>-----Original Message----- >From: Vaidyanathan Srinivasan [mailto:[EMAIL PROTECTED] >Sent: Wednesday, February 27, 2008 12:34 AM >To: Pallipadi, Venkatesh >Cc: Martin Wilck; [email protected] >Subject: Re: [Discuss] SPECpower Linux vs. Windows > >* Pallipadi, Venkatesh <[EMAIL PROTECTED]> >[2008-02-26 09:45:51]: > >> >> >> >-----Original Message----- >> >From: [EMAIL PROTECTED] >> >[mailto:[EMAIL PROTECTED] On Behalf Of Martin Wilck >> >Sent: Tuesday, February 26, 2008 5:56 AM >> >To: [email protected] >> >Subject: Re: [Discuss] SPECpower Linux vs. Windows >> > >> >Pallipadi, Venkatesh wrote: >> > >> >> We have seen sched_mc_power_savings mentioned above help >SPECpower. >> >> Another _hack_ that seems to help the benchmark is to increase the >> >> ondemand sampling rate at >> >> >/sys/devices/system/cpu/cpu0/cpufreq/ondemand/sampling_rate. You may >> >> have to try different values upto max to get best number. >> > >> >Thanks. Is there a similar setting for the cpuidle governor? >> > >> >> No. There are no tunables for idle governor at present. One thing you >> can try is to limit the max_cstate at boot time, with boot option >> processor.max_cstate=1 and use only C1. How many C-states >are supported > >Hi Venki, > >Wouldn't limiting C-states further reduce the power savings? >Are you hinting that this may actually improve performance by reducing >the wakeup latency and thereby give a better SPECpower benchmark? >
Yes. I was not saying reducing the cstates will help. I was rather suggesting that you may have to try playing with it and see whether it affects the benchmark by reducing the latency. Thanks, Venki _______________________________________________ Discuss mailing list [email protected] http://mail.lesswatts.org/mailman/listinfo/discuss
