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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
|
#include "benchmark_helpers.hpp"
#include "../simd_binary_arithmetic.hpp"
#include <cmath>
using namespace nova;
using namespace std;
namespace {
aligned_array<float, 64> out, in, in2;
typedef float afloat __attribute__ ((__aligned__(16)));
afloat * __restrict__ o = out.begin();
afloat * __restrict__ i1 = in.begin();
afloat * __restrict__ i2 = in2.begin();
void __noinline__ bench_1(unsigned int n)
{
plus_vec_simd(o, i1, i2, n);
}
void __noinline__ bench_2(unsigned int n)
{
plus_vec_simd<64>(out.begin(), in.begin(), in2.begin());
}
// void __noinline__ bench_2a(unsigned int n)
// {
// n /= 8;
// float * i = in.begin();
// float * i2 = in2.begin();
// float * o = out.begin();
// do
// {
// plus_vec_simd<8>(o, i, i2);
// i += 8;
// i2 += 8;
// o += 8;
// }
// while (--n);
// }
void __noinline__ bench_2b(unsigned int n)
{
n /= 16;
float * i = in.begin();
float * i2 = in2.begin();
float * o = out.begin();
do
{
plus_vec_simd<16>(o, i, i2);
i += 16;
i2 += 16;
o += 16;
}
while (--n);
}
void __noinline__ bench_2c(unsigned int n)
{
plus_vec_simd<64>(out.begin(), in.begin(), 1.f);
}
void __noinline__ bench_2d(unsigned int n)
{
n /= 32;
float * i = in.begin();
float * o = out.begin();
do
{
plus_vec_simd<32>(o, i, 1.f);
i += 32;
o += 32;
}
while (--n);
}
void __noinline__ bench_2e(unsigned int n)
{
n /= 16;
float * i = in.begin();
float * o = out.begin();
do
{
plus_vec_simd<16>(o, i, 1.f);
i += 16;
o += 16;
}
while (--n);
}
void __noinline__ bench_3(unsigned int n)
{
for (int i = 0; i != n; ++i)
o[i] = i1[i] + i2[i];
}
void __noinline__ bench_3a(unsigned int n)
{
for (std::size_t i = 0; i != n; ++i)
out[i] = in[i] + in2[i];
}
void __noinline__ bench_4(unsigned int n)
{
for (int i = 0; i != n; i+=4)
{
o[i] = i1[i] + i2[i];
o[i+1] = i1[i+1] + i2[i+1];
o[i+2] = i1[i+2] + i2[i+2];
o[i+3] = i1[i+3] + i2[i+3];
}
}
void __noinline__ bench_4a(unsigned int n)
{
for (std::size_t i = 0; i != n; i+=4)
{
out[i] = in[i] + in2[i];
out[i+1] = in[i+1] + in2[i+1];
out[i+2] = in[i+2] + in2[i+2];
out[i+3] = in[i+3] + in2[i+3];
}
}
void __noinline__ bench_5(unsigned int n)
{
for (int i = 0; i != n; i+=8)
{
o[i] = i1[i] + i2[i];
o[i+1] = i1[i+1] + i2[i+1];
o[i+2] = i1[i+2] + i2[i+2];
o[i+3] = i1[i+3] + i2[i+3];
o[i+4] = i1[i+4] + i2[i+4];
o[i+5] = i1[i+5] + i2[i+5];
o[i+6] = i1[i+6] + i2[i+6];
o[i+7] = i1[i+7] + i2[i+7];
}
}
void __noinline__ bench_5a(float * out, const float * in, const float * in2, unsigned int n)
{
for (std::size_t i = 0; i != n; i+=8)
{
out[i] = in[i] + in2[i];
out[i+1] = in[i+1] + in2[i+1];
out[i+2] = in[i+2] + in2[i+2];
out[i+3] = in[i+3] + in2[i+3];
out[i+4] = in[i+4] + in2[i+4];
out[i+5] = in[i+5] + in2[i+5];
out[i+6] = in[i+6] + in2[i+6];
out[i+7] = in[i+7] + in2[i+7];
}
}
void __noinline__ bench_5b(float * out, const float * in, const float * in2, unsigned int n)
{
for (std::size_t i = 0; i != n; i+=4)
{
float i01 = in[i];
float i02 = in2[i];
float i11 = in[i+1];
float i12 = in2[i+1];
float i21 = in[i+2];
float i22 = in2[i+2];
float i31 = in[i+3];
float i32 = in2[i+3];
/* float i41 = in[i+4]; */
/* float i42 = in2[i+4]; */
/* float i51 = in[i+5]; */
/* float i52 = in2[i+5]; */
/* float i61 = in[i+6]; */
/* float i62 = in2[i+6]; */
/* float i71 = in[i+7]; */
/* float i72 = in2[i+7]; */
out[i] = i01 + i02;
out[i+1] = i11 + i12;
out[i+2] = i21 + i22;
out[i+3] = i31 + i32;
/* out[i+4] = i41 + i42; */
/* out[i+5] = i51 + i52; */
/* out[i+6] = i61 + i62; */
/* out[i+7] = i71 + i72; */
}
}
void __noinline__ bench_5c(float * __restrict__ out, const float * __restrict__ in,
const float * __restrict__ in2, unsigned int n)
{
for (std::size_t i = 0; i != n; i+=8)
{
out[i] = in[i] + in2[i];
out[i+1] = in[i+1] + in2[i+1];
out[i+2] = in[i+2] + in2[i+2];
out[i+3] = in[i+3] + in2[i+3];
out[i+4] = in[i+4] + in2[i+4];
out[i+5] = in[i+5] + in2[i+5];
out[i+6] = in[i+6] + in2[i+6];
out[i+7] = in[i+7] + in2[i+7];
}
}
}
int main(void)
{
out.assign(0.f);
in.assign(0.2f);
in2.assign(0.3f);
const unsigned int iterations = 100000000;
run_bench(boost::bind(bench_1, 64), iterations);
run_bench(boost::bind(bench_2, 64), iterations);
/* run_bench(boost::bind(bench_2a, 64), iterations);*/
run_bench(boost::bind(bench_2b, 64), iterations);
run_bench(boost::bind(bench_2c, 64), iterations);
run_bench(boost::bind(bench_2d, 64), iterations);
run_bench(boost::bind(bench_2e, 64), iterations);
/* run_bench(boost::bind(bench_3, 64), iterations); */
/* run_bench(boost::bind(bench_3a, 64), iterations); */
/* run_bench(boost::bind(bench_4, 64), iterations); */
/* run_bench(boost::bind(bench_4a, 64), iterations); */
/* run_bench(boost::bind(bench_5, 64), iterations); */
/* run_bench(boost::bind(bench_5a, out.begin(), in.begin(), in2.begin(), 64), iterations); */
/* run_bench(boost::bind(bench_5b, out.begin(), in.begin(), in2.begin(), 64), iterations); */
/* run_bench(boost::bind(bench_5c, out.begin(), in.begin(), in2.begin(), 64), iterations); */
}
|
