1 /* Copyright (c) 2003-2008 Jean-Marc Valin
2 Copyright (c) 2007-2008 CSIRO
3 Copyright (c) 2007-2009 Xiph.Org Foundation
4 Written by Jean-Marc Valin */
5 /**
6 @file arch.h
7 @brief Various architecture definitions for CELT
8 */
9 /*
10 Redistribution and use in source and binary forms, with or without
11 modification, are permitted provided that the following conditions
12 are met:
13
14 - Redistributions of source code must retain the above copyright
15 notice, this list of conditions and the following disclaimer.
16
17 - Redistributions in binary form must reproduce the above copyright
18 notice, this list of conditions and the following disclaimer in the
19 documentation and/or other materials provided with the distribution.
20
21 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
25 OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
26 EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
27 PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
28 PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
29 LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
30 NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
31 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
32 */
33
34 #ifndef ARCH_H
35 #define ARCH_H
36
37 #include "opus_types.h"
38 #include "common.h"
39
40 # if !defined(__GNUC_PREREQ)
41 # if defined(__GNUC__)&&defined(__GNUC_MINOR__)
42 # define __GNUC_PREREQ(_maj,_min) \
43 ((__GNUC__<<16)+__GNUC_MINOR__>=((_maj)<<16)+(_min))
44 # else
45 # define __GNUC_PREREQ(_maj,_min) 0
46 # endif
47 # endif
48
49 #define CELT_SIG_SCALE 32768.f
50
51 #define celt_fatal(str) _celt_fatal(str, __FILE__, __LINE__);
52 #ifdef ENABLE_ASSERTIONS
53 #include <stdio.h>
54 #include <stdlib.h>
55 #ifdef __GNUC__
56 __attribute__((noreturn))
57 #endif
_celt_fatal(const char * str,const char * file,int line)58 static OPUS_INLINE void _celt_fatal(const char *str, const char *file, int line)
59 {
60 fprintf (stderr, "Fatal (internal) error in %s, line %d: %s\n", file, line, str);
61 abort();
62 }
63 #define celt_assert(cond) {if (!(cond)) {celt_fatal("assertion failed: " #cond);}}
64 #define celt_assert2(cond, message) {if (!(cond)) {celt_fatal("assertion failed: " #cond "\n" message);}}
65 #else
66 #define celt_assert(cond)
67 #define celt_assert2(cond, message)
68 #endif
69
70 #define IMUL32(a,b) ((a)*(b))
71
72 #define MIN16(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum 16-bit value. */
73 #define MAX16(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum 16-bit value. */
74 #define MIN32(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum 32-bit value. */
75 #define MAX32(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum 32-bit value. */
76 #define IMIN(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum int value. */
77 #define IMAX(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum int value. */
78 #define UADD32(a,b) ((a)+(b))
79 #define USUB32(a,b) ((a)-(b))
80
81 /* Set this if opus_int64 is a native type of the CPU. */
82 /* Assume that all LP64 architectures have fast 64-bit types; also x86_64
83 (which can be ILP32 for x32) and Win64 (which is LLP64). */
84 #if defined(__x86_64__) || defined(__LP64__) || defined(_WIN64)
85 #define OPUS_FAST_INT64 1
86 #else
87 #define OPUS_FAST_INT64 0
88 #endif
89
90 #define PRINT_MIPS(file)
91
92 #ifdef FIXED_POINT
93
94 typedef opus_int16 opus_val16;
95 typedef opus_int32 opus_val32;
96 typedef opus_int64 opus_val64;
97
98 typedef opus_val32 celt_sig;
99 typedef opus_val16 celt_norm;
100 typedef opus_val32 celt_ener;
101
102 #define Q15ONE 32767
103
104 #define SIG_SHIFT 12
105 /* Safe saturation value for 32-bit signals. Should be less than
106 2^31*(1-0.85) to avoid blowing up on DC at deemphasis.*/
107 #define SIG_SAT (300000000)
108
109 #define NORM_SCALING 16384
110
111 #define DB_SHIFT 10
112
113 #define EPSILON 1
114 #define VERY_SMALL 0
115 #define VERY_LARGE16 ((opus_val16)32767)
116 #define Q15_ONE ((opus_val16)32767)
117
118 #define SCALEIN(a) (a)
119 #define SCALEOUT(a) (a)
120
121 #define ABS16(x) ((x) < 0 ? (-(x)) : (x))
122 #define ABS32(x) ((x) < 0 ? (-(x)) : (x))
123
SAT16(opus_int32 x)124 static OPUS_INLINE opus_int16 SAT16(opus_int32 x) {
125 return x > 32767 ? 32767 : x < -32768 ? -32768 : (opus_int16)x;
126 }
127
128 #ifdef FIXED_DEBUG
129 #include "fixed_debug.h"
130 #else
131
132 #include "fixed_generic.h"
133
134 #ifdef OPUS_ARM_PRESUME_AARCH64_NEON_INTR
135 #include "arm/fixed_arm64.h"
136 #elif OPUS_ARM_INLINE_EDSP
137 #include "arm/fixed_armv5e.h"
138 #elif defined (OPUS_ARM_INLINE_ASM)
139 #include "arm/fixed_armv4.h"
140 #elif defined (BFIN_ASM)
141 #include "fixed_bfin.h"
142 #elif defined (TI_C5X_ASM)
143 #include "fixed_c5x.h"
144 #elif defined (TI_C6X_ASM)
145 #include "fixed_c6x.h"
146 #endif
147
148 #endif
149
150 #else /* FIXED_POINT */
151
152 typedef float opus_val16;
153 typedef float opus_val32;
154 typedef float opus_val64;
155
156 typedef float celt_sig;
157 typedef float celt_norm;
158 typedef float celt_ener;
159
160 #ifdef FLOAT_APPROX
161 /* This code should reliably detect NaN/inf even when -ffast-math is used.
162 Assumes IEEE 754 format. */
celt_isnan(float x)163 static OPUS_INLINE int celt_isnan(float x)
164 {
165 union {float f; opus_uint32 i;} in;
166 in.f = x;
167 return ((in.i>>23)&0xFF)==0xFF && (in.i&0x007FFFFF)!=0;
168 }
169 #else
170 #ifdef __FAST_MATH__
171 #error Cannot build libopus with -ffast-math unless FLOAT_APPROX is defined. This could result in crashes on extreme (e.g. NaN) input
172 #endif
173 #define celt_isnan(x) ((x)!=(x))
174 #endif
175
176 #define Q15ONE 1.0f
177
178 #define NORM_SCALING 1.f
179
180 #define EPSILON 1e-15f
181 #define VERY_SMALL 1e-30f
182 #define VERY_LARGE16 1e15f
183 #define Q15_ONE ((opus_val16)1.f)
184
185 /* This appears to be the same speed as C99's fabsf() but it's more portable. */
186 #define ABS16(x) ((float)fabs(x))
187 #define ABS32(x) ((float)fabs(x))
188
189 #define QCONST16(x,bits) (x)
190 #define QCONST32(x,bits) (x)
191
192 #define NEG16(x) (-(x))
193 #define NEG32(x) (-(x))
194 #define NEG32_ovflw(x) (-(x))
195 #define EXTRACT16(x) (x)
196 #define EXTEND32(x) (x)
197 #define SHR16(a,shift) (a)
198 #define SHL16(a,shift) (a)
199 #define SHR32(a,shift) (a)
200 #define SHL32(a,shift) (a)
201 #define PSHR32(a,shift) (a)
202 #define VSHR32(a,shift) (a)
203
204 #define PSHR(a,shift) (a)
205 #define SHR(a,shift) (a)
206 #define SHL(a,shift) (a)
207 #define SATURATE(x,a) (x)
208 #define SATURATE16(x) (x)
209
210 #define ROUND16(a,shift) (a)
211 #define SROUND16(a,shift) (a)
212 #define HALF16(x) (.5f*(x))
213 #define HALF32(x) (.5f*(x))
214
215 #define ADD16(a,b) ((a)+(b))
216 #define SUB16(a,b) ((a)-(b))
217 #define ADD32(a,b) ((a)+(b))
218 #define SUB32(a,b) ((a)-(b))
219 #define ADD32_ovflw(a,b) ((a)+(b))
220 #define SUB32_ovflw(a,b) ((a)-(b))
221 #define MULT16_16_16(a,b) ((a)*(b))
222 #define MULT16_16(a,b) ((opus_val32)(a)*(opus_val32)(b))
223 #define MAC16_16(c,a,b) ((c)+(opus_val32)(a)*(opus_val32)(b))
224
225 #define MULT16_32_Q15(a,b) ((a)*(b))
226 #define MULT16_32_Q16(a,b) ((a)*(b))
227
228 #define MULT32_32_Q31(a,b) ((a)*(b))
229
230 #define MAC16_32_Q15(c,a,b) ((c)+(a)*(b))
231 #define MAC16_32_Q16(c,a,b) ((c)+(a)*(b))
232
233 #define MULT16_16_Q11_32(a,b) ((a)*(b))
234 #define MULT16_16_Q11(a,b) ((a)*(b))
235 #define MULT16_16_Q13(a,b) ((a)*(b))
236 #define MULT16_16_Q14(a,b) ((a)*(b))
237 #define MULT16_16_Q15(a,b) ((a)*(b))
238 #define MULT16_16_P15(a,b) ((a)*(b))
239 #define MULT16_16_P13(a,b) ((a)*(b))
240 #define MULT16_16_P14(a,b) ((a)*(b))
241 #define MULT16_32_P16(a,b) ((a)*(b))
242
243 #define DIV32_16(a,b) (((opus_val32)(a))/(opus_val16)(b))
244 #define DIV32(a,b) (((opus_val32)(a))/(opus_val32)(b))
245
246 #define SCALEIN(a) ((a)*CELT_SIG_SCALE)
247 #define SCALEOUT(a) ((a)*(1/CELT_SIG_SCALE))
248
249 #define SIG2WORD16(x) (x)
250
251 #endif /* !FIXED_POINT */
252
253 #ifndef GLOBAL_STACK_SIZE
254 #ifdef FIXED_POINT
255 #define GLOBAL_STACK_SIZE 120000
256 #else
257 #define GLOBAL_STACK_SIZE 120000
258 #endif
259 #endif
260
261 #endif /* ARCH_H */
262