Re: [Gumstix-users] clock()?

[Craig H. <cr...@gu...>]

Here's the function definition in uclibc:

clock_t clock(void)
{
         struct tms xtms;
         unsigned long t;

         times(&xtms);

         t = ((unsigned long) xtms.tms_utime) + xtms.tms_stime;

#ifndef __UCLIBC_CLK_TCK_CONST

# error __UCLIBC_CLK_TCK_CONST not defined!

#elif ((CLOCKS_PER_SEC % __UCLIBC_CLK_TCK_CONST) == 0)

         /* CLOCKS_PER_SEC == k * __UCLIBC_CLK_TCK_CONST for some  
integer k >= 1. */
         return ((t * (CLOCKS_PER_SEC/__UCLIBC_CLK_TCK_CONST)) &  
LONG_MAX);

#else

         /* Unlike the previous case, the scaling factor is not an  
integer.
          * So when tms_utime, tms_stime, or their sum wraps, some of  
the
          * "visible" bits in the return value are affected.  Nothing we
          * can really do about this though other than handle  
tms_utime and
          * tms_stime seperately and then sum.  But since that  
doesn't really
          * buy us much, we don't bother. */

         return ((((t / __UCLIBC_CLK_TCK_CONST) * CLOCKS_PER_SEC)
                          + ((((t % __UCLIBC_CLK_TCK_CONST) *  
CLOCKS_PER_SEC)
                                  / __UCLIBC_CLK_TCK_CONST))
                          ) & LONG_MAX);

#endif
}


That looks OK to me (we fall in the middle category btw, with  
CLOCKS_PER_SEC=1,000,000 and __UCLIBC_CLK_TCK_CONST = 100).  If you  
program isn't doing much other than calling clock() and sleeping in a  
loop, then since the definition of clock() is based on *that  
process*'s CPU activity (user+system), and since while a process is  
waiting in sleep() it's not consuming any user or CPU time, "t" will  
only increment very very very slowly; which means that it unless  
you're actually doing some CPU usage in your app, clock() will appear  
to not be ticking.

If you're looking for wallclock time instead of CPU-usage time, then  
check time() or gettimeofday() instead.

C

On Jan 25, 2006, at 1:25 PM, Bryan Adams wrote:

> hi:
>
> has anyone used the calls to clock() (in time.h) successfully?  i  
> seem to get the same number (50000 on some runs, 60000 on others)  
> on every call to clock().  thoughts?
>
> --bryan
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through  
> log files
> for problems?  Stop!  Download the new AJAX search engine that makes
> searching your log files as easy as surfing the  web.  DOWNLOAD  
> SPLUNK!
> http://sel.as-us.falkag.net/sel? 
> cmd=lnk&kid=103432&bid=230486&dat=121642
> _______________________________________________
> gumstix-users mailing list
> gum...@li...
> https://lists.sourceforge.net/lists/listinfo/gumstix-users