[parsec-users] Running ROI

Can Hankendi hankendi at gmail.com
Mon Feb 14 17:47:31 EST 2011


Actually what I was trying to ask is if we can fastforward to the ROI in a
real system to have a shorter runtime when running some experiments.



2011/2/14 Gagan Singh <gagansingh86 at gmail.com>

> Hey,
>
> This is an easy one. So, I will try answer.
>
> The things outside ROI are actually setting up data, reading arguments,
> initializing stuff, basically the stuff you would need to setup before you
> enter the parallel phase.
>
> I think you are wishing to only time the ROI rather than the complete
> program. I dont know what the timing presently measures the ROI time or the
> full execution time. But, you can always measure time only for ROI
> (start_time() before ROI and end_time() after ROI) to make sure you are only
> measuring performance for ROI.
>
> In a nutshell, running the initial part is vital, not timing it is up to
> you. Hope this helps!
>
> -G
>
>
> 2011/2/14 Can Hankendi <hankendi at gmail.com>
>
>> Hi,
>>
>> Is there any way to run only the region of interest for Parsec benchmarks
>> on a real system.
>>
>> Thanks
>>
>>
>> _______________________________________________
>> parsec-users mailing list
>> parsec-users at lists.cs.princeton.edu
>> https://lists.cs.princeton.edu/mailman/listinfo/parsec-users
>>
>>
>
> _______________________________________________
> parsec-users mailing list
> parsec-users at lists.cs.princeton.edu
> https://lists.cs.princeton.edu/mailman/listinfo/parsec-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/parsec-users/attachments/20110214/8db64ec4/attachment.htm>


More information about the parsec-users mailing list