Mercurial > mplayer.hg
annotate libmpeg2/idct.c @ 11823:98a40d938b91
10l
author | diego |
---|---|
date | Wed, 21 Jan 2004 19:59:33 +0000 |
parents | ec04f41e2480 |
children | d0a8810e155c |
rev | line source |
---|---|
1 | 1 /* |
2 * idct.c | |
9852 | 3 * Copyright (C) 2000-2002 Michel Lespinasse <walken@zoy.org> |
4 * Copyright (C) 1999-2000 Aaron Holtzman <aholtzma@ess.engr.uvic.ca> | |
1 | 5 * |
6 * This file is part of mpeg2dec, a free MPEG-2 video stream decoder. | |
9852 | 7 * See http://libmpeg2.sourceforge.net/ for updates. |
1 | 8 * |
9 * mpeg2dec 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. | |
13 * | |
14 * mpeg2dec 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. | |
18 * | |
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 02111-1307 USA | |
22 */ | |
23 | |
24 #include "config.h" | |
25 | |
9852 | 26 #include <stdlib.h> |
1 | 27 #include <inttypes.h> |
28 | |
9852 | 29 #include "mpeg2.h" |
1 | 30 #include "mpeg2_internal.h" |
9852 | 31 #include "attributes.h" |
1 | 32 |
33 #define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */ | |
34 #define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */ | |
35 #define W3 2408 /* 2048*sqrt (2)*cos (3*pi/16) */ | |
36 #define W5 1609 /* 2048*sqrt (2)*cos (5*pi/16) */ | |
37 #define W6 1108 /* 2048*sqrt (2)*cos (6*pi/16) */ | |
38 #define W7 565 /* 2048*sqrt (2)*cos (7*pi/16) */ | |
39 | |
36 | 40 /* idct main entry point */ |
9852 | 41 void (* mpeg2_idct_copy) (int16_t * block, uint8_t * dest, int stride); |
42 void (* mpeg2_idct_add) (int last, int16_t * block, | |
43 uint8_t * dest, int stride); | |
1 | 44 |
45 static uint8_t clip_lut[1024]; | |
9852 | 46 #define CLIP(i) ((clip_lut+384)[(i)]) |
1 | 47 |
9852 | 48 #if 0 |
49 #define BUTTERFLY(t0,t1,W0,W1,d0,d1) \ | |
50 do { \ | |
51 t0 = W0*d0 + W1*d1; \ | |
52 t1 = W0*d1 - W1*d0; \ | |
53 } while (0) | |
54 #else | |
55 #define BUTTERFLY(t0,t1,W0,W1,d0,d1) \ | |
56 do { \ | |
57 int tmp = W0 * (d0 + d1); \ | |
58 t0 = tmp + (W1 - W0) * d1; \ | |
59 t1 = tmp - (W1 + W0) * d0; \ | |
60 } while (0) | |
61 #endif | |
1 | 62 |
10392 | 63 static inline void idct_row (int16_t * const block) |
1 | 64 { |
9852 | 65 int d0, d1, d2, d3; |
66 int a0, a1, a2, a3, b0, b1, b2, b3; | |
67 int t0, t1, t2, t3; | |
1 | 68 |
69 /* shortcut */ | |
9852 | 70 if (likely (!(block[1] | ((int32_t *)block)[1] | ((int32_t *)block)[2] | |
71 ((int32_t *)block)[3]))) { | |
72 uint32_t tmp = (uint16_t) (block[0] << 3); | |
73 tmp |= tmp << 16; | |
74 ((int32_t *)block)[0] = tmp; | |
75 ((int32_t *)block)[1] = tmp; | |
76 ((int32_t *)block)[2] = tmp; | |
77 ((int32_t *)block)[3] = tmp; | |
1 | 78 return; |
79 } | |
80 | |
9852 | 81 d0 = (block[0] << 11) + 128; |
82 d1 = block[1]; | |
83 d2 = block[2] << 11; | |
84 d3 = block[3]; | |
85 t0 = d0 + d2; | |
86 t1 = d0 - d2; | |
87 BUTTERFLY (t2, t3, W6, W2, d3, d1); | |
88 a0 = t0 + t2; | |
89 a1 = t1 + t3; | |
90 a2 = t1 - t3; | |
91 a3 = t0 - t2; | |
1 | 92 |
9852 | 93 d0 = block[4]; |
94 d1 = block[5]; | |
95 d2 = block[6]; | |
96 d3 = block[7]; | |
97 BUTTERFLY (t0, t1, W7, W1, d3, d0); | |
98 BUTTERFLY (t2, t3, W3, W5, d1, d2); | |
99 b0 = t0 + t2; | |
100 b3 = t1 + t3; | |
101 t0 -= t2; | |
102 t1 -= t3; | |
103 b1 = ((t0 + t1) * 181) >> 8; | |
104 b2 = ((t0 - t1) * 181) >> 8; | |
105 | |
106 block[0] = (a0 + b0) >> 8; | |
107 block[1] = (a1 + b1) >> 8; | |
108 block[2] = (a2 + b2) >> 8; | |
109 block[3] = (a3 + b3) >> 8; | |
110 block[4] = (a3 - b3) >> 8; | |
111 block[5] = (a2 - b2) >> 8; | |
112 block[6] = (a1 - b1) >> 8; | |
113 block[7] = (a0 - b0) >> 8; | |
1 | 114 } |
115 | |
10392 | 116 static inline void idct_col (int16_t * const block) |
1 | 117 { |
9852 | 118 int d0, d1, d2, d3; |
119 int a0, a1, a2, a3, b0, b1, b2, b3; | |
120 int t0, t1, t2, t3; | |
1 | 121 |
9852 | 122 d0 = (block[8*0] << 11) + 65536; |
123 d1 = block[8*1]; | |
124 d2 = block[8*2] << 11; | |
125 d3 = block[8*3]; | |
126 t0 = d0 + d2; | |
127 t1 = d0 - d2; | |
128 BUTTERFLY (t2, t3, W6, W2, d3, d1); | |
129 a0 = t0 + t2; | |
130 a1 = t1 + t3; | |
131 a2 = t1 - t3; | |
132 a3 = t0 - t2; | |
1 | 133 |
9852 | 134 d0 = block[8*4]; |
135 d1 = block[8*5]; | |
136 d2 = block[8*6]; | |
137 d3 = block[8*7]; | |
138 BUTTERFLY (t0, t1, W7, W1, d3, d0); | |
139 BUTTERFLY (t2, t3, W3, W5, d1, d2); | |
140 b0 = t0 + t2; | |
141 b3 = t1 + t3; | |
142 t0 = (t0 - t2) >> 8; | |
143 t1 = (t1 - t3) >> 8; | |
144 b1 = (t0 + t1) * 181; | |
145 b2 = (t0 - t1) * 181; | |
146 | |
147 block[8*0] = (a0 + b0) >> 17; | |
148 block[8*1] = (a1 + b1) >> 17; | |
149 block[8*2] = (a2 + b2) >> 17; | |
150 block[8*3] = (a3 + b3) >> 17; | |
151 block[8*4] = (a3 - b3) >> 17; | |
152 block[8*5] = (a2 - b2) >> 17; | |
153 block[8*6] = (a1 - b1) >> 17; | |
154 block[8*7] = (a0 - b0) >> 17; | |
1 | 155 } |
156 | |
9852 | 157 static void mpeg2_idct_copy_c (int16_t * block, uint8_t * dest, |
158 const int stride) | |
1 | 159 { |
160 int i; | |
161 | |
162 for (i = 0; i < 8; i++) | |
163 idct_row (block + 8 * i); | |
164 for (i = 0; i < 8; i++) | |
165 idct_col (block + i); | |
166 do { | |
167 dest[0] = CLIP (block[0]); | |
168 dest[1] = CLIP (block[1]); | |
169 dest[2] = CLIP (block[2]); | |
170 dest[3] = CLIP (block[3]); | |
171 dest[4] = CLIP (block[4]); | |
172 dest[5] = CLIP (block[5]); | |
173 dest[6] = CLIP (block[6]); | |
174 dest[7] = CLIP (block[7]); | |
175 | |
9852 | 176 block[0] = 0; block[1] = 0; block[2] = 0; block[3] = 0; |
177 block[4] = 0; block[5] = 0; block[6] = 0; block[7] = 0; | |
178 | |
1 | 179 dest += stride; |
180 block += 8; | |
181 } while (--i); | |
182 } | |
183 | |
9852 | 184 static void mpeg2_idct_add_c (const int last, int16_t * block, |
185 uint8_t * dest, const int stride) | |
1 | 186 { |
187 int i; | |
188 | |
9852 | 189 if (last != 129 || (block[0] & 7) == 4) { |
190 for (i = 0; i < 8; i++) | |
191 idct_row (block + 8 * i); | |
192 for (i = 0; i < 8; i++) | |
193 idct_col (block + i); | |
194 do { | |
195 dest[0] = CLIP (block[0] + dest[0]); | |
196 dest[1] = CLIP (block[1] + dest[1]); | |
197 dest[2] = CLIP (block[2] + dest[2]); | |
198 dest[3] = CLIP (block[3] + dest[3]); | |
199 dest[4] = CLIP (block[4] + dest[4]); | |
200 dest[5] = CLIP (block[5] + dest[5]); | |
201 dest[6] = CLIP (block[6] + dest[6]); | |
202 dest[7] = CLIP (block[7] + dest[7]); | |
203 | |
204 block[0] = 0; block[1] = 0; block[2] = 0; block[3] = 0; | |
205 block[4] = 0; block[5] = 0; block[6] = 0; block[7] = 0; | |
1 | 206 |
9852 | 207 dest += stride; |
208 block += 8; | |
209 } while (--i); | |
210 } else { | |
211 int DC; | |
212 | |
213 DC = (block[0] + 4) >> 3; | |
214 block[0] = block[63] = 0; | |
215 i = 8; | |
216 do { | |
217 dest[0] = CLIP (DC + dest[0]); | |
218 dest[1] = CLIP (DC + dest[1]); | |
219 dest[2] = CLIP (DC + dest[2]); | |
220 dest[3] = CLIP (DC + dest[3]); | |
221 dest[4] = CLIP (DC + dest[4]); | |
222 dest[5] = CLIP (DC + dest[5]); | |
223 dest[6] = CLIP (DC + dest[6]); | |
224 dest[7] = CLIP (DC + dest[7]); | |
225 dest += stride; | |
226 } while (--i); | |
227 } | |
228 } | |
1 | 229 |
9852 | 230 void mpeg2_idct_init (uint32_t accel) |
231 { | |
232 #ifdef ARCH_X86 | |
233 if (accel & MPEG2_ACCEL_X86_MMXEXT) { | |
234 mpeg2_idct_copy = mpeg2_idct_copy_mmxext; | |
235 mpeg2_idct_add = mpeg2_idct_add_mmxext; | |
236 mpeg2_idct_mmx_init (); | |
237 } else if (accel & MPEG2_ACCEL_X86_MMX) { | |
238 mpeg2_idct_copy = mpeg2_idct_copy_mmx; | |
239 mpeg2_idct_add = mpeg2_idct_add_mmx; | |
240 mpeg2_idct_mmx_init (); | |
241 } else | |
242 #endif | |
243 #ifdef ARCH_PPC | |
10269
217eb10b2f2d
libmpeg2-altivec patch by Magnus Damm <damm@opensource.se>:
arpi
parents:
9852
diff
changeset
|
244 #ifdef HAVE_ALTIVEC |
9852 | 245 if (accel & MPEG2_ACCEL_PPC_ALTIVEC) { |
246 mpeg2_idct_copy = mpeg2_idct_copy_altivec; | |
247 mpeg2_idct_add = mpeg2_idct_add_altivec; | |
248 mpeg2_idct_altivec_init (); | |
249 } else | |
250 #endif | |
10269
217eb10b2f2d
libmpeg2-altivec patch by Magnus Damm <damm@opensource.se>:
arpi
parents:
9852
diff
changeset
|
251 #endif |
9852 | 252 #ifdef ARCH_ALPHA |
10488 | 253 #ifdef CAN_COMPILE_ALPHA_MVI |
9852 | 254 if (accel & MPEG2_ACCEL_ALPHA_MVI) { |
255 mpeg2_idct_copy = mpeg2_idct_copy_mvi; | |
256 mpeg2_idct_add = mpeg2_idct_add_mvi; | |
257 mpeg2_idct_alpha_init (0); | |
10488 | 258 } else |
259 #endif | |
260 if (accel & MPEG2_ACCEL_ALPHA) { | |
9852 | 261 mpeg2_idct_copy = mpeg2_idct_copy_alpha; |
262 mpeg2_idct_add = mpeg2_idct_add_alpha; | |
263 mpeg2_idct_alpha_init (1); | |
264 } else | |
265 #endif | |
266 #ifdef LIBMPEG2_MLIB | |
267 if (accel & MPEG2_ACCEL_MLIB) { | |
268 mpeg2_idct_copy = mpeg2_idct_copy_mlib_non_ieee; | |
269 mpeg2_idct_add = (getenv ("MLIB_NON_IEEE") ? | |
270 mpeg2_idct_add_mlib_non_ieee : mpeg2_idct_add_mlib); | |
271 } else | |
272 #endif | |
273 { | |
274 extern uint8_t mpeg2_scan_norm[64]; | |
275 extern uint8_t mpeg2_scan_alt[64]; | |
276 int i, j; | |
1 | 277 |
9852 | 278 mpeg2_idct_copy = mpeg2_idct_copy_c; |
279 mpeg2_idct_add = mpeg2_idct_add_c; | |
280 for (i = -384; i < 640; i++) | |
281 clip_lut[i+384] = (i < 0) ? 0 : ((i > 255) ? 255 : i); | |
282 for (i = 0; i < 64; i++) { | |
283 j = mpeg2_scan_norm[i]; | |
284 mpeg2_scan_norm[i] = ((j & 0x36) >> 1) | ((j & 0x09) << 2); | |
285 j = mpeg2_scan_alt[i]; | |
286 mpeg2_scan_alt[i] = ((j & 0x36) >> 1) | ((j & 0x09) << 2); | |
287 } | |
288 } | |
1 | 289 } |