Date: Thu, 23 Mar 2006 06:48:05 -0800 From: Luigi Rizzo <rizzo@icir.org> To: Scott Long <scottl@samsco.org> Cc: current@freebsd.org Subject: Re: interesting(?) data on network interrupt servicing Message-ID: <20060323064805.B67264@xorpc.icir.org> In-Reply-To: <4422B3C8.3080303@samsco.org>; from scottl@samsco.org on Thu, Mar 23, 2006 at 07:42:16AM -0700 References: <20060322122906.A41691@xorpc.icir.org> <20060323001555.GA1811@tin.it> <20060323142518.GA1308@tin.it> <20060323063139.A67037@xorpc.icir.org> <4422B3C8.3080303@samsco.org>
index | next in thread | previous in thread | raw e-mail
On Thu, Mar 23, 2006 at 07:42:16AM -0700, Scott Long wrote: > Luigi Rizzo wrote: > > On Thu, Mar 23, 2006 at 03:25:18PM +0100, Paolo Pisati wrote: ... > >>ok, i updated my CURRENT and rerun the tests (and while here > >>i disabled SMP): > >> > >>phk's optimization to cpu ticks calculation shaved 4k ticks, > > > > this makes it a very good candidate for MFC when 6.1 is out ? ... > I haven't been paying close enough attention, have all of the calcru > problems and other side effects been fixed from phk's work? we should ask phk. As far as i remember the only "problem" is/was that the sys/user times are computed as if the cpu were running at its max speed. But this is in fact a good thing because it is a more consistent measurement of the cost of the CPU work, which decouples us from having to take care of variable cpu speed. luigihome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060323064805.B67264>