i m a student oly.

[kanagaraja]

#include<stdio.h>
#include"time.h"
int main()
{
clock_t start,end;
double value;
start=clock();
/*
my proj code
*/
end=clock();
value=(end-start)/(double)CLOCKS_PER_SEC;
printf("the total time is %g",value);
}

i have done this code in gcc linux i3 core..i got same output zero(0) oly always. could any one help me pls???
and give me correct code..i want to calculate total execution time of this code...pls help me to complete my proj..

[donbock]

Your proj code runs much faster than the resolution of clock(). There are two ways to work around that limitation: either use a timing function that has higher resolution or slow down your proj code. The easiest approach is the second: run your proj code many times (1,000, 10,000, ... or 1,000,000 times) and divide the reported time by that repeat-count.

[kanagaraja]

thank u sir..but i did not get,could you send me sample code.please.

[donbock]

Insert a loop that executes lines 8-10 (your proj code) many times. Modify line 12 to also divide by that number of iterations. Adjust the number of iterations until you get a reasonable answer.

[Banfa]

Note that with code like this you are measuring from the time it started until the time it stopped. This means on a multitasking OS, like Linux or Windows, that you may also include the processing time of any other applications and processes that are also running at that time.

To reduce the effect of this you should shut down as many other applications as possible before running this, you should give the system time to reach an equilibrium, that is for example if as a result of shutting down lots of tasks the OS decides to change the swapfile size you should wait for that to finish. And you should run you measuring application several times and discard any results that appear to be spikes in the data set.

[kanagaraja]

thanku!!i need one more help..i want to split entire process of my coding into each and every core of my system...is there any command available for that in openMP?