time.h & Delay

Sleep in windows will "delay" your program for X milliseconds - as long as your system clock is working correct, it will be precise, no matter what speed your system is (within a few milliseconds assuming no other application loading the system - if your system is heavily loaded, no delay operation is guaranteed anyways). The same applies to the corresponding functions in Linux.

--
Mats

Fine, so what should I do if I want to "delay" my program for X milliseconds (I want it to be precise) no matter if my system clock is working correct?

Quote:

---

Originally Posted by gavra

Fine, so what should I do if I want to "delay" my program for X milliseconds (I want it to be precise) no matter if my system clock is working correct?

---

What do you mean "no matter if my system clock is working correct" - if you have broken hardware, it's not going to be reliable. Period.

--
Mats

XD
hum I checked 2 computers to see how many clocks they do for 1 sec' and the results were diffrent - ?!

Quote:

---

Originally Posted by gavra

XD
hum I checked 2 computers to see how many clocks they do for 1 sec' and the results were diffrent - ?!

---

What do you mean? You mean how many loops of "time2 = clock()" they will do - yes, sure, that will vary from one system to another depending on CPU speed. However, this:

Code:

---

#include <stdio.h>
#include <time.h>
#include <windows.h>

int main()
{
  clock_t t1 = clock(), t2;
  Sleep(1000);
  t2 = clock()-t1;
  printf("time=%d\n", (int)t2);
  return 0;
}

---

should give you the same number on different machines, because your system is ticking at the same rate.

--
Mats

1000 I guess it's the same.
when I do something similar I got 907..

Quote:

---

Originally Posted by gavra

1000 I guess it's the same.
when I do something similar I got 907..

---

Post your code.

But I would also point out that you can not expect the timing in any non-Real-Time-OS to be precise [and RTOS's are only precise if you follow certain rules, e.g. higher priority threads would dictate the real-time behaviour of low priority tasks].

--
Mats

Code:

---

#include <time.h>
#include <stdio.h>

int main(void)
{
  clock_t ticks1 = clock(), ticks2 = ticks1;

  while((ticks2/CLOCKS_PER_SEC-ticks1/CLOCKS_PER_SEC) < 1)
    ticks2=clock();
   
  printf("Took &#37;d ticks to wait one second.\n",ticks2-ticks1);
 
  getchar();
  return 0;
}

---

I don't seem to be able to reproduce it. I rewrote your program to run in an infinite loop and track the min/max values, like this:

Code:

---

       
#include <time.h>
#include <stdio.h>

int main(void)
{
    int minT = 10000, maxT = 0;
    int t = 0;
    do {
        clock_t ticks1 = clock(), ticks2 = ticks1;
           
        while((ticks2/CLOCKS_PER_SEC-ticks1/CLOCKS_PER_SEC) < 1)
            ticks2=clock();
        t = ticks2 - ticks1;
        if (minT > t) minT = t;
        if (maxT < t) maxT = t;
        printf("Took %d ticks to wait one second (%d, %d, min=%d, max=%d).\n",t, (int)ticks1, (int)ticks2, minT, maxT);
  } while(1);
  return 0;
}

---

The first time, I get a start of 15 and end of 1015.

--
Mats

ohh cool it starts with 907 and then all the next get to 1000 - why? XD

Quote:

---

Originally Posted by gavra

ohh cool it starts with 907 and then all the next get to 1000 - why? XD

---

Don't know. I'll play with it a bit and see if I can figure something out.

--
Mats

O:
Ok if this thread won't be here PM me.

Quote:

---

Originally Posted by gavra

O:
Ok if this thread won't be here PM me.

---

I still cant get it BELOW 1015 for the first iteration, but I beleive this is more accurate as it doesn't throw away any of the bits in the input number:

Code:

---

        clock_t ticks1 = clock(), ticks2 = ticks1;
        clock_t target = 1 * CLOCKS_PER_SEC;
           
        while(ticks2-ticks1 < target)
            ticks2=clock();

---

--
Mats