That's what PAPI uses. Is the name parsing case sensitive? If so, the event name would be: "global_power_events:RUNNING" - dan
> -----Original Message----- > From: [EMAIL PROTECTED] [mailto:perfmon- > [EMAIL PROTECTED] On Behalf Of Stephane Eranian > Sent: Thursday, May 10, 2007 11:22 AM > To: Kevin Corry > Cc: [EMAIL PROTECTED]; Vince Weaver > Subject: Re: [perfmon] calculating CPI on Pentium 4 > > Vince, > > I think you need to use GLOBAL_POWER_EVNENTS:RUNNING to get the equivalent > of CPU_CLK_UNHALTED. > > On Thu, May 10, 2007 at 10:08:00AM -0500, Kevin Corry wrote: > > Hi Vince, > > > > On Thu May 10 2007 1:53 pm, Vince Weaver wrote: > > > I've been using perfmon2 to calculate CPI for various programs. This > is > > > fine on Pentium III, Itanium, and Core machines because both a CYCLES > and > > > a RETIRED_INSTRUCTIONS performance counter is available. > > > > > > However, on Pentium 4 there is no CYCLES counter. My question is, can > I > > > use the "unique timestamp" field in the results to somehow get cycle > > > count? From what I can tell this is based on the TSC, which I guess > can > > > cause problems on SMP machines if the TSCs aren't synchronized. > > > > If memory serves correctly, the "global-power-events" event on Pentium4 > is > > equivalent to clock-cycles (at least, when the processor is not > stopped). > > Give that one a try and see if the results look reasonable. > > > > Thanks, > > -- > > Kevin Corry > > [EMAIL PROTECTED] > > http://www.ibm.com/linux/ > > _______________________________________________ > > perfmon mailing list > > [email protected] > > http://www.hpl.hp.com/hosted/linux/mail-archives/perfmon/ > > -- > > -Stephane > _______________________________________________ > perfmon mailing list > [email protected] > http://www.hpl.hp.com/hosted/linux/mail-archives/perfmon/ _______________________________________________ perfmon mailing list [email protected] http://www.hpl.hp.com/hosted/linux/mail-archives/perfmon/
