From owner-freebsd-current Sun Nov 16 21:52:42 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id VAA13719 for current-outgoing; Sun, 16 Nov 1997 21:52:42 -0800 (PST) (envelope-from owner-freebsd-current) Received: from nomis.Simon-Shapiro.ORG (nomis.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id VAA13654 for ; Sun, 16 Nov 1997 21:52:21 -0800 (PST) (envelope-from shimon@i-connect.net) Received: (qmail 14604 invoked by uid 1000); 17 Nov 1997 05:52:30 -0000 Message-ID: X-Mailer: XFMail 1.2-beta-111097 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <199711162043.MAA08177@rah.star-gate.com> Date: Sun, 16 Nov 1997 21:52:30 -0800 (PST) Organization: Atlas Telecom From: Simon Shapiro To: Amancio Hasty Subject: Re: make world time???/ Cc: current@FreeBSD.ORG, Steve Passe , Steve Price Sender: owner-freebsd-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi Amancio Hasty; On 16-Nov-97 you wrote: > > Cool! > > > I am curious as for the make world time on a dual PPro system with > fast disks. It looks like we are very cpu bound with fast disks. > > Tnks, > Amancio I think it bottoms (for a FULL build) at 100 minutes or SO. Disk I/O stops being an issue around there. Memeory is not a problem either. On a P6-200x2, with 256MB of RAM, a DUAL DPT 3334UW, /usr/src/ on a 5x RAID-5 array, /usr/obj on a 8x RAID-0 array, DPT with 64MB of RAM, we are just at 84 minutes. This is with -pipe, noatime, async, -j8, etc. I could not verify putting /tmp /var/tmp on mfs as the system reliably crashes. The ``disk'' I/O is at about 200-400 I/O per second - about 1/9th of the capacity of such a system, memory has over 120MB of free RAM, so we are not gaining much. What I do NOT know, and asked before (maybe I should name names :-), is is it CPU saturation, PCI saturation or RAM saturation? The system reports 80-95% CPU utilization in kernel, 70-80% in user space. I have an AMI Goliath on order. It is supposed to have tow independant PCI busses, on the Orion chipset. Let's see then. Simon