2 * dvb-math provides some complex fixed-point math
3 * operations shared between the dvb related stuff
5 * Copyright (C) 2006 Christoph Pfister (christophpfister@gmail.com)
7 * This library is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU Lesser General Public License as
9 * published by the Free Software Foundation; either version 2.1 of
10 * the License, or (at your option) any later version.
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU Lesser General Public License for more details.
18 #include <linux/bitops.h>
19 #include <linux/kernel.h>
20 #include <linux/module.h>
22 #include <media/dvb_math.h>
24 static const unsigned short logtable
[256] = {
25 0x0000, 0x0171, 0x02e0, 0x044e, 0x05ba, 0x0725, 0x088e, 0x09f7,
26 0x0b5d, 0x0cc3, 0x0e27, 0x0f8a, 0x10eb, 0x124b, 0x13aa, 0x1508,
27 0x1664, 0x17bf, 0x1919, 0x1a71, 0x1bc8, 0x1d1e, 0x1e73, 0x1fc6,
28 0x2119, 0x226a, 0x23ba, 0x2508, 0x2656, 0x27a2, 0x28ed, 0x2a37,
29 0x2b80, 0x2cc8, 0x2e0f, 0x2f54, 0x3098, 0x31dc, 0x331e, 0x345f,
30 0x359f, 0x36de, 0x381b, 0x3958, 0x3a94, 0x3bce, 0x3d08, 0x3e41,
31 0x3f78, 0x40af, 0x41e4, 0x4319, 0x444c, 0x457f, 0x46b0, 0x47e1,
32 0x4910, 0x4a3f, 0x4b6c, 0x4c99, 0x4dc5, 0x4eef, 0x5019, 0x5142,
33 0x526a, 0x5391, 0x54b7, 0x55dc, 0x5700, 0x5824, 0x5946, 0x5a68,
34 0x5b89, 0x5ca8, 0x5dc7, 0x5ee5, 0x6003, 0x611f, 0x623a, 0x6355,
35 0x646f, 0x6588, 0x66a0, 0x67b7, 0x68ce, 0x69e4, 0x6af8, 0x6c0c,
36 0x6d20, 0x6e32, 0x6f44, 0x7055, 0x7165, 0x7274, 0x7383, 0x7490,
37 0x759d, 0x76aa, 0x77b5, 0x78c0, 0x79ca, 0x7ad3, 0x7bdb, 0x7ce3,
38 0x7dea, 0x7ef0, 0x7ff6, 0x80fb, 0x81ff, 0x8302, 0x8405, 0x8507,
39 0x8608, 0x8709, 0x8809, 0x8908, 0x8a06, 0x8b04, 0x8c01, 0x8cfe,
40 0x8dfa, 0x8ef5, 0x8fef, 0x90e9, 0x91e2, 0x92db, 0x93d2, 0x94ca,
41 0x95c0, 0x96b6, 0x97ab, 0x98a0, 0x9994, 0x9a87, 0x9b7a, 0x9c6c,
42 0x9d5e, 0x9e4f, 0x9f3f, 0xa02e, 0xa11e, 0xa20c, 0xa2fa, 0xa3e7,
43 0xa4d4, 0xa5c0, 0xa6ab, 0xa796, 0xa881, 0xa96a, 0xaa53, 0xab3c,
44 0xac24, 0xad0c, 0xadf2, 0xaed9, 0xafbe, 0xb0a4, 0xb188, 0xb26c,
45 0xb350, 0xb433, 0xb515, 0xb5f7, 0xb6d9, 0xb7ba, 0xb89a, 0xb97a,
46 0xba59, 0xbb38, 0xbc16, 0xbcf4, 0xbdd1, 0xbead, 0xbf8a, 0xc065,
47 0xc140, 0xc21b, 0xc2f5, 0xc3cf, 0xc4a8, 0xc580, 0xc658, 0xc730,
48 0xc807, 0xc8de, 0xc9b4, 0xca8a, 0xcb5f, 0xcc34, 0xcd08, 0xcddc,
49 0xceaf, 0xcf82, 0xd054, 0xd126, 0xd1f7, 0xd2c8, 0xd399, 0xd469,
50 0xd538, 0xd607, 0xd6d6, 0xd7a4, 0xd872, 0xd93f, 0xda0c, 0xdad9,
51 0xdba5, 0xdc70, 0xdd3b, 0xde06, 0xded0, 0xdf9a, 0xe063, 0xe12c,
52 0xe1f5, 0xe2bd, 0xe385, 0xe44c, 0xe513, 0xe5d9, 0xe69f, 0xe765,
53 0xe82a, 0xe8ef, 0xe9b3, 0xea77, 0xeb3b, 0xebfe, 0xecc1, 0xed83,
54 0xee45, 0xef06, 0xefc8, 0xf088, 0xf149, 0xf209, 0xf2c8, 0xf387,
55 0xf446, 0xf505, 0xf5c3, 0xf680, 0xf73e, 0xf7fb, 0xf8b7, 0xf973,
56 0xfa2f, 0xfaea, 0xfba5, 0xfc60, 0xfd1a, 0xfdd4, 0xfe8e, 0xff47
59 unsigned int intlog2(u32 value
)
62 * returns: log2(value) * 2^24
63 * wrong result if value = 0 (log2(0) is undefined)
66 unsigned int logentry
;
67 unsigned int significand
;
68 unsigned int interpolation
;
70 if (unlikely(value
== 0)) {
75 /* first detect the msb (count begins at 0) */
79 * now we use a logtable after the following method:
81 * log2(2^x * y) * 2^24 = x * 2^24 + log2(y) * 2^24
82 * where x = msb and therefore 1 <= y < 2
83 * first y is determined by shifting the value left
84 * so that msb is bit 31
85 * 0x00231f56 -> 0x8C7D5800
86 * the result is y * 2^31 -> "significand"
87 * then the highest 9 bits are used for a table lookup
88 * the highest bit is discarded because it's always set
89 * the highest nine bits in our example are 100011000
90 * so we would use the entry 0x18
92 significand
= value
<< (31 - msb
);
93 logentry
= (significand
>> 23) & 0xff;
96 * last step we do is interpolation because of the
97 * limitations of the log table the error is that part of
98 * the significand which isn't used for lookup then we
99 * compute the ratio between the error and the next table entry
100 * and interpolate it between the log table entry used and the
101 * next one the biggest error possible is 0x7fffff
102 * (in our example it's 0x7D5800)
103 * needed value for next table entry is 0x800000
104 * so the interpolation is
105 * (error / 0x800000) * (logtable_next - logtable_current)
106 * in the implementation the division is moved to the end for
107 * better accuracy there is also an overflow correction if
108 * logtable_next is 256
110 interpolation
= ((significand
& 0x7fffff) *
111 ((logtable
[(logentry
+ 1) & 0xff] -
112 logtable
[logentry
]) & 0xffff)) >> 15;
114 /* now we return the result */
115 return ((msb
<< 24) + (logtable
[logentry
] << 8) + interpolation
);
117 EXPORT_SYMBOL(intlog2
);
119 unsigned int intlog10(u32 value
)
122 * returns: log10(value) * 2^24
123 * wrong result if value = 0 (log10(0) is undefined)
127 if (unlikely(value
== 0)) {
132 log
= intlog2(value
);
135 * we use the following method:
136 * log10(x) = log2(x) * log10(2)
139 return (log
* 646456993) >> 31;
141 EXPORT_SYMBOL(intlog10
);