-
Notifications
You must be signed in to change notification settings - Fork 11
/
exprtk_vector_benchmark_multithreaded.cpp
221 lines (180 loc) · 8.38 KB
/
exprtk_vector_benchmark_multithreaded.cpp
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
/*
**************************************************************
* C++ Mathematical Expression Toolkit Library *
* *
* ExprTk Multi-Threaded Vector Processing Benchmark *
* Author: Arash Partow (1999-2023) *
* URL: https://www.partow.net/programming/exprtk/index.html *
* *
* Copyright notice: *
* Free use of the Mathematical Expression Toolkit Library is *
* permitted under the guidelines and in accordance with the *
* most current version of the MIT License. *
* https://www.opensource.org/licenses/MIT *
* *
**************************************************************
*/
#include <cstdio>
#include <fstream>
#include <limits>
#include <numeric>
#include <string>
#include <vector>
#include <thread>
#include "exprtk.hpp"
struct test_expression
{
std::size_t cost;
std::string e;
test_expression(const std::size_t c, const std::string& ex)
: cost(c),
e(ex)
{}
};
const test_expression global_expression_list[] =
{
test_expression( 1,"2 * v0" ),
test_expression( 1,"v0 * 2" ),
test_expression( 2,"2v0 + 3" ),
test_expression( 2,"3 + 2v0" ),
test_expression( 5,"(2v0 + 3) * (2v0 + 3)" ),
test_expression( 5,"(3 + 2v0) * (3 + 2v0)" ),
test_expression( 1,"v0 + v1" ),
test_expression( 3,"(v0 + v1) * (v0 - v1)" ),
test_expression( 3,"2v0 + 3v1" ),
test_expression( 3,"2v0 - v1 / 3" ),
test_expression( 2,"v0 * v1 / v2" ),
test_expression( 3,"2(v0 * v1 / v2)" ),
test_expression( 4,"2(v0 / 3 + v1 / 4)" ),
test_expression( 3,"(2v0 - v1 / v2)" ),
test_expression( 4,"3(2v0 - v1 / v2)" ),
test_expression( 5,"7(5v0 * 3v1 / 2v2)" ),
test_expression( 5,"abs(v0 - v1) * abs(v1 - v0)" ),
test_expression( 7,"abs(2v0 - v1) * abs(v1 - 2v0)" ),
test_expression( 9,"abs(2v0 - 3v1) * abs(3v1 - 5v0)" ),
test_expression( 2,"sum(2 * v0)" ),
test_expression( 2,"sum(v0 * 2)" ),
test_expression( 2,"sum(v0 * v1)" ),
test_expression( 3,"sum(2v0 + 3)" ),
test_expression( 3,"sum(3 + 2v0)" ),
test_expression( 6,"sum((2v0 + 3) * (2v0 + 3))" ),
test_expression( 6,"sum((3 + 2v0) * (3 + 2v0))" ),
test_expression( 2,"sum(v0 + v1)" ),
test_expression( 4,"sum((v0 + v1) * (v0 - v1))" ),
test_expression( 4,"sum(2v0 + 3v1)" ),
test_expression( 4,"sum((2v0 - v1 / v2))" ),
test_expression( 5,"sum(3(2v0 - v1 / v2))" ),
test_expression( 4,"sum(abs(v0 * v1) / v2)" ),
test_expression( 5,"sum(v0 + v1) + avg(v0 - v1)" ),
test_expression( 8,"7(sum(abs(5v0 * 3v1) / 2v2))" ),
test_expression( 9,"sum(2v0 + 3v1) + sum(4 / v0 - 5 / v1)"),
test_expression( 6,"sum(abs(v0 - v1) * abs(v1 - v0))" ),
test_expression( 8,"sum(abs(2v0 - v1) * abs(v1 - 2v0))" ),
test_expression(10,"sum(abs(2v0 - 3v1) * abs(3v1 - 5v0))" ),
test_expression( 2,"axpbz(2,v0,3,v1)" ),
test_expression( 2,"axpy(2,v0,v1)" ),
//test_expression( 4,"sum(v0^2.2 + v1^3.3)" ),
//test_expression( 4,"exp(-1 / (v0^2))" ),
//test_expression( 5,"exp(-1 / (v0^2)) / v1" )
};
const std::size_t global_expression_list_size = sizeof(global_expression_list) / sizeof(test_expression);
const std::size_t rounds = 2000;
typedef double numeric_type;
template <typename T>
void run_expression_benchmark(const std::size_t& vec_size,
const std::string& expr_string,
const std::size_t& cost)
{
typedef exprtk::symbol_table<numeric_type> symbol_table_t;
typedef exprtk::expression<numeric_type> expression_t;
typedef exprtk::parser<numeric_type> parser_t;
std::vector<T> v0(vec_size, T(3.1234567890));
std::vector<T> v1(vec_size, T(5.1234567890));
std::vector<T> v2(vec_size, T(7.1234567890));
exprtk::rtl::vecops::package<T> vecops_package;
symbol_table_t symbol_table;
symbol_table.add_vector("v0",v0);
symbol_table.add_vector("v1",v1);
symbol_table.add_vector("v2",v2);
symbol_table.add_package(vecops_package);
expression_t expression;
expression.register_symbol_table(symbol_table);
parser_t parser;
if (!parser.compile(expr_string,expression))
{
printf("[load_expression] - Parser Error: %s\tExpression: %s\n",
parser.error().c_str(),
expr_string.c_str());
return;
}
exprtk::timer timer;
T total = T(0);
timer.start();
for (std::size_t r = 0; r < rounds; ++r)
{
total += expression.value();
}
timer.stop();
if (T(0) != total)
printf("Total Time:%10.7f Rate:%11.3fevals/sec Perf: %5.3fGFLOPS Expression: %s\n",
timer.time(),
rounds / timer.time(),
(rounds * vec_size * cost) / (1e+9 * timer.time()),
expr_string.c_str());
else
printf("run_expression_benchmark() - Error running benchmark for expression: %s\n",
expr_string.c_str());
}
template <typename T>
void run_threaded_benchmark(const std::size_t& vec_size, const std::size_t& thread_cnt)
{
std::vector<std::thread> thread_list;
exprtk::timer timer;
timer.start();
for (std::size_t i = 0; i < thread_cnt; ++i)
{
thread_list.emplace_back(
std::thread([vec_size]()
{
for (std::size_t i = 0; i < global_expression_list_size; ++i)
{
run_expression_benchmark<T>(vec_size,
global_expression_list[i].e,
global_expression_list[i].cost);
}
}));
}
for (auto& t : thread_list)
{
t.join();
}
timer.stop();
unsigned long long total_flop = 0;
for (std::size_t i = 0; i < global_expression_list_size; ++i)
{
total_flop += global_expression_list[i].cost;
}
total_flop = (total_flop * rounds * vec_size * thread_cnt);
printf("Total Time:%10.7f FLOP: %llu Perf: %7.4fGLOPS\n",
timer.time(),
total_flop,
total_flop / (1e9 * timer.time()));
}
int main(int argc, char* argv[])
{
const std::size_t vec_size = ((argc >= 2) ? atoi(argv[1]) : 100000);
const std::size_t thread_cnt = ((argc == 3) ? atoi(argv[2]) : 1);
if (thread_cnt > 36)
return -1;
run_threaded_benchmark<double>(vec_size,thread_cnt);
return 0;
}
/*
Build command:
c++ -pedantic-errors -Wall -Wextra -Werror -Wno-long-long \
-std=c++11 \
-O3 -mtune=native -march=native -ftree-vectorize \
-o exprtk_vector_benchmark_multithreaded \
exprtk_vector_benchmark_multithreaded.cpp \
-L/usr/lib -lstdc++ -lm -lpthread \
*/