
|
125 |
row[1])) { |
|
|
126 |
uint32_t temp; |
|
|
127 |
if (DC_SHIFT - extra_shift > 0) { |
|
|
128 |
temp = (row[0] << (DC_SHIFT - extra_shift)) & 0xffff; |
|
|
129 |
} else { |
|
|
130 |
temp = (row[0] >> (extra_shift - DC_SHIFT)) & 0xffff; |
|
|
131 |
} |
|
|
132 |
temp += temp << 16; |
|
|
133 |
((uint32_t*)row)[0]=((uint32_t*)row)[1] = |
|
|
134 |
((uint32_t*)row)[2]=((uint32_t*)row)[3] = temp; |
|
|
135 |
return; |
|
|
136 |
} |
|
|
137 |
#endif |
|
|
138 |
|
|
|
139 |
a0 = (W4 * row[0]) + (1 << (ROW_SHIFT - 1)); |
|
|
140 |
a1 = a0; |
|
|
141 |
a2 = a0; |
|
|
142 |
a3 = a0; |
|
|
143 |
|
|
|
144 |
a0 += W2 * row[2]; |
|
|
145 |
a1 += W6 * row[2]; |
|
|
146 |
a2 -= W6 * row[2]; |
|
|
147 |
a3 -= W2 * row[2]; |
|
|
148 |
|
|
|
149 |
b0 = MUL(W1, row[1]); |
|
|
150 |
MAC(b0, W3, row[3]); |
|
|
151 |
b1 = MUL(W3, row[1]); |
|
|
152 |
MAC(b1, -W7, row[3]); |
|
|
153 |
b2 = MUL(W5, row[1]); |
|
|
154 |
MAC(b2, -W1, row[3]); |
|
|
155 |
b3 = MUL(W7, row[1]); |
|
|
156 |
MAC(b3, -W5, row[3]); |
|
|
157 |
|
|
|
158 |
if (AV_RN64A(row + 4)) { |
|
|
159 |
a0 += W4*row[4] + W6*row[6]; |
|
|
160 |
a1 += - W4*row[4] - W2*row[6]; |
|
|
161 |
a2 += - W4*row[4] + W2*row[6]; |
|
|
162 |
a3 += W4*row[4] - W6*row[6]; |
|
|
163 |
|
|
|
164 |
MAC(b0, W5, row[5]); |
|
|
165 |
MAC(b0, W7, row[7]); |
|
|
166 |
|
|
|
167 |
MAC(b1, -W1, row[5]); |
|
|
168 |
MAC(b1, -W5, row[7]); |
|
|
169 |
|
|
|
170 |
MAC(b2, W7, row[5]); |
|
|
171 |
MAC(b2, W3, row[7]); |
|
|
172 |
|
|
|
173 |
MAC(b3, W3, row[5]); |
|
|
174 |
MAC(b3, -W1, row[7]); |
|
|
175 |
} |
|
|
176 |
|
|
|
177 |
row[0] = (a0 + b0) >> (ROW_SHIFT + extra_shift); |
|
|
178 |
row[7] = (a0 - b0) >> (ROW_SHIFT + extra_shift); |
|
|
179 |
row[1] = (a1 + b1) >> (ROW_SHIFT + extra_shift); |
|
|
180 |
row[6] = (a1 - b1) >> (ROW_SHIFT + extra_shift); |
|
|
181 |
row[2] = (a2 + b2) >> (ROW_SHIFT + extra_shift); |
|
|
182 |
row[5] = (a2 - b2) >> (ROW_SHIFT + extra_shift); |
|
|
183 |
row[3] = (a3 + b3) >> (ROW_SHIFT + extra_shift); |
|
|
184 |
row[4] = (a3 - b3) >> (ROW_SHIFT + extra_shift); |
|
|
185 |
} |
|
|
186 |
|
|
|
187 |
#define IDCT_COLS do { \ |
|
|
188 |
a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); \ |
|
|
189 |
a1 = a0; \ |
|
|
190 |
a2 = a0; \ |
|
|
191 |
a3 = a0; \ |
|
|
192 |
\ |
|
|
193 |
a0 += W2*col[8*2]; \ |
|
|
194 |
a1 += W6*col[8*2]; \ |
|
|
195 |
a2 += -W6*col[8*2]; \ |
|
|
196 |
a3 += -W2*col[8*2]; \ |
|
|
197 |
\ |
|
|
198 |
b0 = MUL(W1, col[8*1]); \ |
|
|
199 |
b1 = MUL(W3, col[8*1]); \ |
|
|
200 |
b2 = MUL(W5, col[8*1]); \ |
|
|
201 |
b3 = MUL(W7, col[8*1]); \ |
|
|
202 |
\ |
|
|
203 |
MAC(b0, W3, col[8*3]); \ |
|
|
204 |
MAC(b1, -W7, col[8*3]); \ |
|
|
205 |
MAC(b2, -W1, col[8*3]); \ |
|
|
206 |
MAC(b3, -W5, col[8*3]); \ |
|
|
207 |
\ |
|
|
208 |
if (col[8*4]) { \ |
|
|
209 |
a0 += W4*col[8*4]; \ |
|
|
210 |
a1 += -W4*col[8*4]; \ |
|
|
211 |
a2 += -W4*col[8*4]; \ |
|
|
212 |
a3 += W4*col[8*4]; \ |
|
|
213 |
} \ |
|
|
214 |
\ |
|
|
215 |
if (col[8*5]) { \ |
|
|
216 |
MAC(b0, W5, col[8*5]); \ |
|
|
217 |
MAC(b1, -W1, col[8*5]); \ |
|
|
218 |
MAC(b2, W7, col[8*5]); \ |
|
|
219 |
MAC(b3, W3, col[8*5]); \ |
|
|
220 |
} \ |
|
|
221 |
\ |
|
|
222 |
if (col[8*6]) { \ |
|
|
223 |
a0 += W6*col[8*6]; \ |
|
|
224 |
a1 += -W2*col[8*6]; \ |
|
|
225 |
a2 += W2*col[8*6]; \ |
|
|
226 |
a3 += -W6*col[8*6]; \ |
|
|
227 |
} \ |
|
|
228 |
\ |
|
|
229 |
if (col[8*7]) { \ |
|
|
230 |
MAC(b0, W7, col[8*7]); \ |
|
|
231 |
MAC(b1, -W5, col[8*7]); \ |
|
|
232 |
MAC(b2, W3, col[8*7]); \ |
|
|
233 |
MAC(b3, -W1, col[8*7]); \ |
|
|
234 |
} \ |
|
|
235 |
} while (0) |
|
|
236 |
|
|
|
237 |
static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size, |
|
|
238 |
int16_t *col) |
|
|
239 |
{ |
|
|
240 |
int a0, a1, a2, a3, b0, b1, b2, b3; |
|
|
241 |
|
|
|
242 |
IDCT_COLS; |
|
|
243 |
|
|
|
244 |
dest[0] = av_clip_pixel((a0 + b0) >> COL_SHIFT); |
|
|
245 |
dest += line_size; |
|
|
246 |
dest[0] = av_clip_pixel((a1 + b1) >> COL_SHIFT); |
|
|
247 |
dest += line_size; |
|
|
248 |
dest[0] = av_clip_pixel((a2 + b2) >> COL_SHIFT); |
|
|
249 |
dest += line_size; |
|
|
250 |
dest[0] = av_clip_pixel((a3 + b3) >> COL_SHIFT); |
|
|
251 |
dest += line_size; |
|
|
252 |
dest[0] = av_clip_pixel((a3 - b3) >> COL_SHIFT); |
|
|
253 |
dest += line_size; |
|
|
254 |
dest[0] = av_clip_pixel((a2 - b2) >> COL_SHIFT); |
|
|
255 |
dest += line_size; |
|
|
256 |
dest[0] = av_clip_pixel((a1 - b1) >> COL_SHIFT); |
|
|
257 |
dest += line_size; |
|
|
258 |
dest[0] = av_clip_pixel((a0 - b0) >> COL_SHIFT); |
|
|
259 |
} |
|
|
260 |
|
|
|
261 |
static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size, |
|
|
262 |
int16_t *col) |
|
|
263 |
{ |
|
|
264 |
int a0, a1, a2, a3, b0, b1, b2, b3; |
|
|
265 |
|
|
|
266 |
IDCT_COLS; |
|
|
267 |
|
|
|
268 |
dest[0] = av_clip_pixel(dest[0] + ((a0 + b0) >> COL_SHIFT)); |
|
|
269 |
dest += line_size; |
|
|
270 |
dest[0] = av_clip_pixel(dest[0] + ((a1 + b1) >> COL_SHIFT)); |
|
|
271 |
dest += line_size; |
|
|
272 |
dest[0] = av_clip_pixel(dest[0] + ((a2 + b2) >> COL_SHIFT)); |
|
|
273 |
dest += line_size; |
|
|
274 |
dest[0] = av_clip_pixel(dest[0] + ((a3 + b3) >> COL_SHIFT)); |
|
|
275 |
dest += line_size; |
|
|
276 |
dest[0] = av_clip_pixel(dest[0] + ((a3 - b3) >> COL_SHIFT)); |
|
|
277 |
dest += line_size; |
|
|
278 |
dest[0] = av_clip_pixel(dest[0] + ((a2 - b2) >> COL_SHIFT)); |
|
|
279 |
dest += line_size; |
|
|
280 |
dest[0] = av_clip_pixel(dest[0] + ((a1 - b1) >> COL_SHIFT)); |
|
|
281 |
dest += line_size; |
|
|
282 |
dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT)); |
|
|
283 |
} |
|
|
284 |
|
|
|
285 |
static inline void FUNC(idctSparseCol)(int16_t *col) |
|
|
286 |
{ |
|
|
287 |
int a0, a1, a2, a3, b0, b1, b2, b3; |
|
|
288 |
|
|
|
289 |
IDCT_COLS; |
|
|
290 |
|
|
|
291 |
col[0 ] = ((a0 + b0) >> COL_SHIFT); |
|
|
292 |
col[8 ] = ((a1 + b1) >> COL_SHIFT); |
|
|
293 |
col[16] = ((a2 + b2) >> COL_SHIFT); |
|
|
294 |
col[24] = ((a3 + b3) >> COL_SHIFT); |
|
|
295 |
col[32] = ((a3 - b3) >> COL_SHIFT); |
|
|
296 |
col[40] = ((a2 - b2) >> COL_SHIFT); |
|
|
297 |
col[48] = ((a1 - b1) >> COL_SHIFT); |
|
|
298 |
col[56] = ((a0 - b0) >> COL_SHIFT); |
|
|
299 |
} |
|
|
300 |
|
|
|
301 |
void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, int16_t *block) |
|
|
302 |
{ |
|
|
303 |
pixel *dest = (pixel *)dest_; |
|
|
304 |
int i; |
|
|
305 |
|
|
|
306 |
line_size /= sizeof(pixel); |
|
|
307 |
|
|
|
308 |
for (i = 0; i < 8; i++) |
|
|
309 |
FUNC(idctRowCondDC)(block + i*8, 0); |
|
|
310 |
|
|
|
311 |
for (i = 0; i < 8; i++) |
|
|
312 |
FUNC(idctSparseColPut)(dest + i, line_size, block + i); |
|
|
313 |
} |
|
|
314 |
|
|
|
315 |
void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, int16_t *block) |
|
|
316 |
{ |
|
|
317 |
pixel *dest = (pixel *)dest_; |
|
|
318 |
int i; |
|
|
319 |
|
|
|
320 |
line_size /= sizeof(pixel); |
|
|
321 |
|
|
|
322 |
for (i = 0; i < 8; i++) |
|
|
323 |
FUNC(idctRowCondDC)(block + i*8, 0); |
|
|
324 |
|
|
|
325 |
for (i = 0; i < 8; i++) |
|
|
326 |
FUNC(idctSparseColAdd)(dest + i, line_size, block + i); |
|
|
327 |
} |
|
|
328 |
|
|
|
329 |
void FUNC(ff_simple_idct)(int16_t *block) |
|
|
330 |
{ |
|
|
331 |
int i; |
|
|
332 |
|
|
|
333 |
for (i = 0; i < 8; i++) |
|
|
334 |
FUNC(idctRowCondDC)(block + i*8, 0); |
|
|
335 |
|
|
|
336 |
for (i = 0; i < 8; i++) |
|
|
337 |
FUNC(idctSparseCol)(block + i); |
|
|
338 |
} |