From owner-freebsd-current Tue Dec 9 23:22:24 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id XAA24217 for current-outgoing; Tue, 9 Dec 1997 23:22:24 -0800 (PST) (envelope-from owner-freebsd-current) Received: from paladio.canonware.com (canonware.com [206.184.206.112]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id XAA24207 for ; Tue, 9 Dec 1997 23:22:18 -0800 (PST) (envelope-from jasone@canonware.com) Received: from localhost (jasone@localhost) by paladio.canonware.com (8.8.5/8.8.5) with SMTP id XAA15132; Tue, 9 Dec 1997 23:22:20 -0800 Date: Tue, 9 Dec 1997 23:22:20 -0800 (PST) From: Jason Evans X-Sender: jasone@paladio To: Alex cc: Steve Passe , current Subject: Re: -current, goliath, smp -- problems In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 8 Dec 1997, Alex wrote: > Speaking of more CPUs, did anyone catch the release of HP's newest SMP > server? An 8 CPU (I'm assuming PPro) *drool*. I'm assuming it comes with > gobs of ram and disk space, and it amusingly enough looked like an > overgrown dot-matrix printer with no paper. I didn't realize there were > production machines with more than 4 Intel CPUs *drool*. Well, I don't know that I would necessarily expect this to be as cool as it sounds. I'm not an expert on this by any means, but I've repeatedly heard the following: Tightly coupled MP machines work well until memory bandwidth becomes a bottleneck. Last I heard, the memory bandwidth on Intel-based machines is in the neighborhood of 500 MB/sec, which is enough to feed 4 processors, but quite inadequate for 8. So, even though there are twice as many processors, the performance gain is not impressive. If there's anyone else out there that knows any better, please correct me. =) Jason Jason Evans Email: [jasone@canonware.com] Home phone: [(650) 856-8204] Quote: ["Invention is 1% inspiration, 99% perspiration" - Thomas Edison]