Hi Christian,
Sorry for the long delay. Your mail was lost at the botton of my bag.
Yes, you are correct. In case one alternates with start/stop, a TTimer::Stop
was missing in TBenchmark::Stop.
Now fixed in CVS.
Thanks for reporting.
Rene Brun
Christian Veelken wrote:
>
> Dear ROOTers,
>
> I am using the TBenchmark class with ROOT 3.02_06 to evaluate the run time
> of different functions in a class.
>
> Before calling the function fun1 I do
> TBenchmark::Start("fun1")
> and before calling the next function fun2 I do
> TBenchmark::Stop("fun1")
> TBenchmark::Start("fun2")
> ...
> and so on.
>
> The purpose is to see which of my functions is the most time-consuming.
>
> In the output from
> TBenchmark::Summary
> I get very similar values for all functions which I cannot believe and
> which are also very near the runtime of the entire program each.
>
> It appears to me that in
> TBenchmark::Stop
> there could be a
>
> fTimer[bench].Stop();
>
> statement missing so that each timer is running and running and never
> stops ?
>
> Regards,
>
> Christian Veelken
> H1 Collaboration, DESY
> __________________________________________________________________________
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:42 MET