1 /***********************************************************************
2 Copyright (c) 2006-2011, Skype Limited. All rights reserved.
3 Redistribution and use in source and binary forms, with or without
4 modification, are permitted provided that the following conditions
5 are met:
6 - Redistributions of source code must retain the above copyright notice,
7 this list of conditions and the following disclaimer.
8 - Redistributions in binary form must reproduce the above copyright
9 notice, this list of conditions and the following disclaimer in the
10 documentation and/or other materials provided with the distribution.
11 - Neither the name of Internet Society, IETF or IETF Trust, nor the
12 names of specific contributors, may be used to endorse or promote
13 products derived from this software without specific prior written
14 permission.
15 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
16 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17 IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18 ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
19 LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
20 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
21 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
22 INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
23 CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
24 ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
25 POSSIBILITY OF SUCH DAMAGE.
26 ***********************************************************************/
27
28 /*! \file silk_Inlines.h
29 * \brief silk_Inlines.h defines OPUS_INLINE signal processing functions.
30 */
31
32 #ifndef SILK_FIX_INLINES_H
33 #define SILK_FIX_INLINES_H
34
35 #ifdef __cplusplus
36 extern "C"
37 {
38 #endif
39
40 /* count leading zeros of opus_int64 */
silk_CLZ64(opus_int64 in)41 static OPUS_INLINE opus_int32 silk_CLZ64( opus_int64 in )
42 {
43 opus_int32 in_upper;
44
45 in_upper = (opus_int32)silk_RSHIFT64(in, 32);
46 if (in_upper == 0) {
47 /* Search in the lower 32 bits */
48 return 32 + silk_CLZ32( (opus_int32) in );
49 } else {
50 /* Search in the upper 32 bits */
51 return silk_CLZ32( in_upper );
52 }
53 }
54
55 /* get number of leading zeros and fractional part (the bits right after the leading one */
silk_CLZ_FRAC(opus_int32 in,opus_int32 * lz,opus_int32 * frac_Q7)56 static OPUS_INLINE void silk_CLZ_FRAC(
57 opus_int32 in, /* I input */
58 opus_int32 *lz, /* O number of leading zeros */
59 opus_int32 *frac_Q7 /* O the 7 bits right after the leading one */
60 )
61 {
62 opus_int32 lzeros = silk_CLZ32(in);
63
64 * lz = lzeros;
65 * frac_Q7 = silk_ROR32(in, 24 - lzeros) & 0x7f;
66 }
67
68 /* Approximation of square root */
69 /* Accuracy: < +/- 10% for output values > 15 */
70 /* < +/- 2.5% for output values > 120 */
silk_SQRT_APPROX(opus_int32 x)71 static OPUS_INLINE opus_int32 silk_SQRT_APPROX( opus_int32 x )
72 {
73 opus_int32 y, lz, frac_Q7;
74
75 if( x <= 0 ) {
76 return 0;
77 }
78
79 silk_CLZ_FRAC(x, &lz, &frac_Q7);
80
81 if( lz & 1 ) {
82 y = 32768;
83 } else {
84 y = 46214; /* 46214 = sqrt(2) * 32768 */
85 }
86
87 /* get scaling right */
88 y >>= silk_RSHIFT(lz, 1);
89
90 /* increment using fractional part of input */
91 y = silk_SMLAWB(y, y, silk_SMULBB(213, frac_Q7));
92
93 return y;
94 }
95
96 /* Divide two int32 values and return result as int32 in a given Q-domain */
silk_DIV32_varQ(const opus_int32 a32,const opus_int32 b32,const opus_int Qres)97 static OPUS_INLINE opus_int32 silk_DIV32_varQ( /* O returns a good approximation of "(a32 << Qres) / b32" */
98 const opus_int32 a32, /* I numerator (Q0) */
99 const opus_int32 b32, /* I denominator (Q0) */
100 const opus_int Qres /* I Q-domain of result (>= 0) */
101 )
102 {
103 opus_int a_headrm, b_headrm, lshift;
104 opus_int32 b32_inv, a32_nrm, b32_nrm, result;
105
106 silk_assert( b32 != 0 );
107 silk_assert( Qres >= 0 );
108
109 /* Compute number of bits head room and normalize inputs */
110 a_headrm = silk_CLZ32( silk_abs(a32) ) - 1;
111 a32_nrm = silk_LSHIFT(a32, a_headrm); /* Q: a_headrm */
112 b_headrm = silk_CLZ32( silk_abs(b32) ) - 1;
113 b32_nrm = silk_LSHIFT(b32, b_headrm); /* Q: b_headrm */
114
115 /* Inverse of b32, with 14 bits of precision */
116 b32_inv = silk_DIV32_16( silk_int32_MAX >> 2, silk_RSHIFT(b32_nrm, 16) ); /* Q: 29 + 16 - b_headrm */
117
118 /* First approximation */
119 result = silk_SMULWB(a32_nrm, b32_inv); /* Q: 29 + a_headrm - b_headrm */
120
121 /* Compute residual by subtracting product of denominator and first approximation */
122 /* It's OK to overflow because the final value of a32_nrm should always be small */
123 a32_nrm = silk_SUB32_ovflw(a32_nrm, silk_LSHIFT_ovflw( silk_SMMUL(b32_nrm, result), 3 )); /* Q: a_headrm */
124
125 /* Refinement */
126 result = silk_SMLAWB(result, a32_nrm, b32_inv); /* Q: 29 + a_headrm - b_headrm */
127
128 /* Convert to Qres domain */
129 lshift = 29 + a_headrm - b_headrm - Qres;
130 if( lshift < 0 ) {
131 return silk_LSHIFT_SAT32(result, -lshift);
132 } else {
133 if( lshift < 32){
134 return silk_RSHIFT(result, lshift);
135 } else {
136 /* Avoid undefined result */
137 return 0;
138 }
139 }
140 }
141
142 /* Invert int32 value and return result as int32 in a given Q-domain */
silk_INVERSE32_varQ(const opus_int32 b32,const opus_int Qres)143 static OPUS_INLINE opus_int32 silk_INVERSE32_varQ( /* O returns a good approximation of "(1 << Qres) / b32" */
144 const opus_int32 b32, /* I denominator (Q0) */
145 const opus_int Qres /* I Q-domain of result (> 0) */
146 )
147 {
148 opus_int b_headrm, lshift;
149 opus_int32 b32_inv, b32_nrm, err_Q32, result;
150
151 silk_assert( b32 != 0 );
152 silk_assert( Qres > 0 );
153
154 /* Compute number of bits head room and normalize input */
155 b_headrm = silk_CLZ32( silk_abs(b32) ) - 1;
156 b32_nrm = silk_LSHIFT(b32, b_headrm); /* Q: b_headrm */
157
158 /* Inverse of b32, with 14 bits of precision */
159 b32_inv = silk_DIV32_16( silk_int32_MAX >> 2, silk_RSHIFT(b32_nrm, 16) ); /* Q: 29 + 16 - b_headrm */
160
161 /* First approximation */
162 result = silk_LSHIFT(b32_inv, 16); /* Q: 61 - b_headrm */
163
164 /* Compute residual by subtracting product of denominator and first approximation from one */
165 err_Q32 = silk_LSHIFT( ((opus_int32)1<<29) - silk_SMULWB(b32_nrm, b32_inv), 3 ); /* Q32 */
166
167 /* Refinement */
168 result = silk_SMLAWW(result, err_Q32, b32_inv); /* Q: 61 - b_headrm */
169
170 /* Convert to Qres domain */
171 lshift = 61 - b_headrm - Qres;
172 if( lshift <= 0 ) {
173 return silk_LSHIFT_SAT32(result, -lshift);
174 } else {
175 if( lshift < 32){
176 return silk_RSHIFT(result, lshift);
177 }else{
178 /* Avoid undefined result */
179 return 0;
180 }
181 }
182 }
183
184 #ifdef __cplusplus
185 }
186 #endif
187
188 #endif /* SILK_FIX_INLINES_H */
189