From owner-freebsd-smp Mon Jun 29 17:07:35 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA09877 for freebsd-smp-outgoing; Mon, 29 Jun 1998 17:07:35 -0700 (PDT) (envelope-from owner-freebsd-smp@FreeBSD.ORG) Received: from homer.supersex.com (homer.supersex.com [209.5.1.4]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id RAA09818 for ; Mon, 29 Jun 1998 17:07:02 -0700 (PDT) (envelope-from leo@homer.supersex.com) Received: (from leo@localhost) by homer.supersex.com (8.8.8/8.8.5) id UAA19114; Mon, 29 Jun 1998 20:07:54 -0400 (EDT) Message-ID: <19980629200753.14325@supersex.com> Date: Mon, 29 Jun 1998 20:07:53 -0400 From: Leo Papandreou To: freebsd-smp@FreeBSD.ORG Subject: Re: PPro vs PII References: <9806291802.aa03736@s3.synx.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.89i In-Reply-To: <9806291802.aa03736@s3.synx.com>; from Remy NONNENMACHER on Mon, Jun 29, 1998 at 06:14:13PM +0200 Sender: owner-freebsd-smp@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Mon, Jun 29, 1998 at 06:14:13PM +0200, Remy NONNENMACHER wrote: > On 29 Jun, Christopher G. Petrilli wrote: > > On Sun, 28 Jun 1998, Atipa wrote: > > > >> The P2 will smoke it. Better yet, go up to 350 or 400MHz, then you can > >> utilize 100MHz system bus. > > > > This is a red herring... the system bus was a big restriction back when > > the cache was running at bus speed---any increment made a huge > > difference, but with the current PII architecture, the bus speed has > > long since ceased to be a problem. How many devices do you know that > > can saturate a PCI bus constitently? Don't use this as a reasoning. > > .... > > model of execution (80-90% cache hit rate). Remember, that you can get > > PPros with 512K or 1Mb of L2 cache that is running 1:1 with the chip, > > rather than 2:1. > > > > That's what say Tom's hardware guide: 100Mhz bus give a 3 to 10% > increase. (Now, he uses Win...... benchmarks and it's difficult to > figure out what is really tested.) True. I dont know how many processes are running in the background of a typical windows session but I imagine one person is running one application which is spending nearly all of its time in the CPU cache. When you've spawned upwards of 200 httpd daemons and whatever else not in ram, I bet the "3 to 10%" improvement improves. This is precisely the case where I admin. There is also HEAVY file io spread out across literally hundreds of thousands of different image files which need to be read, stuffed into a socket and often processed to generate thumb- nails on the fly. Whatever was in the CPU cache isnt going to be there next time you blink. All of which to say, Tom's experience with Quake is not necessarily indicative of what a 100Mz bus can do. Fast wide buses are just the next best thing since goldfish. Given that newer PIIs/mobos will soon be running their cache at full speed, just like the Pro, I'd say the Pro is not worth looking at if you can swallow the difference in price for a PII. > > Let's stop the pro/cons war about P2/Pro. I propose that those > interested grab the rc5des client on www.distributed.net and post the > result of -benchmark, along with processor type and speed. This will > close the debate quickly. > Heh. A little anecdotal evidence: we upgraded from a P150 to a PII400/BX combination and believe you me, there's a difference (I just like bragging about how far I can pee :-) I can hardly wait for -CURRENT's support of 2 or more PIIs to become the next -STABLE. > RN. > > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-smp" in the body of the message To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message