c++ 前自增/后自增操作符效率分析

更新時(shí)間：2021年01月07日 09:32:03 作者：xhubobo

這篇文章主要介紹了c++ 前自增/后自增操作符效率分析，幫助大家更好的理解和學(xué)習(xí)c++,感興趣的朋友可以了解下

1、前自增/后自增操作符示例

class Integer
{
public:
    // ++i  first +1,then return new value
    Integer &operator++()
    {
        value_ += 1;
        return *this;
    }
 
    // i++  first save old value,then +1,last return old value
    Integer operator++(int)
    {
        Integer old = *this;
        value_ += 1;
        return old;
    }
 
private:
    int value_;
};

2、分別基于內(nèi)置數(shù)據(jù)類型和自定義數(shù)據(jù)類型做測(cè)試

#include <iostream>
#include <vector>
#include <windows.h>
 
int main()
{
    const int sizeInt = 0x00fffffe;
    const int sizeVec = 0x000ffffe;
 
    LARGE_INTEGER frequency;
    QueryPerformanceFrequency(&frequency);
 
    {
        int* testValue = new int[sizeInt];
 
        LARGE_INTEGER start;
        LARGE_INTEGER stop;
        QueryPerformanceCounter(&start);
        for (int i = 0; i < sizeInt; ++i)
        {
            testValue[i]++;
        }
        QueryPerformanceCounter(&stop);    
 
        const auto interval = static_cast<double>(stop.QuadPart - start.QuadPart);
        const auto timeSpan = interval / frequency.QuadPart * 1000.0; //ms
        std::cout << "i++ " << sizeInt << " times takes " << timeSpan << "ms." << std::endl;
 
        delete[] testValue;
    }
    {
        int* testValue = new int[sizeInt];
 
        LARGE_INTEGER start;
        LARGE_INTEGER stop;
        QueryPerformanceCounter(&start);
        for (int i = 0; i < sizeInt; ++i)
        {
            ++testValue[i];
        }
        QueryPerformanceCounter(&stop);    
 
        const auto interval = static_cast<double>(stop.QuadPart - start.QuadPart);
        const auto timeSpan = interval / frequency.QuadPart * 1000.0; //ms
        std::cout << "++i " << sizeInt << " times takes " << timeSpan << "ms." << std::endl;
 
        delete[] testValue;
    }
 
    {
        const std::vector<int> testVec(sizeVec);
        LARGE_INTEGER start;
        LARGE_INTEGER stop;
        QueryPerformanceCounter(&start);
        for (auto iter = testVec.cbegin(); iter != testVec.cend(); iter++)
        {
        }
        QueryPerformanceCounter(&stop);
 
        const auto interval = static_cast<double>(stop.QuadPart - start.QuadPart);
        const auto timeSpan = interval / frequency.QuadPart * 1000.0; //ms
        std::cout << "iterator++ " << sizeVec << " times takes " << timeSpan << "ms." << std::endl;
    }
    {
        const std::vector<int> testVec(sizeVec);
        LARGE_INTEGER start;
        LARGE_INTEGER stop;
        QueryPerformanceCounter(&start);
        for (auto iter = testVec.cbegin(); iter != testVec.cend(); ++iter)
        {
        }
        QueryPerformanceCounter(&stop);
 
        const auto interval = static_cast<double>(stop.QuadPart - start.QuadPart);
        const auto timeSpan = interval / frequency.QuadPart * 1000.0; //ms
        std::cout << "++iterator " << sizeVec << " times takes " << timeSpan << "ms." << std::endl;
    }
 
    return 0;
}

3、五次執(zhí)行結(jié)果

4、結(jié)果分析及結(jié)論

從上面的執(zhí)行結(jié)果可以看出來(lái)，對(duì)int類型的測(cè)試中前自增和后自增耗費(fèi)時(shí)間基本不變；而對(duì)std::vector中iterator的測(cè)試顯示，前自增所耗費(fèi)的時(shí)間幾乎是后自增的一半。這是因?yàn)?，在后自增的操作中，?huì)首先生成原始對(duì)象的一個(gè)副本，然后將副本中的值加1后返回給調(diào)用者，這樣一來(lái)每執(zhí)行一次后自增操作，就會(huì)增加一個(gè)對(duì)象副本，效率自然降低了。

因此可以得出結(jié)論：對(duì)于C++內(nèi)置類型的自增而言，前自增、后自增的效率相差不大；對(duì)于自定義類型（類、結(jié)構(gòu)體）的自增操作而言，前自增的效率幾乎比后自增大一倍。