使用DTSC（Read Time Stamp Counter）指令做高精度時間處理，比QueryPerformanceCounter效率要高出一個數量級

// Test.cpp : Defines the entry point for the console application.
//

#pragma once

#ifndef _WIN32_WINNT        // Allow use of features specific to Windows XP or later.                  
#define _WIN32_WINNT 0x0501    // Change this to the appropriate value to target other versions of Windows.
#endif                       

#include <stdio.h>
#include <tchar.h>

#include "Windows.h"

__declspec( naked ) inline unsigned __int64 GetCPUCounter()
{
    __asm _emit 0x0F;
    __asm _emit 0x31;
    __asm ret;
}


int _tmain(int argc, _TCHAR* argv[])
{
    LARGE_INTEGER freq;
    QueryPerformanceFrequency(&freq);
   
    unsigned __int64 counter;
   
    LARGE_INTEGER counter1;
    counter = GetCPUCounter();
    QueryPerformanceCounter(&counter1);
   
    unsigned __int64 diff = (GetCPUCounter() - counter);
   
    unsigned __int64 counter2 = GetCPUCounter();
    GetCPUCounter();
    unsigned __int64 diff1 = (GetCPUCounter() - counter2);
   
    printf("diff: %d/n", diff - diff1);
   
    system("pause");
     return 0;
}