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
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
|
#include<simdfuncs.h>
#include<stdio.h>
#include<string.h>
typedef void (*simd_func)(float*);
int check_simd_implementation(float *four,
const float *four_initial,
const char *simd_type,
const float *expected,
simd_func fptr,
const int blocksize) {
int rv = 0;
memcpy(four, four_initial, blocksize*sizeof(float));
printf("Using %s.\n", simd_type);
fptr(four);
for(int i=0; i<blocksize; i++) {
if(four[i] != expected[i]) {
printf("Increment function failed, got %f expected %f.\n", four[i], expected[i]);
rv = 1;
}
}
return rv;
}
int main(void) {
static const float four_initial[4] = {2.0, 3.0, 4.0, 5.0};
ALIGN_16 float four[4];
const float expected[4] = {3.0, 4.0, 5.0, 6.0};
int r=0;
const int blocksize = 4;
/*
* Test all implementations that the current CPU supports.
*/
#if HAVE_NEON
if(neon_available()) {
r += check_simd_implementation(four,
four_initial,
"NEON",
expected,
increment_neon,
blocksize);
}
#endif
#if HAVE_AVX2
if(avx2_available()) {
r += check_simd_implementation(four,
four_initial,
"AVX2",
expected,
increment_avx2,
blocksize);
}
#endif
#if HAVE_AVX
if(avx_available()) {
r += check_simd_implementation(four,
four_initial,
"AVC",
expected,
increment_avx,
blocksize);
}
#endif
#if HAVE_SSE42
if(sse42_available()) {
r += check_simd_implementation(four,
four_initial,
"SSR42",
expected,
increment_sse42,
blocksize);
}
#endif
#if HAVE_SSE41
if(sse41_available()) {
r += check_simd_implementation(four,
four_initial,
"SSE41",
expected,
increment_sse41,
blocksize);
}
#endif
#if HAVE_SSSE3
if(ssse3_available()) {
r += check_simd_implementation(four,
four_initial,
"SSSE3",
expected,
increment_ssse3,
blocksize);
}
#endif
#if HAVE_SSE3
if(sse3_available()) {
r += check_simd_implementation(four,
four_initial,
"SSE3",
expected,
increment_sse3,
blocksize);
}
#endif
#if HAVE_SSE2
if(sse2_available()) {
r += check_simd_implementation(four,
four_initial,
"SSE2",
expected,
increment_sse2,
blocksize);
}
#endif
#if HAVE_SSE
if(sse_available()) {
r += check_simd_implementation(four,
four_initial,
"SSE",
expected,
increment_sse,
blocksize);
}
#endif
#if HAVE_MMX
if(mmx_available()) {
r += check_simd_implementation(four,
four_initial,
"MMX",
expected,
increment_mmx,
blocksize);
}
#endif
r += check_simd_implementation(four,
four_initial,
"fallback",
expected,
increment_fallback,
blocksize);
return r;
}
|
