On Wed, Jun 10, 2009 at 09:38:29AM -0700, Kok, Auke wrote:
> Madhu Sudan Hari wrote:
> > Thanks a lot Auke. One more question. What do you think of using something 
> > like
> > "stress"- a workload generator, for this purpose ?
> 
> 
> I don't have a particular opinion about any workload simulator other than that
> they simulator needs to be appropriate for the research being done. It makes 
> no
> sense to measure data with powertop in a workload which consumes > 10% cpu 
> time
> (or some arbitrary number greater than zero) since powertop is designed to 
> measure
> _idle_ performance of the OS. (IOW "how good can my OS be idle under a 
> minimal or
> zero load").

I beg to differ. I personally used powertop to measure the (energy-)load that a
daemon exerts on my system under different configurations and with different
kernel module options. (IOW: "how often does my daemon wake up my system in
this configuration.")

About stress test: I don't think you'll get any good information out of a
stress test unless the stress load resembles the load that you expect to get
in later use.

cheers
-henrik

_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power

Reply via email to