Can you describe the system are you running this test on? The reason I ask is because the firmware could be doing DVFS in the background and your assumption about always being in P0 might not be valid. Most HP servers for example ship with dynamic power management policy enabled by default in the BIOS.
And yes, by 100% I mean that the ratio is 1. Here are some ACNT and MCNT readings taken at one second interval: aperf=2651211150, mperf=2651211830 aperf=2663606070, mperf=2663641230 aperf=2658386834, mperf=2658554550 aperf=2644989682, mperf=2645542740 As you can see the ratio between these looks pretty close to 1.0. This was done on 2.6Ghz Core 2 Quad 6600. - Andrei On Wed, Jul 23, 2008 at 5:50 PM, Vinay Devadas <Vinay.Devadas at sun.com> wrote: > Hi, > > The physical processor has 2 virtual processors (0 2) > x86 (GenuineIntel 6F6 family 6 model 15 step 6 clock 2660 MHz) > Intel(r) Xeon(r) CPU 5150 @ 2.66GHz > The physical processor has 2 virtual processors (1 3) > x86 (GenuineIntel 6F6 family 6 model 15 step 6 clock 2660 MHz) > Intel(r) Xeon(r) CPU 5150 @ 2.66GHz > > What do you mean 100% ? You mean the ratio is one ? If so can you send me > your test file so that I can compare it with mine. > > Thanks, > Vinay > > ----- Original Message ----- > From: Andrei Dorofeev <dorofeev at gmail.com> > Date: Wednesday, July 23, 2008 5:23 pm > Subject: Re: [tesla-dev] Experiment Results on APERF/MPERF registers - Need > Clarification > To: Vinay Devadas <Vinay.Devadas at Sun.COM> > Cc: Eric Saxe <Eric.Saxe at Sun.COM>, tesla-dev at opensolaris.org > > >> I just tried the same test also and seeing the ratio >> between ACNT and MCNT at 100% when running at P0. >> Can you provide more details about your system and >> processor configuration? >> >> -Andrei >> >> On Wed, Jul 23, 2008 at 2:49 PM, Vinay Devadas <Vinay.Devadas at sun.com> >> wrote: >> > Hi Andrei, >> > >> >>Are you doing anything with throttling (T-states) at the same time? >> > >> > No. >> > >> > -vinay >> > >> > ----- Original Message ----- >> > From: Andrei Dorofeev <dorofeev at gmail.com> >> > Date: Wednesday, July 23, 2008 2:44 pm >> > Subject: Re: [tesla-dev] Experiment Results on APERF/MPERF registers >> - Need Clarification >> > To: Vinay Devadas <Vinay.Devadas at Sun.COM> >> > Cc: Eric Saxe <Eric.Saxe at Sun.COM>, tesla-dev at opensolaris.org >> > >> > >> >> Vinay, >> >> >> >> Are you doing anything with throttling (T-states) at the same time? >> >> >> >> - Andrei >> > >
