Date: Fri, 13 Jul 2007 20:33:43 +0100 From: Doug Rabson <dfr@rabson.org> To: freebsd-smp@freebsd.org Cc: Girish Motwani <girish.motwani@gmail.com>, Robert Watson <rwatson@freebsd.org> Subject: Re: kgmon on multiprocessors Message-ID: <200707132033.43755.dfr@rabson.org> In-Reply-To: <20070711181520.R68820@fledge.watson.org> References: <2779e6080707020843g55be2546w722b08a2023d016a@mail.gmail.com> <20070711181520.R68820@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday 11 July 2007, Robert Watson wrote: > On Mon, 2 Jul 2007, Girish Motwani wrote: > > Does the kernel profiling tool kgmon give accurate > > information on multiprocessor systems ? Or is there some other > > tool for SMP systems ? I tried looking up the archives for > > information on this, but could not find any. > > Girish-- > > Take a look at hwpmc(4), pmccontrol(8), and pmcstat(8). These allow > you to do kernel-only or systemic samplic using performance > monitoring counters, including looking at CPU time, cache misses, > instructions retired, etc. I posted a sixty second howto here: > > > http://lists.freebsd.org/pipermail/freebsd-current/2006-February/0610 >96.html > > Among other things, you can generate gprof output from PMC; right > now, that's without full stack traces (and hence callgraphs), but > Joseph Koshy has work in P4 implementing stack trace support. I > imagine this will go in after the 7.0 freeze lets up. That reminds me - does anyone know what I have to hack to make hwpmc work on my Intel(R) Core(TM)2 Quad CPU (Id = 0x6f7, Stepping = 7)?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200707132033.43755.dfr>