Tony, On Wed, May 02, 2007 at 10:18:32AM -0500, Tony Ernst wrote: > Stephane, > > I don't know whether or not it matters in this case, but I believe Michael > is also using a system with the older perfmon kernel bits. > I used RHEL4U5 for my test, so I was also running the older v2.0 perfmon code.
> > Stephane Eranian wrote: > > Michael, > > > > I did the following test on Montecito REL4U5: > > - pull libpfm and pfmon from CVS > > - compile libpfm > > - compile pfmon as static with PFMROOTDIR pointed at my libpfm dir > > > > All libpfm old examples runs for Montecito. > > > > Then for pfmon, I tried something very close to your cmdline: > > $ ./pfmon --system-wide --smpl-outfile=/tmp/sample.out.31070 > > --smpl-entries=100 -u -k --short-smpl-periods=15940000 > > --smpl-module=compact --events=CPU_OP_CYCLES_ALL --cpu-list=0 -- dd > > if=/dev/zero of=/dev/null count=10000 > > > > And it did not complain about anything. > > > > I do not have a Madison system with the same OS, so I cannot really > > reproduce your environment. Somehow I still think there is something > > wrong with the compile/runtime environment. > > > > _______________________________________________ > 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/
