Question

I am working towards achieving +/- ms timing precision. I want the time between threads to be 10ms, but when I measure the time I get something closer to 15ms.

I am trying to understand if the problem is due to the way I am measuring time or if I am measuring the time accurately and there is some delay introduced by CreateTimerQueueTimer

My code looks like

#include "stdafx.h"
#include <iostream>
#include <Windows.h>
#include <chrono>
#include <ctime>
using namespace std;
int current;
long long* toFill;
void talk()
    {
    chrono::time_point<chrono::system_clock> tp = \
        chrono::system_clock::now();
    toFill[current++]=chrono::duration_cast<chrono::milliseconds>(tp.time_since_epoch()).count() ;
    }

int _tmain(int argc, _TCHAR* argv[])
    {
    HANDLE hTimer;
    current = 0;
    toFill = new long long[1000];
    CreateTimerQueueTimer(&hTimer,NULL, 
        (WAITORTIMERCALLBACK)talk,NULL,
        0,10,0);
    _sleep(3000);
    DeleteTimerQueueTimer(NULL,hTimer,NULL);
    for (int i = 1;i<current;i++)
        {
        cout << i << " : " << toFill[i]-toFill[i-1]<< endl;
        }
    return 0;
    }

Output looks like

...
161 : 16 <-- Should be 10
162 : 15
163 : 16
164 : 16
...

Answer 1

The accuracy of timers and the real time clock in Windows is limited by the clock tick interrupt rate. Which by default tocks 64 times per second, 1/64 sec = 15.625 msec. Just like you saw.

Increasing that rate is actually possible, call timeBeginPeriod(10) at the start of your program to get 10 msec accuracy.