From owner-freebsd-alpha Fri Mar 30 8:35:34 2001 Delivered-To: freebsd-alpha@freebsd.org Received: from feral.com (feral.com [192.67.166.1]) by hub.freebsd.org (Postfix) with ESMTP id 0B59937B71A for ; Fri, 30 Mar 2001 08:35:31 -0800 (PST) (envelope-from mjacob@feral.com) Received: from zeppo.feral.com (IDENT:mjacob@zeppo [192.67.166.71]) by feral.com (8.9.3/8.9.3) with ESMTP id IAA18145; Fri, 30 Mar 2001 08:35:24 -0800 Date: Fri, 30 Mar 2001 08:35:18 -0800 (PST) From: Matthew Jacob Reply-To: mjacob@feral.com To: Doug Rabson Cc: Andrew Gallatin , freebsd-alpha@FreeBSD.ORG Subject: Re: do we care about performance yet? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Fri, 30 Mar 2001, Doug Rabson wrote: > On Thu, 29 Mar 2001, Andrew Gallatin wrote: > > > > > FWIW, I ran a make buildworld this evening on my UP1000 under iprobe, > > as well as with the receive side of a netperf tcp stream. > > > > (for anybody who doesn't know what iprobe is, check out > > http://www.cs.duke.edu/ari/iprobe) > > > > For the buildworld, the system spent about 40% of its time in kernel. > > Roughly 50% of that was in various states of idleness. I've left full > > reports at: > > > > http://www.cs.duke.edu/~gallatin/iprobe_current/ > > > > The "???NOT_FOUND" is userland code I didn't tell iprobe about. > > I think this is just showing a lot of contention for the Giant mutex (and > probably some i/o waiting). Since we don't halt on idle, the cpu will > spend a lot of time buzzlooping on runq_check. That's what I mean about I/O bound- I looked at runq_check and yes, that's really 'idle'. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message