1 /*
2 Copyright (c) 2003-2004, Mark Borgerding
3 
4 All rights reserved.
5 
6 Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
7 
8     * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
9     * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
10     * Neither the author nor the names of any contributors may be used to endorse or promote products derived from this software without specific prior written permission.
11 
12 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
13 */
14 
15 #define MIN(a,b) ((a)<(b) ? (a):(b))
16 #define MAX(a,b) ((a)>(b) ? (a):(b))
17 
18 /* kiss_fft.h
19    defines kiss_fft_scalar as either short or a float type
20    and defines
21    typedef struct { kiss_fft_scalar r; kiss_fft_scalar i; }kiss_fft_cpx; */
22 #include "kiss_fft.h"
23 #include "math_approx.h"
24 
25 #define MAXFACTORS 32
26 /* e.g. an fft of length 128 has 4 factors
27  as far as kissfft is concerned
28  4*4*4*2
29  */
30 
31 struct kiss_fft_state{
32     int nfft;
33     int inverse;
34     int factors[2*MAXFACTORS];
35     kiss_fft_cpx twiddles[1];
36 };
37 
38 /*
39   Explanation of macros dealing with complex math:
40 
41    C_MUL(m,a,b)         : m = a*b
42    C_FIXDIV( c , div )  : if a fixed point impl., c /= div. noop otherwise
43    C_SUB( res, a,b)     : res = a - b
44    C_SUBFROM( res , a)  : res -= a
45    C_ADDTO( res , a)    : res += a
46  * */
47 #ifdef FIXED_POINT
48 #include "arch.h"
49 # define FRACBITS 15
50 # define SAMPPROD spx_int32_t
51 #define SAMP_MAX 32767
52 
53 #define SAMP_MIN -SAMP_MAX
54 
55 #if defined(CHECK_OVERFLOW)
56 #  define CHECK_OVERFLOW_OP(a,op,b)  \
57 	if ( (SAMPPROD)(a) op (SAMPPROD)(b) > SAMP_MAX || (SAMPPROD)(a) op (SAMPPROD)(b) < SAMP_MIN ) { \
58 		fprintf(stderr,"WARNING:overflow @ " __FILE__ "(%d): (%d " #op" %d) = %ld\n",__LINE__,(a),(b),(SAMPPROD)(a) op (SAMPPROD)(b) );  }
59 #endif
60 
61 
62 #   define smul(a,b) ( (SAMPPROD)(a)*(b) )
63 #   define sround( x )  (kiss_fft_scalar)( ( (x) + (1<<(FRACBITS-1)) ) >> FRACBITS )
64 
65 #   define S_MUL(a,b) sround( smul(a,b) )
66 
67 #   define C_MUL(m,a,b) \
68       do{ (m).r = sround( smul((a).r,(b).r) - smul((a).i,(b).i) ); \
69           (m).i = sround( smul((a).r,(b).i) + smul((a).i,(b).r) ); }while(0)
70 
71 #   define C_MUL4(m,a,b) \
72                do{ (m).r = PSHR32( smul((a).r,(b).r) - smul((a).i,(b).i),17 ); \
73                (m).i = PSHR32( smul((a).r,(b).i) + smul((a).i,(b).r),17 ); }while(0)
74 
75 #   define DIVSCALAR(x,k) \
76 	(x) = sround( smul(  x, SAMP_MAX/k ) )
77 
78 #   define C_FIXDIV(c,div) \
79 	do {    DIVSCALAR( (c).r , div);  \
80 		DIVSCALAR( (c).i  , div); }while (0)
81 
82 #   define C_MULBYSCALAR( c, s ) \
83     do{ (c).r =  sround( smul( (c).r , s ) ) ;\
84         (c).i =  sround( smul( (c).i , s ) ) ; }while(0)
85 
86 #else  /* not FIXED_POINT*/
87 
88 #   define S_MUL(a,b) ( (a)*(b) )
89 #define C_MUL(m,a,b) \
90     do{ (m).r = (a).r*(b).r - (a).i*(b).i;\
91         (m).i = (a).r*(b).i + (a).i*(b).r; }while(0)
92 
93 #define C_MUL4(m,a,b) C_MUL(m,a,b)
94 
95 #   define C_FIXDIV(c,div) /* NOOP */
96 #   define C_MULBYSCALAR( c, s ) \
97     do{ (c).r *= (s);\
98         (c).i *= (s); }while(0)
99 #endif
100 
101 #ifndef CHECK_OVERFLOW_OP
102 #  define CHECK_OVERFLOW_OP(a,op,b) /* noop */
103 #endif
104 
105 #define  C_ADD( res, a,b)\
106     do { \
107 	    CHECK_OVERFLOW_OP((a).r,+,(b).r)\
108 	    CHECK_OVERFLOW_OP((a).i,+,(b).i)\
109 	    (res).r=(a).r+(b).r;  (res).i=(a).i+(b).i; \
110     }while(0)
111 #define  C_SUB( res, a,b)\
112     do { \
113 	    CHECK_OVERFLOW_OP((a).r,-,(b).r)\
114 	    CHECK_OVERFLOW_OP((a).i,-,(b).i)\
115 	    (res).r=(a).r-(b).r;  (res).i=(a).i-(b).i; \
116     }while(0)
117 #define C_ADDTO( res , a)\
118     do { \
119 	    CHECK_OVERFLOW_OP((res).r,+,(a).r)\
120 	    CHECK_OVERFLOW_OP((res).i,+,(a).i)\
121 	    (res).r += (a).r;  (res).i += (a).i;\
122     }while(0)
123 
124 #define C_SUBFROM( res , a)\
125     do {\
126 	    CHECK_OVERFLOW_OP((res).r,-,(a).r)\
127 	    CHECK_OVERFLOW_OP((res).i,-,(a).i)\
128 	    (res).r -= (a).r;  (res).i -= (a).i; \
129     }while(0)
130 
131 
132 #ifdef FIXED_POINT
133 #  define KISS_FFT_COS(phase)  floor(MIN(32767,MAX(-32767,.5+32768 * cos (phase))))
134 #  define KISS_FFT_SIN(phase)  floor(MIN(32767,MAX(-32767,.5+32768 * sin (phase))))
135 #  define HALF_OF(x) ((x)>>1)
136 #elif defined(USE_SIMD)
137 #  define KISS_FFT_COS(phase) _mm_set1_ps( cos(phase) )
138 #  define KISS_FFT_SIN(phase) _mm_set1_ps( sin(phase) )
139 #  define HALF_OF(x) ((x)*_mm_set1_ps(.5))
140 #else
141 #  define KISS_FFT_COS(phase) (kiss_fft_scalar) cos(phase)
142 #  define KISS_FFT_SIN(phase) (kiss_fft_scalar) sin(phase)
143 #  define HALF_OF(x) ((x)*.5)
144 #endif
145 
146 #define  kf_cexp(x,phase) \
147 	do{ \
148 		(x)->r = KISS_FFT_COS(phase);\
149 		(x)->i = KISS_FFT_SIN(phase);\
150 	}while(0)
151 #define  kf_cexp2(x,phase) \
152                do{ \
153                (x)->r = spx_cos_norm((phase));\
154                (x)->i = spx_cos_norm((phase)-32768);\
155 }while(0)
156 
157 
158 /* a debugging function */
159 #define pcpx(c)\
160     fprintf(stderr,"%g + %gi\n",(double)((c)->r),(double)((c)->i) )
161