1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
|
/*
** FAAD2 - Freeware Advanced Audio (AAC) Decoder including SBR decoding
** Copyright (C) 2003-2004 M. Bakker, Ahead Software AG, http://www.nero.com
**
** This program is free software; you can redistribute it and/or modify
** it under the terms of the GNU General Public License as published by
** the Free Software Foundation; either version 2 of the License, or
** (at your option) any later version.
**
** This program is distributed in the hope that it will be useful,
** but WITHOUT ANY WARRANTY; without even the implied warranty of
** MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
** GNU General Public License for more details.
**
** You should have received a copy of the GNU General Public License
** along with this program; if not, write to the Free Software
** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
**
** Any non-GPL usage of this software or parts of this software is strictly
** forbidden.
**
** Commercial non-GPL licensing of this software is possible.
** For more info contact Ahead Software through Mpeg4AAClicense@nero.com.
**
** $Id$
**/
#ifndef __FIXED_H__
#define __FIXED_H__
#ifdef __cplusplus
extern "C" {
#endif
#if defined(_WIN32_WCE) && defined(_ARM_)
#include <cmnintrin.h>
#endif
#define COEF_BITS 28
#define COEF_PRECISION (1 << COEF_BITS)
#define REAL_BITS 14 // MAXIMUM OF 14 FOR FIXED POINT SBR
#define REAL_PRECISION (1 << REAL_BITS)
/* FRAC is the fractional only part of the fixed point number [0.0..1.0) */
#define FRAC_SIZE 32 /* frac is a 32 bit integer */
#define FRAC_BITS 31
#define FRAC_PRECISION ((uint32_t)(1 << FRAC_BITS))
#define FRAC_MAX 0x7FFFFFFF
typedef int32_t real_t;
#define UFIX_CONST(A,PRECISION) ((uint32_t)((A)*(PRECISION)+0.5))
#define FIX_CONST(A,PRECISION) (((A) >= 0) ? ((real_t)((A)*(PRECISION)+0.5)) : ((real_t)((A)*(PRECISION)-0.5)))
#define REAL_CONST(A) FIX_CONST((A),(REAL_PRECISION))
#define COEF_CONST(A) FIX_CONST((A),(COEF_PRECISION))
#define FRAC_CONST(A) (((A) == 1.00) ? ((real_t)FRAC_MAX) : FIX_CONST((A),(FRAC_PRECISION)))
//#define FRAC_CONST(A) (((A) >= 0) ? ((real_t)((A)*(FRAC_PRECISION)+0.5)) : ((real_t)((A)*(FRAC_PRECISION)-0.5)))
#define Q2_BITS 22
#define Q2_PRECISION (1 << Q2_BITS)
#define Q2_CONST(A) (((A) >= 0) ? ((real_t)((A)*(Q2_PRECISION)+0.5)) : ((real_t)((A)*(Q2_PRECISION)-0.5)))
#if defined(CPU_COLDFIRE)
static INLINE real_t MUL_F(real_t A, real_t B)
{
asm volatile (
"mac.l %[A], %[B], %%acc0\n\t"
"movclr.l %%acc0, %[A]"
: [A] "+&r" (A) : [B] "r" (B)
);
return A;
}
static INLINE real_t MUL_C(real_t A, real_t B)
{
asm volatile (
"mac.l %[A], %[B], %%acc0\n\t"
"movclr.l %%acc0, %[A]\n\t"
: [A] "+&d" (A) : [B] "r" (B)
);
return A << 3;
}
/* MUL_R needs too many shifts for us to just operate on the top 32 bits the
emac unit gives as usual, so we do a full 64 bit mul here. */
static INLINE real_t MUL_R(real_t x, real_t y)
{
real_t t1, t2;
asm volatile (
"mac.l %[x],%[y],%%acc0\n" /* multiply */
"mulu.l %[y],%[x] \n" /* get lower half, avoid emac stall */
"movclr.l %%acc0,%[t1] \n" /* get higher half */
"moveq.l #17,%[t2] \n"
"asl.l %[t2],%[t1] \n" /* hi <<= 17, plus one free */
"moveq.l #14,%[t2] \n"
"lsr.l %[t2],%[x] \n" /* (unsigned)lo >>= 14 */
"or.l %[x],%[t1] \n" /* combine result */
: /* outputs */
[t1]"=&d"(t1),
[t2]"=&d"(t2),
[x] "+d" (x)
: /* inputs */
[y] "d" (y)
);
return t1;
}
static INLINE void ComplexMult(real_t *y1, real_t *y2,
real_t x1, real_t x2, real_t c1, real_t c2)
{
asm volatile(
"mac.l %[x1], %[c1], %%acc0\n\t"
"mac.l %[x2], %[c2], %%acc0\n\t"
"mac.l %[x2], %[c1], %%acc1\n\t"
"msac.l %[x1], %[c2], %%acc1\n\t"
"movclr.l %%acc0, %[x1]\n\t"
"move.l %[x1], (%[y1])\n\t"
"movclr.l %%acc1, %[x1]\n\t"
"move.l %[x1], (%[y2])"
: [x1] "+&r" (x1)
: [x2] "r" (x2), [y1] "a" (y1), [y2] "a" (y2),
[c1] "r" (c1), [c2] "r" (c2)
: "memory"
);
}
/* the following see little or no use, so just ignore them for now */
#define MUL_Q2(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (Q2_BITS-1))) >> Q2_BITS)
#define MUL_SHIFT6(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (6-1))) >> 6)
#define MUL_SHIFT23(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (23-1))) >> 23)
#define DESCALE(A,S) ((S)>0?(((A)>>((S)-1))+1)>>1:(A)<<-(S))
#define DESCALE_SHIFT(A,SH,SC) DESCALE((A),(SC)-(SH)
#elif defined(__GNUC__) && defined (__arm__)
/* taken from MAD */
#define arm_mul(x, y, SCALEBITS) \
({ \
uint32_t __hi; \
uint32_t __lo; \
uint32_t __result; \
asm("smull %0, %1, %3, %4\n\t" \
"movs %0, %0, lsr %5\n\t" \
"adc %2, %0, %1, lsl %6" \
: "=&r" (__lo), "=&r" (__hi), "=r" (__result) \
: "%r" (x), "r" (y), \
"M" (SCALEBITS), "M" (32 - (SCALEBITS)) \
: "cc"); \
__result; \
})
static INLINE real_t MUL_R(real_t A, real_t B)
{
return arm_mul(A, B, REAL_BITS);
}
static INLINE real_t MUL_C(real_t A, real_t B)
{
return arm_mul(A, B, COEF_BITS);
}
static INLINE real_t MUL_Q2(real_t A, real_t B)
{
return arm_mul(A, B, Q2_BITS);
}
static INLINE real_t MUL_SHIFT6(real_t A, real_t B)
{
return arm_mul(A, B, 6);
}
static INLINE real_t MUL_SHIFT23(real_t A, real_t B)
{
return arm_mul(A, B, 23);
}
static INLINE real_t _MulHigh(real_t x, real_t y)
{
uint32_t __lo;
uint32_t __hi;
asm("smull\t%0, %1, %2, %3"
: "=&r"(__lo),"=&r"(__hi)
: "%r"(x),"r"(y)
: "cc");
return __hi;
}
static INLINE real_t MUL_F(real_t A, real_t B)
{
return _MulHigh(A, B) << (FRAC_SIZE-FRAC_BITS);
}
/* Complex multiplication */
static INLINE void ComplexMult(real_t *y1, real_t *y2,
real_t x1, real_t x2, real_t c1, real_t c2)
{
int32_t tmp, yt1, yt2;
asm("smull %0, %1, %4, %6\n\t"
"smlal %0, %1, %5, %7\n\t"
"rsb %3, %4, #0\n\t"
"smull %0, %2, %5, %6\n\t"
"smlal %0, %2, %3, %7"
: "=&r" (tmp), "=&r" (yt1), "=&r" (yt2), "=r" (x1)
: "3" (x1), "r" (x2), "r" (c1), "r" (c2)
: "cc" );
*y1 = yt1 << (FRAC_SIZE-FRAC_BITS);
*y2 = yt2 << (FRAC_SIZE-FRAC_BITS);
}
static inline real_t DESCALE_SHIFT(unsigned val, int shift, int scale)
{
unsigned out;
if ((out = val >> (scale - shift - 1)))
{
out++;
out >>= 1;
} else
out = val << (shift - scale);
return out;
}
#else
/* multiply with real shift */
#define MUL_R(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (REAL_BITS-1))) >> REAL_BITS)
/* multiply with coef shift */
#define MUL_C(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (COEF_BITS-1))) >> COEF_BITS)
/* multiply with fractional shift */
#if defined(_WIN32_WCE) && defined(_ARM_)
/* eVC for PocketPC has an intrinsic function that returns only the high 32 bits of a 32x32 bit multiply */
static INLINE real_t MUL_F(real_t A, real_t B)
{
return _MulHigh(A,B) << (32-FRAC_BITS);
}
#else
#define _MulHigh(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (FRAC_SIZE-1))) >> FRAC_SIZE)
#define MUL_F(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (FRAC_BITS-1))) >> FRAC_BITS)
#endif
#define MUL_Q2(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (Q2_BITS-1))) >> Q2_BITS)
#define MUL_SHIFT6(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (6-1))) >> 6)
#define MUL_SHIFT23(A,B) (real_t)(((int64_t)(A)*(int64_t)(B)+(1 << (23-1))) >> 23)
#define DESCALE(A,S) ((S)>0?(((A)>>((S)-1))+1)>>1:(A)<<-(S))
#define DESCALE_SHIFT(A,SH,SC) DESCALE((A),(SC)-(SH)
/* Complex multiplication */
static INLINE void ComplexMult(real_t *y1, real_t *y2,
real_t x1, real_t x2, real_t c1, real_t c2)
{
*y1 = (_MulHigh(x1, c1) + _MulHigh(x2, c2))<<(FRAC_SIZE-FRAC_BITS);
*y2 = (_MulHigh(x2, c1) - _MulHigh(x1, c2))<<(FRAC_SIZE-FRAC_BITS);
}
#endif
#ifdef __cplusplus
}
#endif
#endif
|
