法一： ros

#include <ros/ros.h>
ros::Time begin = ros::Time::now();
... do some work ...
ros::Time end = ros::Time::now();
ros::Duration diff = begin - end;
double sec = diff.toSec();
std::cout << "Time used:  " << sec << " s";
1
2
3
4
5
6
7

另外ros还有睡眠和指定频率循环的功能：

ros::Duration(0.5).sleep(); // sleep for half a second

/// 以10Hz的频率执行以下循环
ros::Rate r(10); // 10 hz
while (ros::ok())
{
  ... do some work ...
  r.sleep();
}
1
2
3
4
5
6
7
8
9

还可以使用 ros::Timer比 ros::Rate更灵活。具体参见：https://www.ncnynl.com/archives/201701/1284.html

法二： chrono

#include <chrono>
std::chrono::steady_clock::time_point start = std::chrono::steady_clock::now();
... do some work ...
std::chrono::steady_clock::time_point end = std::chrono::steady_clock::now();
std::chrono::duration<double> diff = end - start;
std::cout << "Time used:  " << diff.count() << " s";
1
2
3
4
5
6

法三： omp

基于OpenMP，没有类的封装，用起来更直观一些。（可能效率也更高点？）

#include <omp.h>
double start = omp_get_wtime(); //单位秒
... do some work ...
double end = omp_get_wtime(); //单位秒
double diff = end - start;
std::cout << "Time used:  " << diff << " s";
1
2
3
4
5
6

测试1

//
// Created by daybeha on 2022/6/27.
//

#include <iostream>
#include <chrono>
#include <ros/ros.h>
#include <omp.h>

using namespace std;

int main(int argc, char** argv){
    ros::init(argc,argv, "time_acquire");
    ros::NodeHandle n;
    ros::Rate r(50000);

    // 法一
    cout << "ros:" << endl;
    ros::Time begin_r = ros::Time::now();
    std::cout << "Start  " << begin_r.toSec() << " s"<< endl;
    r.sleep();
    ros::Time end_r = ros::Time::now();
    std::cout << "End  " << end_r.toSec() << " s"<< endl;
    ros::Duration diff_r = end_r - begin_r;
    double sec = diff_r.toSec();
    std::cout << "Time used:  " << sec << " s" << endl;


    // 法二
    cout << endl << "chrono:" << endl;
    std::chrono::steady_clock::time_point start = std::chrono::steady_clock::now();
    std::cout << "Start  " << start.time_since_epoch().count() << " s"<< endl;
    r.sleep();
    std::chrono::steady_clock::time_point end = std::chrono::steady_clock::now();
    std::cout << "End  " << end.time_since_epoch().count() << " s"<< endl;
    std::chrono::duration<double> diff = end - start;
    std::cout << "Time used:  " << diff.count() << " s"<< endl;

    // 法三
    cout << endl << "OpenMP:" << endl;
    double start_o = omp_get_wtime(); //单位秒
    std::cout << "Start  " << start_o << " s"<< endl;
    r.sleep();
    double end_o = omp_get_wtime(); //单位秒
    std::cout << "End  " << end_o << " s"<< endl;
    double diff_o = end_o - start_o;
    std::cout << "Time used:  " << diff_o << " s"<< endl;
}
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

rate=100Hz

rate=1000Hz

rate=10000Hz

rate=50000Hz

rate=100000Hz

可以看到，随着频率增大，三种方法的时间都不怎么准了（当然r.sleep应该也不准了……），其中chrono最差。 ros整体耗时更高。

测试2

//
// Created by daybeha on 2022/6/27.
//

#include <iostream>
#include <chrono>
#include <ros/ros.h>
#include <omp.h>

using namespace std;

void test(){
    for (int i = 0; i < 10000; ++i) {
        for (int j = 0; j < 1000; ++j) {
            int a = i*j;
        }
    }
}

int main(int argc, char** argv){
    ros::init(argc,argv, "time_acquire");
    ros::NodeHandle n;
    ros::Rate r(100000);

    // 法一
    cout << "ros:" << endl;
    ros::Time begin_r = ros::Time::now();
    std::cout << "Start  " << begin_r.toSec() << " s"<< endl;
//    r.sleep();
    test();
    ros::Time end_r = ros::Time::now();
    std::cout << "End  " << end_r.toSec() << " s"<< endl;
    ros::Duration diff_r = end_r - begin_r;
    double sec = diff_r.toSec();
    std::cout << "Time used:  " << sec << " s" << endl;


    // 法二
    cout << endl << "chrono:" << endl;
    std::chrono::steady_clock::time_point start = std::chrono::steady_clock::now();
    std::cout << "Start  " << start.time_since_epoch().count() << " s"<< endl;
//    r.sleep();
    test();
    std::chrono::steady_clock::time_point end = std::chrono::steady_clock::now();
    std::cout << "End  " << end.time_since_epoch().count() << " s"<< endl;
    std::chrono::duration<double> diff = end - start;
    std::cout << "Time used:  " << diff.count() << " s"<< endl;

    // 法三
    cout << endl << "OpenMP:" << endl;
    double start_o = omp_get_wtime(); //单位秒
    std::cout << "Start  " << start_o << " s"<< endl;
//    r.sleep();
    test();
    double end_o = omp_get_wtime(); //单位秒
    std::cout << "End  " << end_o << " s"<< endl;
    double diff_o = end_o - start_o;
    std::cout << "Time used:  " << diff_o << " s"<< endl;
}
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

结论： chrono获取时间的效率最高，OpenMP其次，ros最差！

Reference

C++ API文档
ROS与C++入门教程-Time(时间)
omp_get_wtime | Microsoft Docs