From owner-freebsd-smp Sun Nov 9 06:09:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id GAA12501 for smp-outgoing; Sun, 9 Nov 1997 06:09:15 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from mail.webspan.net (root@mail.webspan.net [206.154.70.7]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id GAA12493 for ; Sun, 9 Nov 1997 06:09:12 -0800 (PST) (envelope-from gpalmer@orion.webspan.net) Received: from orion.webspan.net (orion.webspan.net [206.154.70.5]) by mail.webspan.net (WEBSPAN/970608) with ESMTP id JAA13035 for ; Sun, 9 Nov 1997 09:09:11 -0500 (EST) Received: from orion.webspan.net (localhost [127.0.0.1]) by orion.webspan.net (WEBSPAN/970608) with ESMTP id JAA09293 for ; Sun, 9 Nov 1997 09:09:10 -0500 (EST) To: freebsd-smp@freebsd.org From: "Gary Palmer" Subject: Best processor? Date: Sun, 09 Nov 1997 09:09:10 -0500 Message-ID: <9268.879084550@orion.webspan.net> Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Which is the best processor for a SMP system? PPro or PII? Also, as an aside, looking at the Tyan Titan Pro board, I see you need `VRM's, but none of the online catalogues have such things. Do they come with the board? Thanks, Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info From owner-freebsd-smp Sun Nov 9 06:57:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id GAA14807 for smp-outgoing; Sun, 9 Nov 1997 06:57:29 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id GAA14798; Sun, 9 Nov 1997 06:57:21 -0800 (PST) (envelope-from vanmaren@fast.cs.utah.edu) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id HAA07223; Sun, 9 Nov 1997 07:57:18 -0700 (MST) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id HAA20600; Sun, 9 Nov 1997 07:57:18 -0700 Date: Sun, 9 Nov 1997 07:57:18 -0700 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199711091457.HAA20600@fast.cs.utah.edu> To: freebsd-smp@freebsd.org, gpalmer@freebsd.org Subject: Re: Best processor? Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk The best processor (currently) is the Pentium Pro/200 with 1MB of L2 cache. However, it is $2700 each in large quantities. It is the fastest in a 4-processor config. For a `budget; dual, probably the Pentium II/300. The chipset advantages and single-processor performance probably outweigh the slower L2 cache, although the Pro/200 w/512k is still a good choice. (The II/300 is cheaper). The Tyan board should come with dual VRMs. Ours did. Kevin From owner-freebsd-smp Sun Nov 9 07:39:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id HAA16508 for smp-outgoing; Sun, 9 Nov 1997 07:39:25 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from picnic.mat.net (picnic.mat.net [206.246.122.117]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id HAA16497; Sun, 9 Nov 1997 07:39:11 -0800 (PST) (envelope-from chuckr@glue.umd.edu) Received: from localhost (chuckr@localhost) by picnic.mat.net (8.8.7/8.8.5) with SMTP id JAA26831; Sun, 9 Nov 1997 09:36:21 -0500 (EST) X-Authentication-Warning: picnic.mat.net: chuckr owned process doing -bs Date: Sun, 9 Nov 1997 09:36:19 -0500 (EST) From: Chuck Robey X-Sender: chuckr@localhost To: Gary Palmer cc: freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: <9268.879084550@orion.webspan.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 9 Nov 1997, Gary Palmer wrote: > > Which is the best processor for a SMP system? PPro or PII? > > Also, as an aside, looking at the Tyan Titan Pro board, I see you need > `VRM's, but none of the online catalogues have such things. Do they > come with the board? They did when I bought one, one Voltage Regulator Module per processor installed. Little boards with huge toroids on them. > > Thanks, > > Gary > -- > Gary Palmer FreeBSD Core Team Member > FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@glue.umd.edu | communications topic, C programming, and Unix. 213 Lakeside Drive Apt T-1 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+----------------------------------------------- From owner-freebsd-smp Sun Nov 9 08:40:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id IAA21503 for smp-outgoing; Sun, 9 Nov 1997 08:40:15 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from server.local.sunyit.edu (A-T34.rh.sunyit.edu [150.156.210.241]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id IAA21495; Sun, 9 Nov 1997 08:40:07 -0800 (PST) (envelope-from perlsta@cs.sunyit.edu) Received: from localhost (perlsta@localhost) by server.local.sunyit.edu (8.8.7/8.8.5) with SMTP id MAA18772; Sun, 9 Nov 1997 12:44:55 -0500 (EST) X-Authentication-Warning: server.local.sunyit.edu: perlsta owned process doing -bs Date: Sun, 9 Nov 1997 12:44:55 -0500 (EST) From: Alfred Perlstein X-Sender: perlsta@server.local.sunyit.edu To: Gary Palmer cc: freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: <9268.879084550@orion.webspan.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk dual 300mhz PIIs will beat dual PPro 200mhz, but i think i't almost better to get a dual PPro 200mhz+512k instead of dual PII 233mhz. -Alfred On Sun, 9 Nov 1997, Gary Palmer wrote: > > Which is the best processor for a SMP system? PPro or PII? > > Also, as an aside, looking at the Tyan Titan Pro board, I see you need > `VRM's, but none of the online catalogues have such things. Do they > come with the board? > > Thanks, > > Gary > -- > Gary Palmer FreeBSD Core Team Member > FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info > From owner-freebsd-smp Sun Nov 9 09:12:07 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id JAA23023 for smp-outgoing; Sun, 9 Nov 1997 09:12:07 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from picnic.mat.net (picnic.mat.net [206.246.122.117]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id JAA22994; Sun, 9 Nov 1997 09:11:56 -0800 (PST) (envelope-from chuckr@glue.umd.edu) Received: from localhost (chuckr@localhost) by picnic.mat.net (8.8.7/8.8.5) with SMTP id LAA27031; Sun, 9 Nov 1997 11:08:47 -0500 (EST) X-Authentication-Warning: picnic.mat.net: chuckr owned process doing -bs Date: Sun, 9 Nov 1997 11:08:46 -0500 (EST) From: Chuck Robey X-Sender: chuckr@localhost To: Alfred Perlstein cc: Gary Palmer , freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 9 Nov 1997, Alfred Perlstein wrote: > dual 300mhz PIIs will beat dual PPro 200mhz, but i think i't almost better > to get a dual PPro 200mhz+512k instead of dual PII 233mhz. Where did you get that information? Because the difference in cache (not the size, the architecture) I'd have suspected that the Ppro would be faster in both cases. > > -Alfred > > On Sun, 9 Nov 1997, Gary Palmer wrote: > > > > > Which is the best processor for a SMP system? PPro or PII? > > > > Also, as an aside, looking at the Tyan Titan Pro board, I see you need > > `VRM's, but none of the online catalogues have such things. Do they > > come with the board? > > > > Thanks, > > > > Gary > > -- > > Gary Palmer FreeBSD Core Team Member > > FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info > > > > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@glue.umd.edu | communications topic, C programming, and Unix. 213 Lakeside Drive Apt T-1 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+----------------------------------------------- From owner-freebsd-smp Sun Nov 9 09:44:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id JAA24463 for smp-outgoing; Sun, 9 Nov 1997 09:44:03 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from onyx.atipa.com (user8713@ns.atipa.com [208.128.22.10]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id JAA24444 for ; Sun, 9 Nov 1997 09:43:58 -0800 (PST) (envelope-from freebsd@atipa.com) Received: (qmail-queue invoked by uid 1018); 9 Nov 1997 17:49:33 -0000 Date: Sun, 9 Nov 1997 10:49:32 -0700 (MST) From: Atipa X-Sender: freebsd@dot.ishiboo.com To: Gary Palmer cc: freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: <9268.879084550@orion.webspan.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 9 Nov 1997, Gary Palmer wrote: > > Which is the best processor for a SMP system? PPro or PII? > > Also, as an aside, looking at the Tyan Titan Pro board, I see you need > `VRM's, but none of the online catalogues have such things. Do they > come with the board? > In a nutshell: 1) PPro is faster than P-II (at THE SAME CLOCK RATE) for native 32-bit code --- BUT --- 2) PII is faster IN GENERAL after you consider the higher clock rate (233 is faster that Pro-200; 266 and 300 kick its ass) 3) P-Pro is currently "unsupported" by Intel. They are not making any other boards, BIOSes, etc., etc. They do not want this line to continue due to high production costs. P-II pricing is going down, but PPro pricing is actually raising (poor supply in the market). 4) Pentium II is more expandible (up to 500+ MHz) 5) Pentium II uses MMX extensions 6) Pentium II chipsets support AGP, UDMA, SDRAM, etc... 7) Pentium II uses Dual Independent Bus (DIB) and has twice the on-chip L1 cache (although L2 is running at 1/2 speed) 8) It has now been "tried and true" That's my personal opinion. I know that Terry and some other's wont agree :) Kevin From owner-freebsd-smp Sun Nov 9 13:32:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id NAA06108 for smp-outgoing; Sun, 9 Nov 1997 13:32:09 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from ns.mt.sri.com (SRI-56K-FR.mt.net [206.127.65.42]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id NAA06052; Sun, 9 Nov 1997 13:31:58 -0800 (PST) (envelope-from nate@rocky.mt.sri.com) Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100]) by ns.mt.sri.com (8.8.7/8.8.7) with ESMTP id OAA11187; Sun, 9 Nov 1997 14:31:49 -0700 (MST) Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id OAA05995; Sun, 9 Nov 1997 14:31:45 -0700 (MST) Date: Sun, 9 Nov 1997 14:31:45 -0700 (MST) Message-Id: <199711092131.OAA05995@rocky.mt.sri.com> From: Nate Williams MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: Alfred Perlstein Cc: Gary Palmer , freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: References: <9268.879084550@orion.webspan.net> X-Mailer: VM 6.29 under 19.15 XEmacs Lucid Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > dual 300mhz PIIs will beat dual PPro 200mhz >From the noise I've been hearing lately on the mailing lists, this suprises me. Do you have #'s to back it up? Nate From owner-freebsd-smp Sun Nov 9 13:59:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id NAA07961 for smp-outgoing; Sun, 9 Nov 1997 13:59:48 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id NAA07937; Sun, 9 Nov 1997 13:59:30 -0800 (PST) (envelope-from toor@dyson.iquest.net) Received: (from root@localhost) by dyson.iquest.net (8.8.7/8.8.5) id QAA27125; Sun, 9 Nov 1997 16:59:16 -0500 (EST) From: "John S. Dyson" Message-Id: <199711092159.QAA27125@dyson.iquest.net> Subject: Re: Best processor? In-Reply-To: <199711092131.OAA05995@rocky.mt.sri.com> from Nate Williams at "Nov 9, 97 02:31:45 pm" To: nate@mt.sri.com (Nate Williams) Date: Sun, 9 Nov 1997 16:59:16 -0500 (EST) Cc: perlsta@cs.sunyit.edu, gpalmer@freebsd.org, freebsd-smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Nate Williams said: > > dual 300mhz PIIs will beat dual PPro 200mhz > > >From the noise I've been hearing lately on the mailing lists, this > suprises me. Do you have #'s to back it up? > I am not responding with numbers, but if you look at it, it is likely true: 1) The PIIs have 512K cache, while the PPro has (normally) 256K cache. Therefore bus utilization is likely less with the PII. Even in the case of a 512K cache, the bus utilization is going to be nearly the same. In a DP system, bus utilization is likely less important than in 4-way systems anyway. 2) Expect about 3-5% miss rate with an 8K or 16K 1st level cache. (I have really measured it on real applications.) Miss rate can be much lower than that though. The miss rate does not scale linearly downward with 1st level cache size, but it does go down (especially with n-way associative cache schemes.) 3) Single processor PIIs at 300MHz are almost always (if not always) faster than a PPro at 200MHz running real code. PII MB's can support SDRAM now, and that really does help mitigate the aggregate performance loss due to the 1/2 speed 2nd level cache. If you are talking about 233MHz PII processors vs. 200MHz PPro processors, it is harder to decide on which processor is faster, but I do think that the PII will win out on average. Clock your PPro at 233MHz, and the PPro will win out, except for MMX type apps, or on memory intensive apps, where the LX chipset and SDRAM memory speeds help. The huge gotcha in all of this is that the PII only caches the first 512MB of memory. -- John dyson@freebsd.org jdyson@nc.com From owner-freebsd-smp Sun Nov 9 14:14:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id OAA08998 for smp-outgoing; Sun, 9 Nov 1997 14:14:36 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from picnic.mat.net (picnic.mat.net [206.246.122.117]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id OAA08976; Sun, 9 Nov 1997 14:14:21 -0800 (PST) (envelope-from chuckr@glue.umd.edu) Received: from localhost (chuckr@localhost) by picnic.mat.net (8.8.7/8.8.5) with SMTP id QAA28534; Sun, 9 Nov 1997 16:10:11 -0500 (EST) X-Authentication-Warning: picnic.mat.net: chuckr owned process doing -bs Date: Sun, 9 Nov 1997 16:10:10 -0500 (EST) From: Chuck Robey X-Sender: chuckr@picnic.mat.net To: "John S. Dyson" cc: Nate Williams , perlsta@cs.sunyit.edu, gpalmer@freebsd.org, freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: <199711092159.QAA27125@dyson.iquest.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 9 Nov 1997, John S. Dyson wrote: > If you are talking about 233MHz PII processors vs. 200MHz PPro processors, it > is harder to decide on which processor is faster, but I do think that the PII > will win out on average. Clock your PPro at 233MHz, and the PPro will win out, > except for MMX type apps, or on memory intensive apps, where the LX chipset and > SDRAM memory speeds help. I can't remember where I read it (because I read a protected mode list also, and game producers comment on this a lot) but I'd read where the MMX instructions are not proving to be any real help there. Exactly how much L1 cache is on the PPro 512K L2 chips vs. the PII 512K L2 chips, do you know? > > The huge gotcha in all of this is that the PII only caches the first 512MB of > memory. > > > -- > John > dyson@freebsd.org > jdyson@nc.com > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@glue.umd.edu | communications topic, C programming, and Unix. 213 Lakeside Drive Apt T-1 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+----------------------------------------------- From owner-freebsd-smp Sun Nov 9 14:26:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id OAA09784 for smp-outgoing; Sun, 9 Nov 1997 14:26:15 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id OAA09775; Sun, 9 Nov 1997 14:26:08 -0800 (PST) (envelope-from toor@dyson.iquest.net) Received: (from root@localhost) by dyson.iquest.net (8.8.7/8.8.5) id RAA27206; Sun, 9 Nov 1997 17:25:55 -0500 (EST) From: "John S. Dyson" Message-Id: <199711092225.RAA27206@dyson.iquest.net> Subject: Re: Best processor? In-Reply-To: from Chuck Robey at "Nov 9, 97 04:10:10 pm" To: chuckr@Glue.umd.edu (Chuck Robey) Date: Sun, 9 Nov 1997 17:25:54 -0500 (EST) Cc: toor@dyson.iquest.net, nate@mt.sri.com, perlsta@cs.sunyit.edu, gpalmer@freebsd.org, freebsd-smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Chuck Robey said: > On Sun, 9 Nov 1997, John S. Dyson wrote: > > > If you are talking about 233MHz PII processors vs. 200MHz PPro processors, it > > is harder to decide on which processor is faster, but I do think that the PII > > will win out on average. Clock your PPro at 233MHz, and the PPro will win out, > > except for MMX type apps, or on memory intensive apps, where the LX chipset and > > SDRAM memory speeds help. > > I can't remember where I read it (because I read a protected mode list > also, and game producers comment on this a lot) but I'd read where the MMX > instructions are not proving to be any real help there. > Well, with MMX you do have to seriously recode your apps. The biggest problem with MMX that I see, is that it doesn't appear to be of much help with 32Bit DSP ops. When working in 16Bit land for certain apps, one has to be much more careful. We (FreeBSD) don't support MMX yet, other than it will probably work if you want to try it. > > Exactly how much L1 cache is on the PPro 512K L2 chips vs. the PII 512K L2 > chips, do you know? > The PII has 16K+16K as opposed to PPro 8K+8K. I seem to remember that the associativity might be better on the PII also. Since that is less in the area of diminishing returns (unlike the upgrade from 256K to 512K in a single cpu system, which mostly helps just a little), I would expect that the larger L1 makes a substantial difference, especially since the PII has slower L2 access, and the cost of a miss is higher. Note that the PPro has only 2 pages of L1 data cache, while the PII has 4 pages. That has to help the multi user system performance when all new pages have to be zeroed, since one is a little less likely to fully erase the valuable cache. (Just a guess.) The 16K+16K must really help when the benchmark doesn't fit in the 8K+8K though :-). -- John dyson@freebsd.org jdyson@nc.com From owner-freebsd-smp Sun Nov 9 14:45:58 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id OAA11043 for smp-outgoing; Sun, 9 Nov 1997 14:45:58 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from icicle.winternet.com (adm@icicle.winternet.com [198.174.169.13]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id OAA11006; Sun, 9 Nov 1997 14:45:42 -0800 (PST) (envelope-from mestery@mail.winternet.com) Received: (from adm@localhost) by icicle.winternet.com (8.8.8/8.8.8) id QAA26647; Sun, 9 Nov 1997 16:45:30 -0600 (CST) Received: from tundra.winternet.com(198.174.169.11) by icicle.winternet.com via smap (V2.0) id xma026552; Sun, 9 Nov 97 16:44:53 -0600 Received: from localhost (mestery@localhost) by tundra.winternet.com (8.8.7/8.8.4) with SMTP id QAA27614; Sun, 9 Nov 1997 16:44:52 -0600 (CST) X-Authentication-Warning: tundra.winternet.com: mestery owned process doing -bs Date: Sun, 9 Nov 1997 16:44:52 -0600 (CST) From: Kyle Mestery To: Atipa cc: Gary Palmer , freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 9 Nov 1997, Atipa wrote: 1) PII can only scale to 2 CPUs. 2) PII can only cache 512MB RAM Kyle Mestery StorageTek's Network Systems Group 7600 Boone Ave. N., Minneapolis, MN 55428 mesteka@anubis.network.com, mestery@winternet.com "You do not greet Death, you punch him in the throat repeatedly until he drags you away." --No Fear > In a nutshell: > > 1) PPro is faster than P-II (at THE SAME CLOCK RATE) for native > 32-bit code > > --- BUT --- > > 2) PII is faster IN GENERAL after you consider the higher clock rate > (233 is faster that Pro-200; 266 and 300 kick its ass) > > 3) P-Pro is currently "unsupported" by Intel. They are not making any > other boards, BIOSes, etc., etc. They do not want this line > to continue due to high production costs. P-II pricing is going > down, but PPro pricing is actually raising (poor supply in > the market). > > 4) Pentium II is more expandible (up to 500+ MHz) > > 5) Pentium II uses MMX extensions > > 6) Pentium II chipsets support AGP, UDMA, SDRAM, etc... > > 7) Pentium II uses Dual Independent Bus (DIB) and has twice the > on-chip L1 cache (although L2 is running at 1/2 speed) > > 8) It has now been "tried and true" > > > That's my personal opinion. I know that Terry and some other's wont agree > :) > > Kevin > From owner-freebsd-smp Sun Nov 9 15:40:13 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id PAA14949 for smp-outgoing; Sun, 9 Nov 1997 15:40:13 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from ns.mt.sri.com (SRI-56K-FR.mt.net [206.127.65.42]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id PAA14930; Sun, 9 Nov 1997 15:40:04 -0800 (PST) (envelope-from nate@rocky.mt.sri.com) Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100]) by ns.mt.sri.com (8.8.7/8.8.7) with ESMTP id QAA12001; Sun, 9 Nov 1997 16:39:55 -0700 (MST) Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id QAA06530; Sun, 9 Nov 1997 16:39:53 -0700 (MST) Date: Sun, 9 Nov 1997 16:39:53 -0700 (MST) Message-Id: <199711092339.QAA06530@rocky.mt.sri.com> From: Nate Williams MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: "John S. Dyson" Cc: nate@mt.sri.com (Nate Williams), perlsta@cs.sunyit.edu, gpalmer@freebsd.org, freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: <199711092159.QAA27125@dyson.iquest.net> References: <199711092131.OAA05995@rocky.mt.sri.com> <199711092159.QAA27125@dyson.iquest.net> X-Mailer: VM 6.29 under 19.15 XEmacs Lucid Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > > dual 300mhz PIIs will beat dual PPro 200mhz > > > > >From the noise I've been hearing lately on the mailing lists, this > > suprises me. Do you have #'s to back it up? > > > I am not responding with numbers, but if you look at it, it is likely true: > > 1) The PIIs have 512K cache, while the PPro has (normally) 256K cache. The big cache runs at half-speed though, which is a *huge* performance hit. (They have a bigger L1 cache, which is a win, more on that below.) > Therefore bus utilization is likely less with the PII. Even in the > case of a 512K cache, the bus utilization is going to be nearly the > same. Not quite. The PII has to 'spin' alot more waiting for data since it can't get to it at bus-speeds, while the PPro doesn't have to. Going from 256 -> 512K doesn't equal a double in cache performance (I'd suspect somewhere around 15-20% at best), so I would think the two #'s would be close to break-even. If you get a 512K PPro it would be a big win. > In a DP system, bus utilization is likely less important than > in 4-way systems anyway. DP? Distributed Processing? SMP? Help me out here. > 2) Expect about 3-5% miss rate with an 8K or 16K 1st level cache. (I > have really measured it on real applications.) Heck, let's use the #'s from Hennessy and Patterson, considered to be 'THE' hardware/cache reference in many folks minds. (The processor in this case is one of the later VAX sets, but it's architecture is similar enough for cache performance to be pretty close). Cache/size vs. miss rate: 8K: 8 - 16% 16K: 7 - 11% 32K: 2 - 6% 64K: 1 - 3% > Miss rate can be much lower than that though. The miss rate does not > scale linearly downward with 1st level cache size, but it does go down > (especially with n-way associative cache schemes.) This is for the L1 cache numbers, and the numbers given assume a data + instruction combined cache. > 3) Single processor PIIs at 300MHz are almost always (if not always) > faster than a PPro at 200MHz running real code. PII MB's can support > SDRAM now, and that really does help mitigate the aggregate > performance loss due to the 1/2 speed 2nd level cache. I'd like to see real #'s to back that up. > If you are talking about 233MHz PII processors vs. 200MHz PPro processors, it > is harder to decide on which processor is faster, but I do think that the PII > will win out on average. We're talking about SMP support, not UP support. For UP stuff, there's no doubt that the high-clock PII chips will outperform a (relatively speaking) low-clock PPro chip, but for SMP everything I've read and seen tells me that the PPro *kills* the PII for SMP work, mostly due to the L2 cache (and motherboard design??) Nate From owner-freebsd-smp Sun Nov 9 15:42:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id PAA15107 for smp-outgoing; Sun, 9 Nov 1997 15:42:23 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from ns.mt.sri.com (SRI-56K-FR.mt.net [206.127.65.42]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id PAA15087; Sun, 9 Nov 1997 15:42:15 -0800 (PST) (envelope-from nate@rocky.mt.sri.com) Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100]) by ns.mt.sri.com (8.8.7/8.8.7) with ESMTP id QAA12032; Sun, 9 Nov 1997 16:42:03 -0700 (MST) Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id QAA06582; Sun, 9 Nov 1997 16:42:01 -0700 (MST) Date: Sun, 9 Nov 1997 16:42:01 -0700 (MST) Message-Id: <199711092342.QAA06582@rocky.mt.sri.com> From: Nate Williams MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: Chuck Robey Cc: "John S. Dyson" , Nate Williams , gpalmer@freebsd.org, freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: References: <199711092159.QAA27125@dyson.iquest.net> X-Mailer: VM 6.29 under 19.15 XEmacs Lucid Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Chuck Robey writes: > I can't remember where I read it (because I read a protected mode list > also, and game producers comment on this a lot) but I'd read where the MMX > instructions are not proving to be any real help there. I could see this easily. Many 'high-end' programs are attempting to take advantage of the FPU for some of the work, and MMX makes using the FPU too expensive, so they trade-off MMX instructions for FPU instructions, so in the end it's a wash. Nate From owner-freebsd-smp Sun Nov 9 16:00:22 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA16202 for smp-outgoing; Sun, 9 Nov 1997 16:00:22 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id QAA16170; Sun, 9 Nov 1997 16:00:07 -0800 (PST) (envelope-from toor@dyson.iquest.net) Received: (from root@localhost) by dyson.iquest.net (8.8.7/8.8.5) id SAA27521; Sun, 9 Nov 1997 18:59:42 -0500 (EST) From: "John S. Dyson" Message-Id: <199711092359.SAA27521@dyson.iquest.net> Subject: Re: Best processor? In-Reply-To: <199711092339.QAA06530@rocky.mt.sri.com> from Nate Williams at "Nov 9, 97 04:39:53 pm" To: nate@mt.sri.com (Nate Williams) Date: Sun, 9 Nov 1997 18:59:42 -0500 (EST) Cc: toor@dyson.iquest.net, nate@mt.sri.com, perlsta@cs.sunyit.edu, gpalmer@freebsd.org, freebsd-smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Nate Williams said: > > > > dual 300mhz PIIs will beat dual PPro 200mhz > > > > > > >From the noise I've been hearing lately on the mailing lists, this > > > suprises me. Do you have #'s to back it up? > > > > > > I am not responding with numbers, but if you look at it, it is likely true: > > > > 1) The PIIs have 512K cache, while the PPro has (normally) 256K cache. > > The big cache runs at half-speed though, which is a *huge* performance > hit. (They have a bigger L1 cache, which is a win, more on that below.) > > > Therefore bus utilization is likely less with the PII. Even in the > > case of a 512K cache, the bus utilization is going to be nearly the > > same. > > Not quite. The PII has to 'spin' alot more waiting for data since it > can't get to it at bus-speeds, while the PPro doesn't have to. Going > from 256 -> 512K doesn't equal a double in cache performance (I'd > suspect somewhere around 15-20% at best), so I would think the two #'s > would be close to break-even. If you get a 512K PPro it would be a big > win. > Bus utilization doesn't have as much to do with the processor as what the processor appears to be to the memory subsystem. A 512K PPro should have a bus utilization similar to a 512K PII. Sure, the traffic between the processor and 2nd level cache will be slower (due to the 1/2 speed) and different (due to the double sized 1st level cache.) That isn't what I said though. > > > In a DP system, bus utilization is likely less important than > > in 4-way systems anyway. > > DP? Distributed Processing? SMP? Help me out here. > Dual processor -- take a look at the comparison with a 4-way system. > > > 2) Expect about 3-5% miss rate with an 8K or 16K 1st level cache. (I > > have really measured it on real applications.) > > Heck, let's use the #'s from Hennessy and Patterson, considered to be > 'THE' hardware/cache reference in many folks minds. (The processor in > this case is one of the later VAX sets, but it's architecture is similar > enough for cache performance to be pretty close). > Sorry, but I measured it running real programs, like gcc, etc. Note that I seldom saw an 11% L1 miss rate (of course, you can make it miss using synthetic benchmarks, but that is not what I am talking about.) Try running some tests with the P6 performance counters. It doesn't measure the miss rates, etc directly, but with a few documented calculations, you can get them. I am going to be out of town for the rest of the week, but I have given others a copy of the code. > Cache/size vs. miss rate: > > 8K: 8 - 16% > 16K: 7 - 11% > 32K: 2 - 6% > 64K: 1 - 3% > Interesting numbers. They are too high for the workloads that I have measured though. What was the line size on those caches? Maybe I'll finally have to buy a copy of H&P to see what they are talking about. > > > Miss rate can be much lower than that though. The miss rate does not > > scale linearly downward with 1st level cache size, but it does go down > > (especially with n-way associative cache schemes.) > > This is for the L1 cache numbers, and the numbers given assume a data + > instruction combined cache. > A seperate cache for data + instruction has both advantages and disadvantages. > > I'd like to see real #'s to back that up. > Well, someone just posted a benchmark that showed that at least on PII was faster than my PPro (I think that it was the semspeed benchmark.) The results and experiment were not controlled, but it was within the range that I would expect. > > > If you are talking about 233MHz PII processors vs. 200MHz PPro processors, it > > is harder to decide on which processor is faster, but I do think that the PII > > will win out on average. > > We're talking about SMP support, not UP support. For UP stuff, there's > no doubt that the high-clock PII chips will outperform a (relatively > speaking) low-clock PPro chip, but for SMP everything I've read and seen > tells me that the PPro *kills* the PII for SMP work, mostly due to the > L2 cache (and motherboard design??) > Even with SDRAM and an LX chipset? A 512K PII shouldn't look that different to a bus subsystem from a 512K P6, unless someone made a very bad mistake. Maybe they broke the MESI protocol? Now, the only thing that I can believe to be fact at this point: a 4-WAY P6 beats all PII configs. a P6 (per MHz) is mostly faster than a PII. a P6 system is likely faster than a PII system, if you need more than 512MB. a P6-233 is seldom slower than a PII-233. I don't think that any claims can be made that in general: a dual P6-200 Natoma sys is faster than a dual PII-233 LX system. It would be nice to see some numbers that say that a 2XP6-200 Natoma is faster in the same hardware configuration (incl diskdrives/memory) than a 2XPII-233 LX. I don't think that it will be so though. A good, fast test would be a recompile of GCC from the FSF sources. (That is faster than worldstone.) If we find out that a dual PII/233 or PII/266 is slower than a dual P6-512K, that would be VERY INTERESTING!!! Anyone willing to take up the challenge? -- John dyson@freebsd.org jdyson@nc.com From owner-freebsd-smp Sun Nov 9 16:01:39 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA16315 for smp-outgoing; Sun, 9 Nov 1997 16:01:39 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id QAA16310; Sun, 9 Nov 1997 16:01:32 -0800 (PST) (envelope-from toor@dyson.iquest.net) Received: (from root@localhost) by dyson.iquest.net (8.8.7/8.8.5) id TAA27531; Sun, 9 Nov 1997 19:00:22 -0500 (EST) From: "John S. Dyson" Message-Id: <199711100000.TAA27531@dyson.iquest.net> Subject: Re: Best processor? In-Reply-To: <199711092342.QAA06582@rocky.mt.sri.com> from Nate Williams at "Nov 9, 97 04:42:01 pm" To: nate@mt.sri.com (Nate Williams) Date: Sun, 9 Nov 1997 19:00:22 -0500 (EST) Cc: chuckr@Glue.umd.edu, toor@dyson.iquest.net, nate@mt.sri.com, gpalmer@freebsd.org, freebsd-smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Nate Williams said: > Chuck Robey writes: > > > I can't remember where I read it (because I read a protected mode list > > also, and game producers comment on this a lot) but I'd read where the MMX > > instructions are not proving to be any real help there. > > I could see this easily. Many 'high-end' programs are attempting to > take advantage of the FPU for some of the work, and MMX makes using the > FPU too expensive, so they trade-off MMX instructions for FPU > instructions, so in the end it's a wash. > Some of the MMX clones appear to have very fast MMX/FPU switching. I wonder what difference that means. -- John dyson@freebsd.org jdyson@nc.com From owner-freebsd-smp Sun Nov 9 16:10:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA16822 for smp-outgoing; Sun, 9 Nov 1997 16:10:17 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from ns.mt.sri.com (SRI-56K-FR.mt.net [206.127.65.42]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id QAA16814 for ; Sun, 9 Nov 1997 16:10:09 -0800 (PST) (envelope-from nate@rocky.mt.sri.com) Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100]) by ns.mt.sri.com (8.8.7/8.8.7) with ESMTP id RAA12238; Sun, 9 Nov 1997 17:10:07 -0700 (MST) Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id RAA06734; Sun, 9 Nov 1997 17:10:05 -0700 (MST) Date: Sun, 9 Nov 1997 17:10:05 -0700 (MST) Message-Id: <199711100010.RAA06734@rocky.mt.sri.com> From: Nate Williams MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: "John S. Dyson" Cc: nate@mt.sri.com (Nate Williams), freebsd-smp@freebsd.org Subject: Re: Best processor? In-Reply-To: <199711092359.SAA27521@dyson.iquest.net> References: <199711092339.QAA06530@rocky.mt.sri.com> <199711092359.SAA27521@dyson.iquest.net> X-Mailer: VM 6.29 under 19.15 XEmacs Lucid Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > > Therefore bus utilization is likely less with the PII. Even in the > > > case of a 512K cache, the bus utilization is going to be nearly the > > > same. > > > > Not quite. The PII has to 'spin' alot more waiting for data since it > > can't get to it at bus-speeds, while the PPro doesn't have to. Going > > from 256 -> 512K doesn't equal a double in cache performance (I'd > > suspect somewhere around 15-20% at best), so I would think the two #'s > > would be close to break-even. If you get a 512K PPro it would be a big > > win. > > Bus utilization doesn't have as much to do with the processor as what > the processor appears to be to the memory subsystem. A 512K PPro should > have a bus utilization similar to a 512K PII. Sure, the traffic between > the processor and 2nd level cache will be slower (due to the 1/2 speed) > and different (due to the double sized 1st level cache.) That isn't what > I said though. True. But, the 'speed' of the system is only partially related to the bus utilization. > > > 2) Expect about 3-5% miss rate with an 8K or 16K 1st level cache. (I > > > have really measured it on real applications.) > > > > Heck, let's use the #'s from Hennessy and Patterson... > > Sorry, but I measured it running real programs, like gcc, etc. Note that > I seldom saw an 11% L1 miss rate (of course, you can make it miss using > synthetic benchmarks, but that is not what I am talking about.) My version of the book is pretty old, and much has been done since then, but the principles are still valid. .. > measured though. What was the line size on those caches? Maybe I'll > finally have to buy a copy of H&P to see what they are talking about. Highly recommended. > > I'd like to see real #'s to back that up. > > Well, someone just posted a benchmark that showed that at least on PII was > faster than my PPro (I think that it was the semspeed benchmark.) That was a 266Mhz PII in UP setup, wasn't it? I expect it to beat it, but in DP (or more), no-one has made any benchmarks. > Now, the only thing that I can believe to be fact at this point: > > a 4-WAY P6 beats all PII configs. > a P6 (per MHz) is mostly faster than a PII. > a P6 system is likely faster than a PII system, if you need > more than 512MB. > a P6-233 is seldom slower than a PII-233. > > I don't think that any claims can be made that in general: > > a dual P6-200 Natoma sys is faster than a dual PII-233 LX system. I remember seeing someone claiming that that was indeed the case, and they had #'s to back it up. And not just a bit faster, *Significantly* faster, so much that if I remember right I was thinking that a 2*PPro@200 would *still* be faster than a 2*PII-266, and possibly even at 300, though that would be a harder one to judge. > If we find out that a dual PII/233 or PII/266 is slower than a dual P6-512K, > that would be VERY INTERESTING!!! Anyone willing to take up the challenge? If I remember right, there are numbers in the archives. Does anyone remember them? I don't have time to wade through the archives (playing with mame right now. :) Nate From owner-freebsd-smp Mon Nov 10 07:13:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id HAA05246 for smp-outgoing; Mon, 10 Nov 1997 07:13:15 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id HAA05239 for ; Mon, 10 Nov 1997 07:13:13 -0800 (PST) (envelope-from vanmaren@fast.cs.utah.edu) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id IAA27428; Mon, 10 Nov 1997 08:13:09 -0700 (MST) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id IAA21564; Mon, 10 Nov 1997 08:13:09 -0700 Date: Mon, 10 Nov 1997 08:13:09 -0700 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199711101513.IAA21564@fast.cs.utah.edu> To: freebsd-smp@freebsd.org Subject: Re: Best processor? Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk My 2 cents here. We went from a Pentium Pro/200 w/512k to a Pentium II/266 and it significantly faster (uniprocessor). (Before LX chipset was out) Price estimates (non-binding ;-) A Pentium Pro/200 w/512k is ~ $1100 A Pentium II/300 w/512k is ~ $800 A Pentium Pro/200 w/256k is ~ $500 A Pentium II/266 w/512k is ~ $550 For the price and options available, I'd go with the Pentium II, unless you really need features of the Pentium Pro (like 4-way SMP). But Intel is killing the Pro, except for 1MB chips for big servers. (That is only until Slot 2 is out). The L2 cache is slower, which may cause problems. But the processors only communicate at 66MHz anyway, so running the cache at twice the external bus isn't too bad. Yes, performance will suffer when the working set is > L1 and < L2 relative to the Pro. But a Pentium II/300 has a 150MHz L2 cache -- not much slower than the 200Mhz on the Pentium Pro/200 (okay, the Pentium Pro/166). BTW, intel is planning on selling the Pentium II w/o a L2 cache. With AGP, SDRAM, USB, MMX, blah, blah, blah, who cares about the actual *performance* -- you have all the buzzwords. Slot 2 (II?) will have a full-speed L2 Cache and support 4-way smp. It will be targeted at server-only marker. Kevin From owner-freebsd-smp Mon Nov 10 07:26:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id HAA06073 for smp-outgoing; Mon, 10 Nov 1997 07:26:29 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id HAA06068 for ; Mon, 10 Nov 1997 07:26:26 -0800 (PST) (envelope-from vanmaren@fast.cs.utah.edu) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id IAA27678; Mon, 10 Nov 1997 08:26:24 -0700 (MST) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id IAA23187; Mon, 10 Nov 1997 08:26:23 -0700 Date: Mon, 10 Nov 1997 08:26:23 -0700 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199711101526.IAA23187@fast.cs.utah.edu> To: freebsd-smp@freebsd.org Subject: Re: Best processor? Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Shoot -- I forgot to make my point. Two Pentium II/266's are about the same price as a Pentium Pro/200 w/512k cache. This is (I believe) a much better way of spending money. We have better performance on uni-processor (between those CPUs), and I would expect SMP to be faster (not slower) than the uni-processor case, so it is a win. Kevin From owner-freebsd-smp Mon Nov 10 10:01:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id KAA15867 for smp-outgoing; Mon, 10 Nov 1997 10:01:55 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from george.arc.nasa.gov (george.arc.nasa.gov [128.102.194.142]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id KAA15862 for ; Mon, 10 Nov 1997 10:01:53 -0800 (PST) (envelope-from lamaster@george.arc.nasa.gov) From: lamaster@george.arc.nasa.gov Received: (from lamaster@localhost) by george.arc.nasa.gov (8.8.7/8.8.7) id JAA06997 for freebsd-smp@FreeBSD.ORG; Mon, 10 Nov 1997 09:59:41 -0800 (PST) Date: Mon, 10 Nov 1997 09:59:41 -0800 (PST) Message-Id: <199711101759.JAA06997@george.arc.nasa.gov> To: freebsd-smp@freebsd.org Subject: Re: Best processor? Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > We went from a Pentium Pro/200 w/512k to a Pentium II/266 and it > significantly faster (uniprocessor). (Before LX chipset was out) : > The L2 cache is slower, which may cause problems. But the processors : > Slot 2 (II?) will have a full-speed L2 Cache and support 4-way smp. > It will be targeted at server-only marker. So far, I have not seen any numbers for the definitive benchmark: "make world". For the numbers to be meaningful, we need to define the conditions. *********************** converge on benchmark standard ******************* PPro or PII@nMHz, 1 or 2 processors, at least 64 MB EDO DRAM Orion (anybody using Orion?), Natoma, or LX (most common disk controller here: Symbios 3C875-Ultra-wide-based ????) (kernel configured with SCSI controller tagged queueing on) (most common disks here: Quantum Atlas II Ultra-Wide ???) /usr/src mounted: normally /usr/obj mounted: async,noatime (What "world" we are making: I assume 3.0-current) /etc/make.conf: CFLAGS= -O -pipe make -j4 -DNOCLEANDIR world ********************** any comments or corrections? ********************** Once we get the numbers for the make world benchmark in a standard configuration, we will know which processor is really, truly faster. ;-) BTW, most of the numbers I have seen have made it pretty much a wash, that is, at the same clock, PPro and PII seem to be about the same performance. On the PPro, an L1 cache miss is handled faster, but, on a PII, you have a bigger L1 cache. And the basic CPU cores are apparently nearly identical. One thing I am not clear on is the SMP performance. The PPro was notoriously slow at main-memory writes because of the cache-coherence protocol. I'm not sure if a PII on LX might be faster? If so, this could be a big advantage for PII/LX on SMP performance. Plus, SDRAM on an LX should help a little, too. So, let's see those benchmark numbers. From owner-freebsd-smp Tue Nov 11 15:53:05 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id PAA05161 for smp-outgoing; Tue, 11 Nov 1997 15:53:05 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from home.dragondata.com (toasty@home.dragondata.com [204.137.237.2]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id PAA05154 for ; Tue, 11 Nov 1997 15:53:00 -0800 (PST) (envelope-from toasty@home.dragondata.com) Received: (from toasty@localhost) by home.dragondata.com (8.8.5/8.8.5) id RAA07302 for freebsd-smp@freebsd.org; Tue, 11 Nov 1997 17:52:54 -0600 (CST) From: Kevin Day Message-Id: <199711112352.RAA07302@home.dragondata.com> Subject: Time drift. To: freebsd-smp@freebsd.org Date: Tue, 11 Nov 1997 17:52:53 -0600 (CST) X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I'm experimenting with the SMP kernel on a 3.0-SNAP release. The clock seems to be drifting about 30-45 seconds a day now, where it was less than 1 sec/week before. Anyone seen this before? Kevin Day From owner-freebsd-smp Tue Nov 11 16:28:45 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA07609 for smp-outgoing; Tue, 11 Nov 1997 16:28:45 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from ix.netcom.com (sil-wa4-47.ix.netcom.com [207.93.136.111]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id QAA07604 for ; Tue, 11 Nov 1997 16:28:42 -0800 (PST) (envelope-from tomdean@ix.netcom.com) Received: (from tomdean@localhost) by ix.netcom.com (8.8.8/8.8.8) id QAA00831; Tue, 11 Nov 1997 16:28:33 -0800 (PST) (envelope-from tomdean) Date: Tue, 11 Nov 1997 16:28:33 -0800 (PST) Message-Id: <199711120028.QAA00831@ix.netcom.com> From: Thomas Dean To: toasty@home.dragondata.com CC: freebsd-smp@freebsd.org In-reply-to: <199711112352.RAA07302@home.dragondata.com> (message from Kevin Day on Tue, 11 Nov 1997 17:52:53 -0600 (CST)) Subject: Re: Time drift. Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I am using FreeBSD celebris 3.0-CURRENT FreeBSD 3.0-CURRENT #0:\ Mon Nov 10 18:01:06 PST 1997 \ tomdean@celebris:/usr/src/sys/compile/CELEBRIS-SMP i386 cvsup'd Nov 10 at about 1000 PST. I use ntpdate bigben.cac.washington.edu Over 24 hours, I see 11 Nov 16:25:35 ntpdate: step time server 140.142.16.34 offset 5.938484 tomdean From owner-freebsd-smp Tue Nov 11 22:31:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id WAA16780 for smp-outgoing; Tue, 11 Nov 1997 22:31:28 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from merkur.dbft.daimlerbenz.com (merkur.DaimlerBenz.com [141.113.7.195]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id WAA16771 for ; Tue, 11 Nov 1997 22:31:24 -0800 (PST) (envelope-from andreas.strobel@dbag.ulm.DaimlerBenz.COM) Received: (from uucp@localhost) by merkur.dbft.daimlerbenz.com (8.7.1/8.7.1) id HAA10396 for ; Wed, 12 Nov 1997 07:33:07 +0100 (MET) Received: from sophie-scholl.dbag.ulm.daimlerbenz.com(53.16.8.3) by merkur.dbft.daimlerbenz.com via smap (3.2) id xma010394; Wed, 12 Nov 97 07:33:06 +0100 Received: from dagobert.dbag.ulm.DaimlerBenz.COM by sophie-scholl.dbag.ulm.DaimlerBenz.COM (5.x/SMI-SVR4-23.9.1997-e) id AA11961; Wed, 12 Nov 1997 07:33:41 +0100 Received: from mickey by dagobert.dbag.ulm.DaimlerBenz.COM (4.1/SMI-4.1-18.9.1995-gm) id AA23656; Wed, 12 Nov 97 07:26:09 +0100 Message-Id: <34694C00.41C67EA6@dbag.ulm.daimlerbenz.com> Date: Wed, 12 Nov 1997 07:26:08 +0100 From: Andreas Strobel Organization: Daimler Benz AG X-Mailer: Mozilla 3.0Gold (X11; I; SunOS 4.1.3 sun4c) Mime-Version: 1.0 To: freebsd-smp@freebsd.org Subject: motherboard Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, i want to build up a SMP-System with FreeBSD. I plan to buy the S1668 Titan Pro ATX motherboard. Works this board well with the SMP-Kernel? Or is a better alternative on the market? Thanks Andreas From owner-freebsd-smp Wed Nov 12 03:23:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id DAA26471 for smp-outgoing; Wed, 12 Nov 1997 03:23:41 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from bagpuss.visint.co.uk (bagpuss.visint.co.uk [194.207.134.1]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id DAA26463 for ; Wed, 12 Nov 1997 03:23:32 -0800 (PST) (envelope-from steve@visint.co.uk) Received: from dylan.visint.co.uk (dylan.visint.co.uk [194.207.134.180]) by bagpuss.visint.co.uk (8.7.5/8.7.3) with SMTP id LAA04577; Wed, 12 Nov 1997 11:23:04 GMT Date: Wed, 12 Nov 1997 11:23:25 +0000 (GMT) From: Stephen Roome To: Kevin Day cc: freebsd-smp@freebsd.org Subject: Re: Time drift. In-Reply-To: <199711112352.RAA07302@home.dragondata.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Tue, 11 Nov 1997, Kevin Day wrote: > > I'm experimenting with the SMP kernel on a 3.0-SNAP release. > > The clock seems to be drifting about 30-45 seconds a day now, where it was > less than 1 sec/week before. Actually, someone pointed this out to me this morning that my clock was wrong: Wed Nov 12 11:33:32 GMT 1997 The actual time is... Wed Nov 12 11:16:21 GMT 1997 The second machine that I'm using for reference is running xntpd, so I guess it shouldn't have gone out much, but the first one was set 34 days ago (I remember setting the clock last reboot).. That's 30 seconds out a day so it fits with your observation.. That's running 3.0-970618-SNAP (SMP), the reference machine is a very stable router which (needs an upgrade!) is running 3.0-970209-SNAP (UP) So I guess I'm going to be putting xntpd on this machine to find out what's going on.. > Anyone seen this before? [Well, I saw it before I read your post, but only by coincidence was it ten minutes before.. but then my clock is wrong so who's judging which is before =)] Steve -- Steve Roome - Vision Interactive Ltd. Tel:+44(0)117 9730597 Home:+44(0)976 241342 WWW: http://dylan.visint.co.uk/ From owner-freebsd-smp Wed Nov 12 16:16:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA23237 for smp-outgoing; Wed, 12 Nov 1997 16:16:09 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from abby.skypoint.net (abby.skypoint.net [199.86.32.252]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id QAA23204 for ; Wed, 12 Nov 1997 16:15:51 -0800 (PST) (envelope-from bruce@zuhause.mn.org) Received: (from uucp@localhost) by abby.skypoint.net (8.8.7/jl 1.3) with UUCP id SAA25898; Wed, 12 Nov 1997 18:15:29 -0600 (CST) Received: (from bruce@localhost) by zuhause.mn.org (8.8.7/8.8.7) id SAA23066; Wed, 12 Nov 1997 18:14:00 -0600 (CST) Date: Wed, 12 Nov 1997 18:14:00 -0600 (CST) Message-Id: <199711130014.SAA23066@zuhause.mn.org> From: Bruce Albrecht To: Andreas Strobel Cc: freebsd-smp@freebsd.org Subject: Re: motherboard In-Reply-To: <34694C00.41C67EA6@dbag.ulm.daimlerbenz.com> References: <34694C00.41C67EA6@dbag.ulm.daimlerbenz.com> X-Mailer: VM 6.34 under 19.16 "New York" XEmacs Lucid Mime-Version: 1.0 (generated by tm-edit 7.106) Content-Type: text/plain; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Andreas Strobel writes: > Hi, > i want to build up a SMP-System with FreeBSD. I plan to buy the S1668 > Titan Pro ATX motherboard. Works this board well with the > SMP-Kernel? Or is a better alternative on the market? I'm using it. The only problem I've had is with getting it to work well with 32MBx36 DRAM in ECC mode. I've found that I've had to do a bit of mix-n-match with DRAM to find some that work, and the losers at RC Systems (which I recommend you avoid) don't understand that it should work. From owner-freebsd-smp Wed Nov 12 16:55:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA26237 for smp-outgoing; Wed, 12 Nov 1997 16:55:55 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from monk.via.net (monk.via.net [140.174.204.10]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id QAA26223 for ; Wed, 12 Nov 1997 16:55:50 -0800 (PST) (envelope-from joe@via.net) Received: (from joe@localhost) by monk.via.net (8.6.11/8.6.12) id QAA20398 for smp@freebsd.org; Wed, 12 Nov 1997 16:45:32 -0800 Date: Wed, 12 Nov 1997 16:45:32 -0800 From: Joe McGuckin Message-Id: <199711130045.QAA20398@monk.via.net> To: smp@freebsd.org X-Sun-Charset: US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I'm running 2.2.5 on one of my machines. Would it be possible to run the SMP kernel on this machine - or would it require a complete reinstallation of the OS? Thanks! joe From owner-freebsd-smp Wed Nov 12 17:29:05 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id RAA29488 for smp-outgoing; Wed, 12 Nov 1997 17:29:05 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from monk.via.net (monk.via.net [140.174.204.10]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id RAA29483 for ; Wed, 12 Nov 1997 17:29:03 -0800 (PST) (envelope-from joe@via.net) Received: (from joe@localhost) by monk.via.net (8.6.11/8.6.12) id RAA20627 for smp@freebsd.org; Wed, 12 Nov 1997 17:13:41 -0800 Date: Wed, 12 Nov 1997 17:13:41 -0800 From: Joe McGuckin Message-Id: <199711130113.RAA20627@monk.via.net> To: smp@freebsd.org Subject: TYAN S1682/D X-Sun-Charset: US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Has this board sucessfully run the SMP kernel ? From owner-freebsd-smp Wed Nov 12 18:45:01 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id SAA05339 for smp-outgoing; Wed, 12 Nov 1997 18:45:01 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from monk.via.net (monk.via.net [140.174.204.10]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id SAA05328 for ; Wed, 12 Nov 1997 18:44:57 -0800 (PST) (envelope-from joe@via.net) Received: (from joe@localhost) by monk.via.net (8.6.11/8.6.12) id SAA21196 for smp@freebsd.org; Wed, 12 Nov 1997 18:34:37 -0800 Date: Wed, 12 Nov 1997 18:34:37 -0800 From: Joe McGuckin Message-Id: <199711130234.SAA21196@monk.via.net> To: smp@freebsd.org Subject: Heavily loaded SMP server X-Sun-Charset: US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I`d like to talk to someone who is using their SMP machine for heavy duty WWW serving, etc. joe From owner-freebsd-smp Wed Nov 12 19:33:06 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id TAA08468 for smp-outgoing; Wed, 12 Nov 1997 19:33:06 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from source.isd.state.in.us (source.isd.state.in.us [206.158.11.5]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id TAA08463 for ; Wed, 12 Nov 1997 19:33:01 -0800 (PST) (envelope-from sfra47@source.isd.state.in.us) Received: (from sfra47@localhost) by source.isd.state.in.us (8.8.5/8.8.3) id WAA28428; Wed, 12 Nov 1997 22:34:08 -0500 (EST) From: "Raymond L. Gilbert" Message-Id: <199711130334.WAA28428@source.isd.state.in.us> Subject: Re: Heavily loaded SMP server To: joe@via.net (Joe McGuckin) Date: Wed, 12 Nov 1997 22:34:07 -0500 (EST) Cc: smp@freebsd.org In-Reply-To: <199711130234.SAA21196@monk.via.net> from "Joe McGuckin" at Nov 12, 97 06:34:37 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Thus it was recorded by the prophets, Joe McGuckin said: >I`d like to talk to someone who is using their SMP machine for >heavy duty WWW serving, etc. Yeah so would I! Heavy duty WWW serving seems to imply putting SMP in a *production* environment. I've been running the SMP kernel for about a month now with few problems (none related to SMP code specifically), and I'd like to use it as a heavy-duty WWW server but I'm not sure that running -current on a production system is wise. Has anyone out there settled on a good, recent SNAP that could be used for heavy WWW traffic? -- Raymond L. Gilbert | "...the calculation of pi to 100 or 500 decimal Systems Administrator | places is wholly useless." State of Indiana | - Hermann Schubert on pi, 1889, Hamburg IDOA/ISD -*-*- pi@isd.state.in.us -*-*- http://source.isd.state.in.us/pi From owner-freebsd-smp Wed Nov 12 21:01:12 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id VAA13978 for smp-outgoing; Wed, 12 Nov 1997 21:01:12 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from home.dragondata.com (toasty@home.dragondata.com [204.137.237.2]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id VAA13969 for ; Wed, 12 Nov 1997 21:01:09 -0800 (PST) (envelope-from toasty@home.dragondata.com) Received: (from toasty@localhost) by home.dragondata.com (8.8.5/8.8.5) id XAA17161; Wed, 12 Nov 1997 23:00:48 -0600 (CST) From: Kevin Day Message-Id: <199711130500.XAA17161@home.dragondata.com> Subject: Re: Heavily loaded SMP server In-Reply-To: <199711130334.WAA28428@source.isd.state.in.us> from "Raymond L. Gilbert" at "Nov 12, 97 10:34:07 pm" To: sfra47@source.isd.state.in.us (Raymond L. Gilbert) Date: Wed, 12 Nov 1997 23:00:48 -0600 (CST) Cc: joe@via.net, smp@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Thus it was recorded by the prophets, Joe McGuckin said: > >I`d like to talk to someone who is using their SMP machine for > >heavy duty WWW serving, etc. > > Yeah so would I! Heavy duty WWW serving seems to imply putting SMP in > a *production* environment. I've been running the SMP kernel for > about a month now with few problems (none related to SMP code > specifically), and I'd like to use it as a heavy-duty WWW server but > I'm not sure that running -current on a production system is wise. > Has anyone out there settled on a good, recent SNAP that could > be used for heavy WWW traffic? > One of my machines is now running the 1108-SNAP version of the SMP kernel... Dual Pentium/200's. It's currently running about 30 eggdrops with no problem, for my users, and if it keeps running well this week, we're going to try moving one of the heavier used web pages we host to it. (www.mk4.com). The only problems we've had so far, are 1) Lots of programs are still expecting *.so.2.1 libraries, when all it has are the 3.0's. 2) We had a terrible time getting NIS to work with a 2.x server and a 3.x client. It turned out to be something to do with the libcrypt's, but I never really figured it all out, it just works now. 3) analog takes *longer* to run on a SMP kernel than a standard kernel... I'm still trying to figure that one out. 4) One user actually complained, because he thought gcc was broken, because it was compiling so fast. :) 5) Now I have no excuse to go buy a dual P/II machine. :) All these problems are 3.0 related, not SMP related though... KEvin Day DragonData From owner-freebsd-smp Wed Nov 12 22:22:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id WAA18625 for smp-outgoing; Wed, 12 Nov 1997 22:22:27 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from gratis.grondar.za (gratis.grondar.za [196.7.18.133]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id WAA18620 for ; Wed, 12 Nov 1997 22:22:23 -0800 (PST) (envelope-from mark@greenpeace.grondar.za) Received: from greenpeace.grondar.za (greenpeace.grondar.za [196.7.18.132]) by gratis.grondar.za (8.8.7/8.8.7) with ESMTP id IAA06117; Thu, 13 Nov 1997 08:22:22 +0200 (SAT) (envelope-from mark@greenpeace.grondar.za) Received: from greenpeace.grondar.za (localhost [127.0.0.1]) by greenpeace.grondar.za (8.8.7/8.8.7) with ESMTP id IAA11959; Thu, 13 Nov 1997 08:23:55 +0200 (SAST) Message-Id: <199711130623.IAA11959@greenpeace.grondar.za> X-Mailer: exmh version 2.0zeta 7/24/97 To: Joe McGuckin cc: smp@freebsd.org Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 13 Nov 1997 08:23:54 +0200 From: Mark Murray Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Joe McGuckin wrote: > I'm running 2.2.5 on one of my machines. Would it be possible to > run the SMP kernel on this machine - or would it require a complete > reinstallation of the OS? You'll need to reinstall the OS. 3.0-CURRENT has a new syscall. M -- Mark Murray Join the anti-SPAM movement: http://www.cauce.org From owner-freebsd-smp Thu Nov 13 05:53:43 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id FAA14982 for smp-outgoing; Thu, 13 Nov 1997 05:53:43 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from bagpuss.visint.co.uk (bagpuss.visint.co.uk [194.207.134.1]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id FAA14964 for ; Thu, 13 Nov 1997 05:53:37 -0800 (PST) (envelope-from steve@visint.co.uk) Received: from dylan.visint.co.uk (dylan.visint.co.uk [194.207.134.180]) by bagpuss.visint.co.uk (8.7.5/8.7.3) with SMTP id NAA09323; Thu, 13 Nov 1997 13:53:22 GMT Date: Thu, 13 Nov 1997 13:53:22 +0000 (GMT) From: Stephen Roome To: Kevin Day cc: "Raymond L. Gilbert" , joe@via.net, smp@freebsd.org Subject: Re: Heavily loaded SMP server In-Reply-To: <199711130500.XAA17161@home.dragondata.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Wed, 12 Nov 1997, Kevin Day wrote: > > Thus it was recorded by the prophets, Joe McGuckin said: > > >I`d like to talk to someone who is using their SMP machine for > > >heavy duty WWW serving, etc. > > > > Yeah so would I! Heavy duty WWW serving seems to imply putting SMP in > > a *production* environment. I've been running the SMP kernel for > > about a month now with few problems (none related to SMP code > > specifically), and I'd like to use it as a heavy-duty WWW server but > > I'm not sure that running -current on a production system is wise. > > Has anyone out there settled on a good, recent SNAP that could > > be used for heavy WWW traffic? I'm not too sure about WWW, our main web server here is still running 3.0-970618-SNAP, seems okay, it's a dual P133, and the load on it is miserably small at the moment (5k hits a day). So this machine is fine, doesn't crash/go wrong, well, it doesn't really do anything! However, I'm running 3.0-current as of last sunday on an identical machine which is being used as an intranet webserver, fileserver (apple & pc), shell accounts etc.. general purpose workhorse.. This machine does have some problems every now and again, and usually with high disk activity or heavy load to netatalk/appletalk. Before I upgraded from 3.0-970618-SNAP to current (on sunday) it was crashing about twice a week. Just freezing up, or spontaneous reboots (and I never got to the console in time)... My guess is this was an appletalk problem, although that's as good as any guess. This machine gets only about 1k hits a day, to the internal only website, but they are all fairly intensive cgi's and it handles this okay. YMMV, but don't fool yourself into thinking it's okay too soon.. I'd keep everything ready on whatever you do the transfer from so that you can just switch back by bringing the interface back up on the other machine. Could get nasty otherwise =) > 1) Lots of programs are still expecting *.so.2.1 libraries, when all it has > are the 3.0's. I haven't seen this problem yet.. (is this from not making world ?) Steve. -- Steve Roome - Vision Interactive Ltd. Tel:+44(0)117 9730597 Home:+44(0)976 241342 WWW: http://dylan.visint.co.uk/ From owner-freebsd-smp Thu Nov 13 07:11:02 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id HAA19419 for smp-outgoing; Thu, 13 Nov 1997 07:11:02 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from time.cdrom.com (root@time.cdrom.com [204.216.27.226]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id HAA19409 for ; Thu, 13 Nov 1997 07:11:00 -0800 (PST) (envelope-from jkh@time.cdrom.com) Received: from time.cdrom.com (jkh@localhost.cdrom.com [127.0.0.1]) by time.cdrom.com (8.8.7/8.6.9) with ESMTP id HAA20459 for ; Thu, 13 Nov 1997 07:10:54 -0800 (PST) To: smp@freebsd.org Subject: No CPU speed identification in CPU case? Date: Thu, 13 Nov 1997 07:10:54 -0800 Message-ID: <20455.879433854@time.cdrom.com> From: "Jordan K. Hubbard" Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk During boot, I noticed the following difference between uni-processor and SMP CPU identification: 1. Non-SMP case: CPU: Pentium Pro (232.67-MHz 686-class CPU) 2. SMP case: CPU: Pentium Pro (686-class CPU) Is there some problem with identifying the CPU speed in the SMP case? Just wondering. Thanks! Jordan From owner-freebsd-smp Thu Nov 13 23:13:38 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id XAA05405 for smp-outgoing; Thu, 13 Nov 1997 23:13:38 -0800 (PST) (envelope-from owner-freebsd-smp) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id XAA05397 for ; Thu, 13 Nov 1997 23:13:35 -0800 (PST) (envelope-from bde@zeta.org.au) Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.7/8.6.9) id SAA21788; Fri, 14 Nov 1997 18:11:45 +1100 Date: Fri, 14 Nov 1997 18:11:45 +1100 From: Bruce Evans Message-Id: <199711140711.SAA21788@godzilla.zeta.org.au> To: jkh@time.cdrom.com, smp@freebsd.org Subject: Re: No CPU speed identification in CPU case? Sender: owner-freebsd-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >1. Non-SMP case: > > CPU: Pentium Pro (232.67-MHz 686-class CPU) > >2. SMP case: > > CPU: Pentium Pro (686-class CPU) > >Is there some problem with identifying the CPU speed in the SMP case? >Just wondering. Thanks! The speed is kept in the global variable i586_ctr_freq, and everything related to this variable is disabled for SMP, since it needs to be per-CPU, or maybe an average, and code to initialize and use it per-cpu hasn't been written. Bruce