blob: b6ed0024d085dfc214b100f73eed06d20c84ec35 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
|
/*-----------------------------------------------------------------
LOG
GEM - Graphics Environment for Multimedia
include file for SIMD
Copyright (c) 1997-1999 Mark Danks. mark@danks.org
Copyright (c) Günther Geiger. geiger@epy.co.at
Copyright (c) 2001-2011 IOhannes m zmölnig. forum::für::umläute. IEM. zmoelnig@iem.at
For information on usage and redistribution, and for a DISCLAIMER OF ALL
WARRANTIES, see the file, "GEM.LICENSE.TERMS" in this distribution.
-----------------------------------------------------------------*/
/*
* compiler-issues:
*
* gcc: when gcc is invoked with "-mmmx" (or "-msse2" or "-maltivec")
* the defines __MMX__ (or corresponding) will be defined automatically
*
* vc6: you will have to install the microsoft processor-pack to use MMX/SSE2
* you have to have the sp5 for vc6 installed (note: do not install sp6!!)
* vc6/vc7: (i think) you need to define __MMX__ (and friends) by hand
*/
#ifndef _INCLUDE__GEM_UTILS_SIMD_H_
#define _INCLUDE__GEM_UTILS_SIMD_H_
#define GEM_VECTORALIGNMENT 128
const int GEM_SIMD_NONE=0;
const int GEM_SIMD_MMX=1;
const int GEM_SIMD_SSE2=2;
const int GEM_SIMD_ALTIVEC=3;
#if defined __APPLE__ && defined __VEC__
# ifndef __APPLE_ALTIVEC__
# undef __VEC__
# endif
#endif
/* include for SIMD on PC's */
#ifdef __SSE2__
#include <emmintrin.h>
// for icc this should be <dvec.h>
typedef union{
unsigned char c[16];
__m128i v;
} vector_128;
#elif defined __VEC__
/* for AltiVec (PowerPC) */
typedef union{
unsigned char c[16];
vector unsigned char v;
} vector_128;
#endif
#if defined __MMX__
# include <mmintrin.h>
// for icc this should be <ivec.h>
typedef union{
__m64 v; unsigned char c[8];
} vector64i;
#endif
#ifdef __SSE__
#include <xmmintrin.h>
typedef union{
__m128 m; float f[4];
} vector128f;
#endif
#include "Gem/ExportDef.h"
/* this is a help-class to query the capabilities of the cpu
* whenever you want to use SIMD-optimized code, you should
* make sure you chose the code-block based on the "cpuid" value
* of this class and NOT simply on preprocessor defines.
*
* this class needs only be instantiated once (and it is done in GemMan)
* this sets the cpuid
*/
class GEM_EXTERN GemSIMD
{
public:
GemSIMD(void);
virtual ~GemSIMD(void);
/* this gets the "cpuid" (something like GEM_SIMD_NONE) */
static int getCPU(void);
/* change the cpuid returned by getCPU()
* you can only set the cpuid to something that is actually supported
* by your processor
*/
static int requestCPU(int cpuid);
/* performs a runtime-check (if possible) to determine the capabilities
* of the CPU
* sets realcpuid appropriately and returns this value
*/
static int simd_runtime_check(void);
private:
/* this is the maximum capability of the CPU */
static int realcpuid;
/* this is the current choosen capability (normally this equals realcpuid) */
static int cpuid;
};
#endif /* _INCLUDE__GEM_UTILS_SIMD_H_ */
|
