Timing C program always returns 0 seconds.

Hi,

I was looking at the programming FAQ and I saw an example method of timing a program in c. So, I tried out the example below and compiled it using the plain gcc under ubuntu. eg.
gcc version 4.1.2 20060928 (prerelease) (Ubuntu 4.1.1-13ubuntu5). After running the compiled executable from below - I always get 0.000 seconds despite waiting a few seconds before hitting return. Would I have overlooked something?

Code:

---

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

int main ( void )
{
  clock_t start, end;

  /* Start the timer */
  start = clock();

  printf ( "Please wait a few moments and hit return\n" );
  getchar();

  /* End the timer */
  end = clock();

  /* Print out the difference */
  printf ( "The interval was: %f seconds\n",
    (double)( end - start ) / (double)CLOCKS_PER_SEC );

  return 0;
}

---

clock() measures CPU time used by the program.
The seconds which tick by waiting for the user to type in something don't count to this total.

May be time function would get you the actual time. But you could get the time in terms of second. Perhaps check the following code.

Code:

---

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

int main ( void )
{
  clock_t start, end;

  /* Start the timer */
  start = time(NULL);

  printf ( "Please wait a few moments and hit return\n" );
  getchar();

  /* End the timer */
  end = time(NULL);

  /* Print out the difference */
  printf ( "The interval was: %f seconds\n",
    (double)( end - start ) / (double)CLOCKS_PER_SEC );

    getchar();
  return 0;
}
/* my ouput
Please wait a few moments and hit return

The interval was: 0.003000 seconds
*/

---

ssharish2005

> clock_t start, end;
time() returns time_t, not clock_t

> ( end - start ) / (double)CLOCKS_PER_SEC );
Use difftime() to calculate the difference between two times in seconds.

If you want to try the clock() difference on a real-world example, I provide you with my code for measuring different sort methods:

Code:

---

#define ARRAY_SIZE 10000

void swap(int *a, int *b)
{
    int tmp;
   
    tmp = *a;
    *a = *b;
    *b = tmp;
}

void bubbleSort(int *array, int size)
{
    int swapped = 0;
    int x,tmp;   
    do
    {
        swapped = 0;
        for (x = 0; x < size-1;x++)
        {
            if (array[x]>array[x+1])
            {
                swap(&array[x], &array[x+1]);
                swapped = 1;
            }
        }
       
    } while (swapped);
}

int main(int argc, char *argv[])
{
    int array[ARRAY_SIZE];
    int i;
    clock_t start, end;
   
    /* initialize array with ARRAY_SIZE random numbers */
    srand(time(NULL));
    for (i = 0; i < ARRAY_SIZE; i++)
    {
        array[i] = rand(); // random number between 0 and RAND_MAX
    }
    /* start the timer */
    start = clock();
    bubbleSort(array, ARRAY_SIZE);
    end = clock();
   
    /* Print out the difference */
    printf ( "The interval was: %f seconds\n", (double)( end - start ) / (double)CLOCKS_PER_SEC );

    getchar();
    return 0;
}

---

Quote:

---

Originally Posted by Salem

clock() measures CPU time used by the program.
The seconds which tick by waiting for the user to type in something don't count to this total.

---

Aaah yes - thanks! So, the code above is correct for cpu time usage rather than measure the time passed since start and end.

So to really test the above, I would need to put something cpu intensive enough in a for loop or something to get a measure of cpu time usage.

And thanks Salem - that gives me the actual time passed between start and finish.

I guess from a program duration pov we are interested in the cpu time taken to run a program - at least that what I was trying to do. :)

That's nice of you Koni. Appreciated! I will have some fun playing with the timings. Makes a change friom SQL tuning. :)