|
1 /* Copyright (c) 2003-2008 Timothy B. Terriberry |
|
2 Copyright (c) 2008 Xiph.Org Foundation */ |
|
3 /* |
|
4 Redistribution and use in source and binary forms, with or without |
|
5 modification, are permitted provided that the following conditions |
|
6 are met: |
|
7 |
|
8 - Redistributions of source code must retain the above copyright |
|
9 notice, this list of conditions and the following disclaimer. |
|
10 |
|
11 - Redistributions in binary form must reproduce the above copyright |
|
12 notice, this list of conditions and the following disclaimer in the |
|
13 documentation and/or other materials provided with the distribution. |
|
14 |
|
15 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
|
16 ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
|
17 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
|
18 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER |
|
19 OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
|
20 EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
|
21 PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
|
22 PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF |
|
23 LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING |
|
24 NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS |
|
25 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
|
26 */ |
|
27 |
|
28 /*Some common macros for potential platform-specific optimization.*/ |
|
29 #include "opus_types.h" |
|
30 #include <math.h> |
|
31 #include <limits.h> |
|
32 #include "arch.h" |
|
33 #if !defined(_ecintrin_H) |
|
34 # define _ecintrin_H (1) |
|
35 |
|
36 /*Some specific platforms may have optimized intrinsic or OPUS_INLINE assembly |
|
37 versions of these functions which can substantially improve performance. |
|
38 We define macros for them to allow easy incorporation of these non-ANSI |
|
39 features.*/ |
|
40 |
|
41 /*Modern gcc (4.x) can compile the naive versions of min and max with cmov if |
|
42 given an appropriate architecture, but the branchless bit-twiddling versions |
|
43 are just as fast, and do not require any special target architecture. |
|
44 Earlier gcc versions (3.x) compiled both code to the same assembly |
|
45 instructions, because of the way they represented ((_b)>(_a)) internally.*/ |
|
46 # define EC_MINI(_a,_b) ((_a)+(((_b)-(_a))&-((_b)<(_a)))) |
|
47 |
|
48 /*Count leading zeros. |
|
49 This macro should only be used for implementing ec_ilog(), if it is defined. |
|
50 All other code should use EC_ILOG() instead.*/ |
|
51 #if defined(_MSC_VER) && (_MSC_VER >= 1400) |
|
52 # include <intrin.h> |
|
53 /*In _DEBUG mode this is not an intrinsic by default.*/ |
|
54 # pragma intrinsic(_BitScanReverse) |
|
55 |
|
56 static __inline int ec_bsr(unsigned long _x){ |
|
57 unsigned long ret; |
|
58 _BitScanReverse(&ret,_x); |
|
59 return (int)ret; |
|
60 } |
|
61 # define EC_CLZ0 (1) |
|
62 # define EC_CLZ(_x) (-ec_bsr(_x)) |
|
63 #elif defined(ENABLE_TI_DSPLIB) |
|
64 # include "dsplib.h" |
|
65 # define EC_CLZ0 (31) |
|
66 # define EC_CLZ(_x) (_lnorm(_x)) |
|
67 #elif __GNUC_PREREQ(3,4) |
|
68 # if INT_MAX>=2147483647 |
|
69 # define EC_CLZ0 ((int)sizeof(unsigned)*CHAR_BIT) |
|
70 # define EC_CLZ(_x) (__builtin_clz(_x)) |
|
71 # elif LONG_MAX>=2147483647L |
|
72 # define EC_CLZ0 ((int)sizeof(unsigned long)*CHAR_BIT) |
|
73 # define EC_CLZ(_x) (__builtin_clzl(_x)) |
|
74 # endif |
|
75 #endif |
|
76 |
|
77 #if defined(EC_CLZ) |
|
78 /*Note that __builtin_clz is not defined when _x==0, according to the gcc |
|
79 documentation (and that of the BSR instruction that implements it on x86). |
|
80 The majority of the time we can never pass it zero. |
|
81 When we need to, it can be special cased.*/ |
|
82 # define EC_ILOG(_x) (EC_CLZ0-EC_CLZ(_x)) |
|
83 #else |
|
84 int ec_ilog(opus_uint32 _v); |
|
85 # define EC_ILOG(_x) (ec_ilog(_x)) |
|
86 #endif |
|
87 #endif |