CPP中速度性能的测试功能

Question

我创建了一个简单的函数来将任何小写字母 az 转换为大写，这个问题可能不是问题，但每个测试都返回 0。如果我添加 system("pause") 我可以看到一个新值，指示长度暂停。

是否有更准确的方法来测试速度，或者这实际上是否正确？我想将它与其他函数进行比较，看看它是否比标准函数转换得更快。

char* ToUppercase(char* Input)
{
    int Len = Length(Input);
    for (int i = 0; i < Len; i++)
    {
        short keycode = static_cast<short>(Input[i]);
        if (keycode >= 97 && keycode <= 122)
            Input[i] -= 32;
    }
    return Input;
}

我用来测试的当前计时器是（由其他人创建的）

template<typename TimeT = std::chrono::milliseconds>
struct measure
{
    template<typename F, typename ...Args>
    static typename TimeT::rep execution(F func, Args&&... args)
    {
        auto start = std::chrono::system_clock::now();
        func(std::forward<Args>(args)...);
        auto duration = std::chrono::duration_cast< TimeT>
            (std::chrono::system_clock::now() - start);
        return duration.count();
    }
};

打电话给我：

void Debug()
{
    char Buffer[10000] = "aaaa /..../ aaaa";
    MyStringControl::ToUppercase(Buffer);
}
int main()
{
    std::cout << measure<std::chrono::nanoseconds>::execution(Debug);
}

Answer 1

你看过std::chrono::high_resolution_clock吗？

这是一个例子：

#include <iostream>
#include <ctime>
#include <ratio>
#include <chrono>

template<typename TimeT = std::chrono::milliseconds>
struct measure
{
    template<typename F, typename ...Args>
    static typename TimeT::rep execution(F func, Args&&... args)
    {
        auto start = std::chrono::high_resolution_clock::now();
        func(std::forward<Args>(args)...);
        auto duration = std::chrono::duration_cast< TimeT>
                (std::chrono::high_resolution_clock::now() - start);
        return duration.count();
    }
};

int total = 0;

void test()
{
    int foo = 0;
    for (int i=0; i<1000; ++i) ++foo;
    total += foo;
}

int main ()
{
    using namespace std::chrono;

    for (int i = 0; i < 30; ++i)
    {
        total = 0;
        auto t = measure<std::chrono::nanoseconds>::execution(test);
        std::cout << "Calculated total = " << total << " in " << t << " ns." << std::endl;
    }    
    return 0;
}

这给出了：

Calculated total = 1000 in 64 ns.
Calculated total = 1000 in 21 ns.
Calculated total = 1000 in 22 ns.
Calculated total = 1000 in 21 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 15 ns.
Calculated total = 1000 in 13 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 13 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 13 ns.
Calculated total = 1000 in 21 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 15 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 15 ns.
Calculated total = 1000 in 22 ns.
Calculated total = 1000 in 21 ns.
Calculated total = 1000 in 20 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 14 ns.
Calculated total = 1000 in 20 ns.
Calculated total = 1000 in 20 ns.
Calculated total = 1000 in 21 ns.
Calculated total = 1000 in 20 ns.
Calculated total = 1000 in 15 ns.
Calculated total = 1000 in 15 ns.
Calculated total = 1000 in 15 ns.
Calculated total = 1000 in 14 ns.

Answer 2

函数运行 1000000 次，结果除以 1000000。你可以使用高精度定时器，但由于硬件的古怪，它更容易出现不准确的情况。

编辑：

您希望对函数本身进行 1000,000 次调用，并且只对计时器进行一次调用：

    auto start = std::chrono::system_clock::now();
    for (size_t counter = 0; counter<1000000; ++counter)
         func(std::forward<Args>(args)...);
    auto duration = std::chrono::duration_cast< TimeT>
        (std::chrono::system_clock::now() - start)/1000000;
    return duration.count();

Answer 3

您的函数 Debug 什么都不做，您的编译器可能会弄清楚这一点，因此您要做的就是计时您可以连续两次调用 now 的速度。

采取措施确保您尝试计时的代码不会被优化掉。例如以某种方式使用它的输出，或者给它__attribute__((noinline))（如果你也不介意计算实际函数调用的成本）或其他东西。

（此外，如果您希望从计时中获得任何有用的精度，则您的功能需要比时钟分辨率长得多）