summaryrefslogtreecommitdiff
path: root/libavcodec/simple_idct_template.c (plain)
blob: f532313441c05426679a378d629819416d3540cc
1/*
2 * Simple IDCT
3 *
4 * Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at>
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/* Based upon some commented-out C code from mpeg2dec (idct_mmx.c
29 * written by Aaron Holtzman <aholtzma@ess.engr.uvic.ca>). */
30
31#include "simple_idct.h"
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 22725 // 90901
69#define W2 21407 // 85627
70#define W3 19265 // 77062
71#define W4 16384 // 65535
72#define W5 12873 // 51491
73#define W6 8867 // 35468
74#define W7 4520 // 18081
75
76# ifdef EXTRA_SHIFT
77#define ROW_SHIFT 13
78#define COL_SHIFT 18
79#define DC_SHIFT 1
80# else
81#define ROW_SHIFT 12
82#define COL_SHIFT 19
83#define DC_SHIFT 2
84# endif
85
86# else
87#define W1 45451
88#define W2 42813
89#define W3 38531
90#define W4 32767
91#define W5 25746
92#define W6 17734
93#define W7 9041
94
95#define ROW_SHIFT 16
96#define COL_SHIFT 17
97#define DC_SHIFT -1
98# endif
99
100#define MUL(a, b) ((a) * (b))
101#define MAC(a, b, c) ((a) += (b) * (c))
102
103#else
104
105#error "Unsupported bitdepth"
106
107#endif
108
109#ifdef EXTRA_SHIFT
110static inline void FUNC(idctRowCondDC_extrashift)(int16_t *row, int extra_shift)
111#else
112static inline void FUNC(idctRowCondDC)(int16_t *row, int extra_shift)
113#endif
114{
115 SUINT a0, a1, a2, a3, b0, b1, b2, b3;
116
117#if HAVE_FAST_64BIT
118#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN)
119 if (((AV_RN64A(row) & ~ROW0_MASK) | AV_RN64A(row+4)) == 0) {
120 uint64_t temp;
121 if (DC_SHIFT - extra_shift >= 0) {
122 temp = (row[0] * (1 << (DC_SHIFT - extra_shift))) & 0xffff;
123 } else {
124 temp = ((row[0] + (1<<(extra_shift - DC_SHIFT-1))) >> (extra_shift - DC_SHIFT)) & 0xffff;
125 }
126 temp += temp * (1 << 16);
127 temp += temp * ((uint64_t) 1 << 32);
128 AV_WN64A(row, temp);
129 AV_WN64A(row + 4, temp);
130 return;
131 }
132#else
133 if (!(AV_RN32A(row+2) |
134 AV_RN32A(row+4) |
135 AV_RN32A(row+6) |
136 row[1])) {
137 uint32_t temp;
138 if (DC_SHIFT - extra_shift >= 0) {
139 temp = (row[0] * (1 << (DC_SHIFT - extra_shift))) & 0xffff;
140 } else {
141 temp = ((row[0] + (1<<(extra_shift - DC_SHIFT-1))) >> (extra_shift - DC_SHIFT)) & 0xffff;
142 }
143 temp += temp * (1 << 16);
144 AV_WN32A(row, temp);
145 AV_WN32A(row+2, temp);
146 AV_WN32A(row+4, temp);
147 AV_WN32A(row+6, temp);
148 return;
149 }
150#endif
151
152 a0 = (W4 * row[0]) + (1 << (ROW_SHIFT + extra_shift - 1));
153 a1 = a0;
154 a2 = a0;
155 a3 = a0;
156
157 a0 += W2 * row[2];
158 a1 += W6 * row[2];
159 a2 -= W6 * row[2];
160 a3 -= W2 * row[2];
161
162 b0 = MUL(W1, row[1]);
163 MAC(b0, W3, row[3]);
164 b1 = MUL(W3, row[1]);
165 MAC(b1, -W7, row[3]);
166 b2 = MUL(W5, row[1]);
167 MAC(b2, -W1, row[3]);
168 b3 = MUL(W7, row[1]);
169 MAC(b3, -W5, row[3]);
170
171 if (AV_RN64A(row + 4)) {
172 a0 += W4*row[4] + W6*row[6];
173 a1 += - W4*row[4] - W2*row[6];
174 a2 += - W4*row[4] + W2*row[6];
175 a3 += W4*row[4] - W6*row[6];
176
177 MAC(b0, W5, row[5]);
178 MAC(b0, W7, row[7]);
179
180 MAC(b1, -W1, row[5]);
181 MAC(b1, -W5, row[7]);
182
183 MAC(b2, W7, row[5]);
184 MAC(b2, W3, row[7]);
185
186 MAC(b3, W3, row[5]);
187 MAC(b3, -W1, row[7]);
188 }
189
190 row[0] = (int)(a0 + b0) >> (ROW_SHIFT + extra_shift);
191 row[7] = (int)(a0 - b0) >> (ROW_SHIFT + extra_shift);
192 row[1] = (int)(a1 + b1) >> (ROW_SHIFT + extra_shift);
193 row[6] = (int)(a1 - b1) >> (ROW_SHIFT + extra_shift);
194 row[2] = (int)(a2 + b2) >> (ROW_SHIFT + extra_shift);
195 row[5] = (int)(a2 - b2) >> (ROW_SHIFT + extra_shift);
196 row[3] = (int)(a3 + b3) >> (ROW_SHIFT + extra_shift);
197 row[4] = (int)(a3 - b3) >> (ROW_SHIFT + extra_shift);
198}
199
200#define IDCT_COLS do { \
201 a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); \
202 a1 = a0; \
203 a2 = a0; \
204 a3 = a0; \
205 \
206 a0 += W2*col[8*2]; \
207 a1 += W6*col[8*2]; \
208 a2 += -W6*col[8*2]; \
209 a3 += -W2*col[8*2]; \
210 \
211 b0 = MUL(W1, col[8*1]); \
212 b1 = MUL(W3, col[8*1]); \
213 b2 = MUL(W5, col[8*1]); \
214 b3 = MUL(W7, col[8*1]); \
215 \
216 MAC(b0, W3, col[8*3]); \
217 MAC(b1, -W7, col[8*3]); \
218 MAC(b2, -W1, col[8*3]); \
219 MAC(b3, -W5, col[8*3]); \
220 \
221 if (col[8*4]) { \
222 a0 += W4*col[8*4]; \
223 a1 += -W4*col[8*4]; \
224 a2 += -W4*col[8*4]; \
225 a3 += W4*col[8*4]; \
226 } \
227 \
228 if (col[8*5]) { \
229 MAC(b0, W5, col[8*5]); \
230 MAC(b1, -W1, col[8*5]); \
231 MAC(b2, W7, col[8*5]); \
232 MAC(b3, W3, col[8*5]); \
233 } \
234 \
235 if (col[8*6]) { \
236 a0 += W6*col[8*6]; \
237 a1 += -W2*col[8*6]; \
238 a2 += W2*col[8*6]; \
239 a3 += -W6*col[8*6]; \
240 } \
241 \
242 if (col[8*7]) { \
243 MAC(b0, W7, col[8*7]); \
244 MAC(b1, -W5, col[8*7]); \
245 MAC(b2, W3, col[8*7]); \
246 MAC(b3, -W1, col[8*7]); \
247 } \
248 } while (0)
249
250#ifdef EXTRA_SHIFT
251static inline void FUNC(idctSparseCol_extrashift)(int16_t *col)
252#else
253static inline void FUNC(idctSparseColPut)(pixel *dest, ptrdiff_t line_size,
254 int16_t *col)
255{
256 SUINT a0, a1, a2, a3, b0, b1, b2, b3;
257
258 IDCT_COLS;
259
260 dest[0] = av_clip_pixel((int)(a0 + b0) >> COL_SHIFT);
261 dest += line_size;
262 dest[0] = av_clip_pixel((int)(a1 + b1) >> COL_SHIFT);
263 dest += line_size;
264 dest[0] = av_clip_pixel((int)(a2 + b2) >> COL_SHIFT);
265 dest += line_size;
266 dest[0] = av_clip_pixel((int)(a3 + b3) >> COL_SHIFT);
267 dest += line_size;
268 dest[0] = av_clip_pixel((int)(a3 - b3) >> COL_SHIFT);
269 dest += line_size;
270 dest[0] = av_clip_pixel((int)(a2 - b2) >> COL_SHIFT);
271 dest += line_size;
272 dest[0] = av_clip_pixel((int)(a1 - b1) >> COL_SHIFT);
273 dest += line_size;
274 dest[0] = av_clip_pixel((int)(a0 - b0) >> COL_SHIFT);
275}
276
277static inline void FUNC(idctSparseColAdd)(pixel *dest, ptrdiff_t line_size,
278 int16_t *col)
279{
280 int a0, a1, a2, a3, b0, b1, b2, b3;
281
282 IDCT_COLS;
283
284 dest[0] = av_clip_pixel(dest[0] + ((a0 + b0) >> COL_SHIFT));
285 dest += line_size;
286 dest[0] = av_clip_pixel(dest[0] + ((a1 + b1) >> COL_SHIFT));
287 dest += line_size;
288 dest[0] = av_clip_pixel(dest[0] + ((a2 + b2) >> COL_SHIFT));
289 dest += line_size;
290 dest[0] = av_clip_pixel(dest[0] + ((a3 + b3) >> COL_SHIFT));
291 dest += line_size;
292 dest[0] = av_clip_pixel(dest[0] + ((a3 - b3) >> COL_SHIFT));
293 dest += line_size;
294 dest[0] = av_clip_pixel(dest[0] + ((a2 - b2) >> COL_SHIFT));
295 dest += line_size;
296 dest[0] = av_clip_pixel(dest[0] + ((a1 - b1) >> COL_SHIFT));
297 dest += line_size;
298 dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT));
299}
300
301static inline void FUNC(idctSparseCol)(int16_t *col)
302#endif
303{
304 int a0, a1, a2, a3, b0, b1, b2, b3;
305
306 IDCT_COLS;
307
308 col[0 ] = ((a0 + b0) >> COL_SHIFT);
309 col[8 ] = ((a1 + b1) >> COL_SHIFT);
310 col[16] = ((a2 + b2) >> COL_SHIFT);
311 col[24] = ((a3 + b3) >> COL_SHIFT);
312 col[32] = ((a3 - b3) >> COL_SHIFT);
313 col[40] = ((a2 - b2) >> COL_SHIFT);
314 col[48] = ((a1 - b1) >> COL_SHIFT);
315 col[56] = ((a0 - b0) >> COL_SHIFT);
316}
317
318#ifndef EXTRA_SHIFT
319void FUNC(ff_simple_idct_put)(uint8_t *dest_, ptrdiff_t line_size, int16_t *block)
320{
321 pixel *dest = (pixel *)dest_;
322 int i;
323
324 line_size /= sizeof(pixel);
325
326 for (i = 0; i < 8; i++)
327 FUNC(idctRowCondDC)(block + i*8, 0);
328
329 for (i = 0; i < 8; i++)
330 FUNC(idctSparseColPut)(dest + i, line_size, block + i);
331}
332
333void FUNC(ff_simple_idct_add)(uint8_t *dest_, ptrdiff_t line_size, int16_t *block)
334{
335 pixel *dest = (pixel *)dest_;
336 int i;
337
338 line_size /= sizeof(pixel);
339
340 for (i = 0; i < 8; i++)
341 FUNC(idctRowCondDC)(block + i*8, 0);
342
343 for (i = 0; i < 8; i++)
344 FUNC(idctSparseColAdd)(dest + i, line_size, block + i);
345}
346
347void FUNC(ff_simple_idct)(int16_t *block)
348{
349 int i;
350
351 for (i = 0; i < 8; i++)
352 FUNC(idctRowCondDC)(block + i*8, 0);
353
354 for (i = 0; i < 8; i++)
355 FUNC(idctSparseCol)(block + i);
356}
357#endif
358
generated by cgit at 2024-05-17 03:08:38 (GMT)