From owner-freebsd-smp Sun Apr 27 04:41:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id EAA08837 for smp-outgoing; Sun, 27 Apr 1997 04:41:34 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id EAA08832 for ; Sun, 27 Apr 1997 04:41:29 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id NAA24332 for smp@FreeBSD.org; Sun, 27 Apr 1997 13:30:46 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id KAA00833; Sun, 27 Apr 1997 10:25:54 +0200 (CEST) Message-ID: <19970427102554.50784@klemm.gtn.com> Date: Sun, 27 Apr 1997 10:25:54 +0200 From: Andreas Klemm To: Andreas Klemm Cc: smp@FreeBSD.org Subject: Re: compilation problems under -current of today References: <19970426172158.46581@klemm.gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 In-Reply-To: <19970426172158.46581@klemm.gtn.com>; from Andreas Klemm on Sat, Apr 26, 1997 at 05:21:58PM +0200 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Sat, Apr 26, 1997 at 05:21:58PM +0200, Andreas Klemm wrote: > FreeBSD-current of today, fbsd-smp of today: > > Somebody other seen this ? My configuration ? > Well a simple GENERIC kernel produced similar messages. > > make -ki > > cc -c -x assembler-with-cpp -DLOCORE -nostdinc -I- -I. -I../.. -I/usr/include -DTELES_HAS_MEMCPYB -DMD5 -DCOMPAT_43 -DNSWAPDEV=3 -DMFS -DFFS -DINET -DKERNEL ../../i386/i386/exception.s > machine/smpasm.h: Assembler messages: > machine/smpasm.h:28: Error: no such 386 instruction: `extern' The problems went away after using current's SMP ... -- andreas@klemm.gtn.com /\/\___ Wiechers & Partner Datentechnik GmbH Andreas Klemm ___/\/\/ Support Unix -- andreas.klemm@wup.de pgp p-key http://www-swiss.ai.mit.edu/~bal/pks-toplev.html >>> powered by <<< ftp://sunsite.unc.edu/pub/Linux/system/Printing/aps-491.tgz >>> FreeBSD <<< From owner-freebsd-smp Sun Apr 27 07:03:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA12623 for smp-outgoing; Sun, 27 Apr 1997 07:03:09 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA12614 for ; Sun, 27 Apr 1997 07:03:02 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id PAA07441 for smp@FreeBSD.ORG; Sun, 27 Apr 1997 15:47:18 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id PAA02858; Sun, 27 Apr 1997 15:34:12 +0200 (CEST) Message-ID: <19970427153411.08790@klemm.gtn.com> Date: Sun, 27 Apr 1997 15:34:11 +0200 From: Andreas Klemm To: smp@FreeBSD.ORG Subject: good success with -current && smp Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi ! Wanted to let you know, that the FreeBSD SMP kernel from -current is running very fine here (Tyan Titan PRO ATX board). After disabling DRAM Fast Leadoff in the BIOS "chipset features setup" I have a real stable system, that doesn't crash even under heavy load. I noticed too late, that the lkm modules doesn't work, so I can't give you some performance numbers now, but later this afternoon ;-) -- andreas@klemm.gtn.com /\/\___ Wiechers & Partner Datentechnik GmbH Andreas Klemm ___/\/\/ Support Unix -- andreas.klemm@wup.de pgp p-key http://www-swiss.ai.mit.edu/~bal/pks-toplev.html >>> powered by <<< ftp://sunsite.unc.edu/pub/Linux/system/Printing/aps-491.tgz >>> FreeBSD <<< From owner-freebsd-smp Sun Apr 27 09:41:20 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA18143 for smp-outgoing; Sun, 27 Apr 1997 09:41:20 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [194.77.0.15]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA18122 for ; Sun, 27 Apr 1997 09:41:11 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id SAA07728 for smp@FreeBSD.ORG; Sun, 27 Apr 1997 18:30:37 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id SAA12953; Sun, 27 Apr 1997 18:27:05 +0200 (CEST) Message-ID: <19970427182705.28628@klemm.gtn.com> Date: Sun, 27 Apr 1997 18:27:05 +0200 From: Andreas Klemm To: smp@FreeBSD.ORG Subject: make world finished Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk That make world was running while I was working on the machine. /etc/make.conf contained: NOLKM= YES /usr/obj was already cleaned... 2578.908u 2749.278s 2:17:40.24 64.5% 249+359k 52840+181686io 12175pf+0w I think I have to re-partition my system so that I'm not too i/o bound since my disks are not the fastest around ;-) -- andreas@klemm.gtn.com /\/\___ Wiechers & Partner Datentechnik GmbH Andreas Klemm ___/\/\/ Support Unix -- andreas.klemm@wup.de pgp p-key http://www-swiss.ai.mit.edu/~bal/pks-toplev.html >>> powered by <<< ftp://sunsite.unc.edu/pub/Linux/system/Printing/aps-491.tgz >>> FreeBSD <<< From owner-freebsd-smp Sun Apr 27 14:35:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA00584 for smp-outgoing; Sun, 27 Apr 1997 14:35:34 -0700 (PDT) Received: from 586quick166.saturn-tech.com ([207.229.19.32]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA00578 for ; Sun, 27 Apr 1997 14:35:31 -0700 (PDT) Received: from localhost (drussell@localhost) by 586quick166.saturn-tech.com (8.8.4/8.8.4) with SMTP id PAA07978; Sun, 27 Apr 1997 15:35:46 -0600 (MDT) X-Authentication-Warning: 586quick166.saturn-tech.com: drussell owned process doing -bs Date: Sun, 27 Apr 1997 15:35:46 -0600 (MDT) From: Doug Russell To: Kevin Van Maren cc: andreas@klemm.gtn.com, freebsd-smp@FreeBSD.ORG Subject: Re: AMI BIOS on Titan PRO In-Reply-To: <199704270512.XAA20562@fast.cs.utah.edu> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Sat, 26 Apr 1997, Kevin Van Maren wrote: > Both BIOSes use the same Flash ROM. Both BIOSes are both stored > in a 128k binary file that is written directly into the ROM. > However, the AMI BIOS uses a flash routine IN the ROM, so ... I would hope that AMI has a dos-based (or whatever) utility to flash their BIOS into the ROM. Perhaps checking their web sight would help? I'm just guessing here, but you would think that they would have come up with a way to do this. :) Later...... From owner-freebsd-smp Sun Apr 27 15:21:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA02481 for smp-outgoing; Sun, 27 Apr 1997 15:21:34 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA02476 for ; Sun, 27 Apr 1997 15:21:31 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id QAA22227 for ; Sun, 27 Apr 1997 16:21:30 -0600 (MDT) Message-Id: <199704272221.QAA22227@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@FreeBSD.org Subject: Re: compilation problems under -current of today In-reply-to: Your message of "Sat, 26 Apr 1997 11:10:14 MDT." <199704261710.LAA16333@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 16:21:30 -0600 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Hi, > Next I am going to cvsup 3.0-current and rebuild the world with the new > SMP kernel... stay tuned. 3.0-current is now the official SMP src tree. make world builds EXCEPT for LKMs, so to make everything BUT LKMs do: make -DNOLKM world OR add: NOLKM= YES to /etc/make.conf --- The IPI_INTS code has gone away, and the XFAST_IPI32 code is now standard. I expect APIC_IO to become standard within a few days. Note that by standard I mean MANDATORY, not "default but possivble to disable". So if your not using it now you had better try it soon... --- Please DO NOT conmtinue to use the smp tree, we're not going to waste time supporting it anymore (I suppose we should hide it) -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Apr 27 16:52:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA06115 for smp-outgoing; Sun, 27 Apr 1997 16:52:23 -0700 (PDT) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id QAA06109 for ; Sun, 27 Apr 1997 16:52:17 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id RAA06685; Sun, 27 Apr 1997 17:52:02 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id RAA08301; Sun, 27 Apr 1997 17:52:01 -0600 Date: Sun, 27 Apr 1997 17:52:01 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199704272352.RAA08301@fast.cs.utah.edu> To: drussell@saturn-tech.com, vanmaren@fast.cs.utah.edu Subject: Re: AMI BIOS on Titan PRO Cc: andreas@klemm.gtn.com, freebsd-smp@FreeBSD.ORG Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk >On Sat, 26 Apr 1997, Kevin Van Maren wrote: > >> Both BIOSes use the same Flash ROM. Both BIOSes are both stored >> in a 128k binary file that is written directly into the ROM. >> However, the AMI BIOS uses a flash routine IN the ROM, so >... > >I would hope that AMI has a dos-based (or whatever) utility to flash their >BIOS into the ROM. Perhaps checking their web sight would help? > >I'm just guessing here, but you would think that they would have come up >with a way to do this. :) > >Later...... > Why would AMI want you to use their BIOS over someone else's? That jsut means that you are using their BIOS w/o paying roaylties... There *is* a DOS utility used to flash the ROM -- however, it uses hooks in the BIOS to do the Flash. This has some advanteges, as it doesn't overwrite the entire BIOS, which is supposed to allow you to recover from a failure. Kevin From owner-freebsd-smp Sun Apr 27 17:07:24 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA07156 for smp-outgoing; Sun, 27 Apr 1997 17:07:24 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id RAA07134; Sun, 27 Apr 1997 17:07:21 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id SAA22684; Sun, 27 Apr 1997 18:06:53 -0600 (MDT) Message-Id: <199704280006.SAA22684@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Wolfgang Helbig cc: Vincent Poy , Peter Wemm , current@freebsd.org, smp@freebsd.org Subject: Re: make world cannot build lkm's - missing opt_smp.h Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 18:06:53 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, >make world fails when it tries to build LKMs because the files > opt_smp.h > opt_smp_invltbl.h >are missing. They are included at least by the following files > ... >I suggest the following patch to /usr/src/share/mk/bsd.kmod.mk as >a temporary workaround > ... done, thanx. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Apr 27 17:13:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA07452 for smp-outgoing; Sun, 27 Apr 1997 17:13:18 -0700 (PDT) Received: from twait.csv.warwick.ac.uk (csubl@twait.csv.warwick.ac.uk [137.205.148.199]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id RAA07446 for ; Sun, 27 Apr 1997 17:13:15 -0700 (PDT) From: Mr M P Searle Message-Id: <2375.199704280013@twait.csv.warwick.ac.uk> Received: by twait.csv.warwick.ac.uk id BAA02375; Mon, 28 Apr 1997 01:13:03 +0100 (BST) Subject: Quad Pro 150 motherboard? To: smp@freebsd.org Date: Mon, 28 Apr 1997 01:12:58 +0100 (BST) 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-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards around, and those that I could find were for large servers (eg Intel Alder, Goliath, etc.) Thanks, Michael. From owner-freebsd-smp Sun Apr 27 17:41:11 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA08659 for smp-outgoing; Sun, 27 Apr 1997 17:41:11 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id RAA08654 for ; Sun, 27 Apr 1997 17:41:02 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id UAA04840; Sun, 27 Apr 1997 20:38:48 -0400 Date: Sun, 27 Apr 1997 20:38:47 -0400 (EDT) From: Ben Black To: Mr M P Searle cc: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <2375.199704280013@twait.csv.warwick.ac.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk a 4 or 6 CPU P6 board for *other* than a large server...think about that for a few minutes. (hint: NT4 Workstation can't handle more than 2CPUs and that is the most popular commercial SMP-capable OS) On Mon, 28 Apr 1997, Mr M P Searle wrote: > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > around, and those that I could find were for large servers (eg Intel Alder, > Goliath, etc.) > > Thanks, Michael. > > From owner-freebsd-smp Sun Apr 27 18:50:40 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id SAA11842 for smp-outgoing; Sun, 27 Apr 1997 18:50:40 -0700 (PDT) Received: from jumping-spider.aracnet.com (root@jumping-spider.aracnet.com [204.188.47.14]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id SAA11837 for ; Sun, 27 Apr 1997 18:50:38 -0700 (PDT) Received: from cbrowni2-home (ppp-r7.aracnet.com [205.238.13.9]) by jumping-spider.aracnet.com (8.8.5/8.8.5) with SMTP id SAA16075; Sun, 27 Apr 1997 18:51:24 -0700 Message-ID: <336401FE.F21@earthling.net> Date: Sun, 27 Apr 1997 18:48:46 -0700 From: Chris Browning Reply-To: brownie@earthling.net X-Mailer: Mozilla 3.01 (WinNT; U) MIME-Version: 1.0 To: Mr M P Searle CC: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Michael, I concur with Ben here. Typically in the PC world, anything above 2 procs is geared towards servers. I can't think of many non-server > 2way systems out there. So, if you want >2way, you will have to go with one of the server class machines, which are typically expensive. In addition, I would not bother with the 150MHz PPP. If I remember correctly, the 150 only comes in the 256k cache size. If you are going to do 4way, do yourself a favor and get the 512k cache PPP. 4way PPP will saturate the processor bus quite quickly, so the more cache the better. I believe they make a 166/512k PPP, so for cost effectiveness, that is what I would recommend. Chris Not speaking for Intel. Ben Black wrote: > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > for a few minutes. > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > popular commercial SMP-capable OS) > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > around, and those that I could find were for large servers (eg Intel Alder, > > Goliath, etc.) > > > > Thanks, Michael. > > > > -- -----BEGIN PGP PUBLIC KEY BLOCK----- Version: 2.6.2 mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ =NxsF -----END PGP PUBLIC KEY BLOCK----- From owner-freebsd-smp Sun Apr 27 19:00:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA12237 for smp-outgoing; Sun, 27 Apr 1997 19:00:28 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA12232 for ; Sun, 27 Apr 1997 19:00:25 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id UAA23188; Sun, 27 Apr 1997 20:00:08 -0600 (MDT) Message-Id: <199704280200.UAA23188@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Ben Black cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Sun, 27 Apr 1997 20:38:47 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 20:00:08 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > a 4 or 6 CPU P6 board for *other* than a large server...think about that > for a few minutes. > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > popular commercial SMP-capable OS) Actually we have several users reporting success with 4-CPU freeBSD SMP systems! A 4-CPU systems makes sense for more than just servers. A development system for one. With our parallel make you can keep 4 CPUs busy with a group of parallel compiles. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Apr 27 19:06:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA12629 for smp-outgoing; Sun, 27 Apr 1997 19:06:03 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA12624 for ; Sun, 27 Apr 1997 19:06:00 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA06703; Sun, 27 Apr 1997 22:03:42 -0400 Date: Sun, 27 Apr 1997 22:03:42 -0400 (EDT) From: Ben Black To: Steve Passe cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704280200.UAA23188@Ilsa.StevesCafe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk my point was that they are not for general consumption. certainly, there are certain applications which make use of 4 CPUs, but my day goes just fine with a single P6-150 even when doing frequent recompiles. b3n On Sun, 27 Apr 1997, Steve Passe wrote: > Hi, > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > for a few minutes. > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > popular commercial SMP-capable OS) > > Actually we have several users reporting success with 4-CPU freeBSD SMP > systems! A 4-CPU systems makes sense for more than just servers. A > development system for one. With our parallel make you can > keep 4 CPUs busy with a group of parallel compiles. > > -- > Steve Passe | powered by > smp@csn.net | Symmetric MultiProcessor FreeBSD > > From owner-freebsd-smp Sun Apr 27 19:08:32 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA12858 for smp-outgoing; Sun, 27 Apr 1997 19:08:32 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA12845 for ; Sun, 27 Apr 1997 19:08:28 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA06723; Sun, 27 Apr 1997 22:06:08 -0400 Date: Sun, 27 Apr 1997 22:06:07 -0400 (EDT) From: Ben Black To: Chris Browning cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <336401FE.F21@earthling.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk i think a good rul eof thumb is about 256K per 64MB RAM (except for servers or where bus speeds far exceed RAM speeds). so, 4 CPUs with 256K cache should be fine in non-server configs. the 512K P6 chips are just not cost effective. last week i ordered a P6-150 for $165. tough to beat that. On Sun, 27 Apr 1997, Chris Browning wrote: > Michael, > I concur with Ben here. Typically in the PC world, anything above > 2 procs is geared towards servers. I can't think of many non-server > > 2way systems out there. So, if you want >2way, you will have > to go with one of the server class machines, which are typically > expensive. In addition, I would not bother with the 150MHz PPP. If > I remember correctly, the 150 only comes in the 256k cache size. If > you are going to do 4way, do yourself a favor and get the 512k cache > PPP. 4way PPP will saturate the processor bus quite quickly, so the > more cache the better. I believe they make a 166/512k PPP, so for > cost effectiveness, that is what I would recommend. > > Chris > Not speaking for Intel. > Ben Black wrote: > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > for a few minutes. > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > popular commercial SMP-capable OS) > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > around, and those that I could find were for large servers (eg Intel Alder, > > > Goliath, etc.) > > > > > > Thanks, Michael. > > > > > > > > -- > -----BEGIN PGP PUBLIC KEY BLOCK----- > Version: 2.6.2 > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > =NxsF > -----END PGP PUBLIC KEY BLOCK----- > From owner-freebsd-smp Sun Apr 27 19:16:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA13385 for smp-outgoing; Sun, 27 Apr 1997 19:16:15 -0700 (PDT) Received: from avatar.avatar.com (avatar.avatar.com [199.33.206.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA13380 for ; Sun, 27 Apr 1997 19:16:11 -0700 (PDT) Received: from avatar.avatar.com (kory@avatar.avatar.com [199.33.206.17]) by avatar.avatar.com (8.7.4/8.6.9) with SMTP id TAA06495; Sun, 27 Apr 1997 19:14:46 -0700 (PDT) Date: Sun, 27 Apr 1997 19:14:44 -0700 (PDT) From: Kory Hamzeh To: Chris Browning cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <336401FE.F21@earthling.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk What about using the system for software developement (lots of compiles, emacs, etc...)? Does having more than 2 cpu make sense? Kory On Sun, 27 Apr 1997, Chris Browning wrote: > Michael, > I concur with Ben here. Typically in the PC world, anything above > 2 procs is geared towards servers. I can't think of many non-server > > 2way systems out there. So, if you want >2way, you will have > to go with one of the server class machines, which are typically > expensive. In addition, I would not bother with the 150MHz PPP. If > I remember correctly, the 150 only comes in the 256k cache size. If > you are going to do 4way, do yourself a favor and get the 512k cache > PPP. 4way PPP will saturate the processor bus quite quickly, so the > more cache the better. I believe they make a 166/512k PPP, so for > cost effectiveness, that is what I would recommend. > > Chris > Not speaking for Intel. > Ben Black wrote: > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > for a few minutes. > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > popular commercial SMP-capable OS) > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > around, and those that I could find were for large servers (eg Intel Alder, > > > Goliath, etc.) > > > > > > Thanks, Michael. > > > > > > > > -- > -----BEGIN PGP PUBLIC KEY BLOCK----- > Version: 2.6.2 > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > =NxsF > -----END PGP PUBLIC KEY BLOCK----- > > From owner-freebsd-smp Sun Apr 27 19:24:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA13616 for smp-outgoing; Sun, 27 Apr 1997 19:24:33 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA13610 for ; Sun, 27 Apr 1997 19:24:31 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id WAA25442 for ; Sun, 27 Apr 1997 22:24:23 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id WAA03754 for ; Sun, 27 Apr 1997 22:24:21 -0400 Date: Sun, 27 Apr 1997 22:23:45 -0400 (EDT) From: Chuck Robey To: FreeBSD-SMP@FreeBSD.org Subject: SMP Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk I am in the throes of writing an OS for an OS class. This is a class project, and not ever going to be made public, but it got me thinking about smp. Specifically, I'm wondering how kernel stuff works with 2 processors .... are there two copies of the kernel running? Or does one processor have to interrupt the other to get the kernel stuff working? I'm curious about that, and about how mutual exclusion type things work when there're two processors that could be doing top level things at once. How are things like virtual memory handled? Do the two processors have to cooperate on VM? How about FS stuff ... are the two processors having ghe same view of the disk (especially vnodes?) I'm not looking for a treatise here, but maybe just a few words on the major changes that the smp code does would be really appreciated. Even one long paragraph would better than nothing (I don't have time to read thru all the code now). ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | 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 Apr 27 19:26:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA13646 for smp-outgoing; Sun, 27 Apr 1997 19:26:15 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA13641 for ; Sun, 27 Apr 1997 19:26:13 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA07155; Sun, 27 Apr 1997 22:23:55 -0400 Date: Sun, 27 Apr 1997 22:23:55 -0400 (EDT) From: Ben Black To: Kory Hamzeh cc: Chris Browning , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk well, is one person capable of changing so many source files between compiles that they need 4 CPUs to get it done in reasonable time? or is it that a *group* of developers all using the system as a compile *server* could do that? b3n On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > What about using the system for software developement (lots of compiles, > emacs, etc...)? Does having more than 2 cpu make sense? > > Kory > > > On Sun, 27 Apr 1997, Chris Browning wrote: > > > Michael, > > I concur with Ben here. Typically in the PC world, anything above > > 2 procs is geared towards servers. I can't think of many non-server > > > 2way systems out there. So, if you want >2way, you will have > > to go with one of the server class machines, which are typically > > expensive. In addition, I would not bother with the 150MHz PPP. If > > I remember correctly, the 150 only comes in the 256k cache size. If > > you are going to do 4way, do yourself a favor and get the 512k cache > > PPP. 4way PPP will saturate the processor bus quite quickly, so the > > more cache the better. I believe they make a 166/512k PPP, so for > > cost effectiveness, that is what I would recommend. > > > > Chris > > Not speaking for Intel. > > Ben Black wrote: > > > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > > for a few minutes. > > > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > popular commercial SMP-capable OS) > > > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > > around, and those that I could find were for large servers (eg Intel Alder, > > > > Goliath, etc.) > > > > > > > > Thanks, Michael. > > > > > > > > > > > > -- > > -----BEGIN PGP PUBLIC KEY BLOCK----- > > Version: 2.6.2 > > > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > > =NxsF > > -----END PGP PUBLIC KEY BLOCK----- > > > > > From owner-freebsd-smp Sun Apr 27 19:32:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA13823 for smp-outgoing; Sun, 27 Apr 1997 19:32:25 -0700 (PDT) Received: from avatar.avatar.com (avatar.avatar.com [199.33.206.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA13817 for ; Sun, 27 Apr 1997 19:32:23 -0700 (PDT) Received: from avatar.avatar.com (kory@avatar.avatar.com [199.33.206.17]) by avatar.avatar.com (8.7.4/8.6.9) with SMTP id TAA06582; Sun, 27 Apr 1997 19:30:57 -0700 (PDT) Date: Sun, 27 Apr 1997 19:30:55 -0700 (PDT) From: Kory Hamzeh To: Ben Black cc: Chris Browning , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Actually, I use pmake and something touching one .h file causes massive compiles, and, I build for many platforms. On Sun, 27 Apr 1997, Ben Black wrote: > well, is one person capable of changing so many source files between > compiles that they need 4 CPUs to get it done in reasonable time? or is > it that a *group* of developers all using the system as a compile > *server* could do that? > > > b3n > > On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > > > > What about using the system for software developement (lots of compiles, > > emacs, etc...)? Does having more than 2 cpu make sense? > > > > Kory > > > > > > On Sun, 27 Apr 1997, Chris Browning wrote: > > > > > Michael, > > > I concur with Ben here. Typically in the PC world, anything above > > > 2 procs is geared towards servers. I can't think of many non-server > > > > 2way systems out there. So, if you want >2way, you will have > > > to go with one of the server class machines, which are typically > > > expensive. In addition, I would not bother with the 150MHz PPP. If > > > I remember correctly, the 150 only comes in the 256k cache size. If > > > you are going to do 4way, do yourself a favor and get the 512k cache > > > PPP. 4way PPP will saturate the processor bus quite quickly, so the > > > more cache the better. I believe they make a 166/512k PPP, so for > > > cost effectiveness, that is what I would recommend. > > > > > > Chris > > > Not speaking for Intel. > > > Ben Black wrote: > > > > > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > > > for a few minutes. > > > > > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > > popular commercial SMP-capable OS) > > > > > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > > > around, and those that I could find were for large servers (eg Intel Alder, > > > > > Goliath, etc.) > > > > > > > > > > Thanks, Michael. > > > > > > > > > > > > > > > > -- > > > -----BEGIN PGP PUBLIC KEY BLOCK----- > > > Version: 2.6.2 > > > > > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > > > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > > > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > > > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > > > =NxsF > > > -----END PGP PUBLIC KEY BLOCK----- > > > > > > > > > From owner-freebsd-smp Sun Apr 27 19:33:44 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA13851 for smp-outgoing; Sun, 27 Apr 1997 19:33:44 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA13842 for ; Sun, 27 Apr 1997 19:33:34 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA07261; Sun, 27 Apr 1997 22:31:14 -0400 Date: Sun, 27 Apr 1997 22:31:13 -0400 (EDT) From: Ben Black To: Kory Hamzeh cc: Chris Browning , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk and enough people do this to drive demand up so much that the price drops below $3k for a quad P6 board? On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > Actually, I use pmake and something touching one .h file causes massive > compiles, and, I build for many platforms. > > > On Sun, 27 Apr 1997, Ben Black wrote: > > > well, is one person capable of changing so many source files between > > compiles that they need 4 CPUs to get it done in reasonable time? or is > > it that a *group* of developers all using the system as a compile > > *server* could do that? > > > > > > b3n > > > > On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > > > > > > > What about using the system for software developement (lots of compiles, > > > emacs, etc...)? Does having more than 2 cpu make sense? > > > > > > Kory > > > > > > > > > On Sun, 27 Apr 1997, Chris Browning wrote: > > > > > > > Michael, > > > > I concur with Ben here. Typically in the PC world, anything above > > > > 2 procs is geared towards servers. I can't think of many non-server > > > > > 2way systems out there. So, if you want >2way, you will have > > > > to go with one of the server class machines, which are typically > > > > expensive. In addition, I would not bother with the 150MHz PPP. If > > > > I remember correctly, the 150 only comes in the 256k cache size. If > > > > you are going to do 4way, do yourself a favor and get the 512k cache > > > > PPP. 4way PPP will saturate the processor bus quite quickly, so the > > > > more cache the better. I believe they make a 166/512k PPP, so for > > > > cost effectiveness, that is what I would recommend. > > > > > > > > Chris > > > > Not speaking for Intel. > > > > Ben Black wrote: > > > > > > > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > > > > for a few minutes. > > > > > > > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > > > popular commercial SMP-capable OS) > > > > > > > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > > > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > > > > around, and those that I could find were for large servers (eg Intel Alder, > > > > > > Goliath, etc.) > > > > > > > > > > > > Thanks, Michael. > > > > > > > > > > > > > > > > > > > > -- > > > > -----BEGIN PGP PUBLIC KEY BLOCK----- > > > > Version: 2.6.2 > > > > > > > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > > > > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > > > > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > > > > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > > > > =NxsF > > > > -----END PGP PUBLIC KEY BLOCK----- > > > > > > > > > > > > > > From owner-freebsd-smp Sun Apr 27 19:36:42 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA14059 for smp-outgoing; Sun, 27 Apr 1997 19:36:42 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA14052 for ; Sun, 27 Apr 1997 19:36:40 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA07358; Sun, 27 Apr 1997 22:34:29 -0400 Date: Sun, 27 Apr 1997 22:34:28 -0400 (EDT) From: Ben Black To: Chuck Robey cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk try: UNIX Systems for Modern Architectures and UNIX Internals: The New Frontier also, check out the VSTA kernel (it is fully threaded) as is Solaris. also, mach papers from CMU provide good info on it. b3n On Sun, 27 Apr 1997, Chuck Robey wrote: > I am in the throes of writing an OS for an OS class. This is a class > project, and not ever going to be made public, but it got me thinking > about smp. Specifically, I'm wondering how kernel stuff works with 2 > processors .... are there two copies of the kernel running? Or does one > processor have to interrupt the other to get the kernel stuff working? > > I'm curious about that, and about how mutual exclusion type things work > when there're two processors that could be doing top level things at once. > How are things like virtual memory handled? Do the two processors have > to cooperate on VM? > > How about FS stuff ... are the two processors having ghe same view of the > disk (especially vnodes?) > > I'm not looking for a treatise here, but maybe just a few words on the > major changes that the smp code does would be really appreciated. Even > one long paragraph would better than nothing (I don't have time to read > thru all the code now). > > ----------------------------+----------------------------------------------- > Chuck Robey | Interests include any kind of voice or data > chuckr@eng.umd.edu | communications topic, C programming, and Unix. > 9120 Edmonston Ct #302 | > 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 Apr 27 19:38:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA14110 for smp-outgoing; Sun, 27 Apr 1997 19:38:54 -0700 (PDT) Received: from avatar.avatar.com (avatar.avatar.com [199.33.206.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA14105 for ; Sun, 27 Apr 1997 19:38:50 -0700 (PDT) Received: from avatar.avatar.com (kory@avatar.avatar.com [199.33.206.17]) by avatar.avatar.com (8.7.4/8.6.9) with SMTP id TAA06658; Sun, 27 Apr 1997 19:37:22 -0700 (PDT) Date: Sun, 27 Apr 1997 19:37:22 -0700 (PDT) From: Kory Hamzeh To: Ben Black cc: Chris Browning , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I'm not going to sit here and argue with you. I am just saying that I have uses for more than two CPU's, and it is not a server. On Sun, 27 Apr 1997, Ben Black wrote: > and enough people do this to drive demand up so much that the price drops > below $3k for a quad P6 board? > > On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > > > > Actually, I use pmake and something touching one .h file causes massive > > compiles, and, I build for many platforms. > > > > > > On Sun, 27 Apr 1997, Ben Black wrote: > > > > > well, is one person capable of changing so many source files between > > > compiles that they need 4 CPUs to get it done in reasonable time? or is > > > it that a *group* of developers all using the system as a compile > > > *server* could do that? > > > > > > > > > b3n > > > > > > On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > > > > > > > > > > What about using the system for software developement (lots of compiles, > > > > emacs, etc...)? Does having more than 2 cpu make sense? > > > > > > > > Kory > > > > > > > > > > > > On Sun, 27 Apr 1997, Chris Browning wrote: > > > > > > > > > Michael, > > > > > I concur with Ben here. Typically in the PC world, anything above > > > > > 2 procs is geared towards servers. I can't think of many non-server > > > > > > 2way systems out there. So, if you want >2way, you will have > > > > > to go with one of the server class machines, which are typically > > > > > expensive. In addition, I would not bother with the 150MHz PPP. If > > > > > I remember correctly, the 150 only comes in the 256k cache size. If > > > > > you are going to do 4way, do yourself a favor and get the 512k cache > > > > > PPP. 4way PPP will saturate the processor bus quite quickly, so the > > > > > more cache the better. I believe they make a 166/512k PPP, so for > > > > > cost effectiveness, that is what I would recommend. > > > > > > > > > > Chris > > > > > Not speaking for Intel. > > > > > Ben Black wrote: > > > > > > > > > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > > > > > for a few minutes. > > > > > > > > > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > > > > popular commercial SMP-capable OS) > > > > > > > > > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > > > > > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > > > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > > > > > around, and those that I could find were for large servers (eg Intel Alder, > > > > > > > Goliath, etc.) > > > > > > > > > > > > > > Thanks, Michael. > > > > > > > > > > > > > > > > > > > > > > > > -- > > > > > -----BEGIN PGP PUBLIC KEY BLOCK----- > > > > > Version: 2.6.2 > > > > > > > > > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > > > > > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > > > > > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > > > > > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > > > > > =NxsF > > > > > -----END PGP PUBLIC KEY BLOCK----- > > > > > > > > > > > > > > > > > > > > From owner-freebsd-smp Sun Apr 27 19:43:24 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA14483 for smp-outgoing; Sun, 27 Apr 1997 19:43:24 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA14471 for ; Sun, 27 Apr 1997 19:43:16 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA07466; Sun, 27 Apr 1997 22:40:58 -0400 Date: Sun, 27 Apr 1997 22:40:57 -0400 (EDT) From: Ben Black To: Kory Hamzeh cc: Chris Browning , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk and the point is that quad CPU boards are expensive because the demand for them as anything but servers (and as servers too, really) IS ALMOST NONEXISTENT. look at all the extra costs of going from 2 to 4 CPUs on a board: you can't fit all 4 on the mainboard, so dauthercards are required, all schemes for this are proprietary. so whether or not *you* can use it is not the issue. the fact is you are greatly in the minority and you know it. think of yourself as the exception that proves the rule. b3n On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > I'm not going to sit here and argue with you. I am just saying that I > have uses for more than two CPU's, and it is not a server. > > On Sun, 27 Apr 1997, Ben Black wrote: > > > and enough people do this to drive demand up so much that the price drops > > below $3k for a quad P6 board? > > > > On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > > > > > > > Actually, I use pmake and something touching one .h file causes massive > > > compiles, and, I build for many platforms. > > > > > > > > > On Sun, 27 Apr 1997, Ben Black wrote: > > > > > > > well, is one person capable of changing so many source files between > > > > compiles that they need 4 CPUs to get it done in reasonable time? or is > > > > it that a *group* of developers all using the system as a compile > > > > *server* could do that? > > > > > > > > > > > > b3n > > > > > > > > On Sun, 27 Apr 1997, Kory Hamzeh wrote: > > > > > > > > > > > > > > What about using the system for software developement (lots of compiles, > > > > > emacs, etc...)? Does having more than 2 cpu make sense? > > > > > > > > > > Kory > > > > > > > > > > > > > > > On Sun, 27 Apr 1997, Chris Browning wrote: > > > > > > > > > > > Michael, > > > > > > I concur with Ben here. Typically in the PC world, anything above > > > > > > 2 procs is geared towards servers. I can't think of many non-server > > > > > > > 2way systems out there. So, if you want >2way, you will have > > > > > > to go with one of the server class machines, which are typically > > > > > > expensive. In addition, I would not bother with the 150MHz PPP. If > > > > > > I remember correctly, the 150 only comes in the 256k cache size. If > > > > > > you are going to do 4way, do yourself a favor and get the 512k cache > > > > > > PPP. 4way PPP will saturate the processor bus quite quickly, so the > > > > > > more cache the better. I believe they make a 166/512k PPP, so for > > > > > > cost effectiveness, that is what I would recommend. > > > > > > > > > > > > Chris > > > > > > Not speaking for Intel. > > > > > > Ben Black wrote: > > > > > > > > > > > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > > > > > > for a few minutes. > > > > > > > > > > > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > > > > > popular commercial SMP-capable OS) > > > > > > > > > > > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > > > > > > > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > > > > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > > > > > > around, and those that I could find were for large servers (eg Intel Alder, > > > > > > > > Goliath, etc.) > > > > > > > > > > > > > > > > Thanks, Michael. > > > > > > > > > > > > > > > > > > > > > > > > > > > > -- > > > > > > -----BEGIN PGP PUBLIC KEY BLOCK----- > > > > > > Version: 2.6.2 > > > > > > > > > > > > mQCNAzJSHeUAAAEEAKQrvftlb+sbnw0hA5vW2Orzq3rCGypldkYxRdXhx0yWx/dY > > > > > > U2PMqxgTwlOeQl3wA1IIWGMaHhbpPp0IegkOm9HIHEvc2G8uWywN5OvkaVFyuIHL > > > > > > juZ6VSem3cd63bqpNe3ZWtWwQdjFivm+YNeQveV220eTPfTuvbz7xZq+b9WZAAUR > > > > > > tCtDaHJpcyBTaGVybWFuIEJyb3duaW5nIDxjYnJvd25AYXJhY25ldC5jb20+ > > > > > > =NxsF > > > > > > -----END PGP PUBLIC KEY BLOCK----- > > > > > > > > > > > > > > > > > > > > > > > > > > > From owner-freebsd-smp Sun Apr 27 19:54:02 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA14905 for smp-outgoing; Sun, 27 Apr 1997 19:54:02 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA14899 for ; Sun, 27 Apr 1997 19:53:58 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id WAA29990; Sun, 27 Apr 1997 22:53:56 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id WAA07558; Sun, 27 Apr 1997 22:53:54 -0400 Date: Sun, 27 Apr 1997 22:53:18 -0400 (EDT) From: Chuck Robey To: Ben Black cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 27 Apr 1997, Ben Black wrote: > try: > UNIX Systems for Modern Architectures > and > UNIX Internals: The New Frontier > > also, check out the VSTA kernel (it is fully threaded) as is Solaris. > also, mach papers from CMU provide good info on it. I am looking for FreeBSD info, not Sun info. Heck, the Frontiers book has a sum total of 4 pages on architectures, none of which is about FreeBSD. I don't want all that much, a parapgraph or two on what I asked would do. Could Steve or Peter do that? I don't need to know how semaphores work. I want to know how FreeBSD SMP semaphores differ from the FreeBSD non-SMP semaphores (for instance). > > > b3n > > On Sun, 27 Apr 1997, Chuck Robey wrote: > > > I am in the throes of writing an OS for an OS class. This is a class > > project, and not ever going to be made public, but it got me thinking > > about smp. Specifically, I'm wondering how kernel stuff works with 2 > > processors .... are there two copies of the kernel running? Or does one > > processor have to interrupt the other to get the kernel stuff working? > > > > I'm curious about that, and about how mutual exclusion type things work > > when there're two processors that could be doing top level things at once. > > How are things like virtual memory handled? Do the two processors have > > to cooperate on VM? > > > > How about FS stuff ... are the two processors having the same view of the > > disk (especially vnodes?) > > > > I'm not looking for a treatise here, but maybe just a few words on the > > major changes that the smp code does would be really appreciated. Even > > one long paragraph would better than nothing (I don't have time to read > > thru all the code now). > > > > ----------------------------+----------------------------------------------- > > Chuck Robey | Interests include any kind of voice or data > > chuckr@eng.umd.edu | communications topic, C programming, and Unix. > > 9120 Edmonston Ct #302 | > > Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD > > (301) 220-2114 | version 3.0 current -- and great FUN! > > ----------------------------+----------------------------------------------- > > > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | 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 Apr 27 19:56:11 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA15042 for smp-outgoing; Sun, 27 Apr 1997 19:56:11 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA15037 for ; Sun, 27 Apr 1997 19:56:08 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id WAA07732; Sun, 27 Apr 1997 22:53:58 -0400 Date: Sun, 27 Apr 1997 22:53:57 -0400 (EDT) From: Ben Black To: Chuck Robey cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk freebsd-smp is not the best example of how to do SMP. it uses the simplest method: one giant kernel lock. i don't know that it is particularly representative of advanced SMP operating systems (though linux also uses a giant kernel lock). b3n On Sun, 27 Apr 1997, Chuck Robey wrote: > On Sun, 27 Apr 1997, Ben Black wrote: > > > try: > > UNIX Systems for Modern Architectures > > and > > UNIX Internals: The New Frontier > > > > also, check out the VSTA kernel (it is fully threaded) as is Solaris. > > also, mach papers from CMU provide good info on it. > > I am looking for FreeBSD info, not Sun info. Heck, the Frontiers book > has a sum total of 4 pages on architectures, none of which is about > FreeBSD. I don't want all that much, a parapgraph or two on what I asked > would do. Could Steve or Peter do that? > > I don't need to know how semaphores work. I want to know how FreeBSD SMP > semaphores differ from the FreeBSD non-SMP semaphores (for instance). > > > > > > > b3n > > > > On Sun, 27 Apr 1997, Chuck Robey wrote: > > > > > I am in the throes of writing an OS for an OS class. This is a class > > > project, and not ever going to be made public, but it got me thinking > > > about smp. Specifically, I'm wondering how kernel stuff works with 2 > > > processors .... are there two copies of the kernel running? Or does one > > > processor have to interrupt the other to get the kernel stuff working? > > > > > > I'm curious about that, and about how mutual exclusion type things work > > > when there're two processors that could be doing top level things at once. > > > How are things like virtual memory handled? Do the two processors have > > > to cooperate on VM? > > > > > > How about FS stuff ... are the two processors having the same view of the > > > disk (especially vnodes?) > > > > > > I'm not looking for a treatise here, but maybe just a few words on the > > > major changes that the smp code does would be really appreciated. Even > > > one long paragraph would better than nothing (I don't have time to read > > > thru all the code now). > > > > > > ----------------------------+----------------------------------------------- > > > Chuck Robey | Interests include any kind of voice or data > > > chuckr@eng.umd.edu | communications topic, C programming, and Unix. > > > 9120 Edmonston Ct #302 | > > > Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD > > > (301) 220-2114 | version 3.0 current -- and great FUN! > > > ----------------------------+----------------------------------------------- > > > > > > > > > ----------------------------+----------------------------------------------- > Chuck Robey | Interests include any kind of voice or data > chuckr@eng.umd.edu | communications topic, C programming, and Unix. > 9120 Edmonston Ct #302 | > 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 Apr 27 20:04:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA15565 for smp-outgoing; Sun, 27 Apr 1997 20:04:23 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA15557 for ; Sun, 27 Apr 1997 20:04:20 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id XAA01565; Sun, 27 Apr 1997 23:04:18 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id XAA08741; Sun, 27 Apr 1997 23:04:15 -0400 Date: Sun, 27 Apr 1997 23:03:38 -0400 (EDT) From: Chuck Robey To: Ben Black cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 27 Apr 1997, Ben Black wrote: > freebsd-smp is not the best example of how to do SMP. it uses the > simplest method: one giant kernel lock. i don't know that it is > particularly representative of advanced SMP operating systems (though > linux also uses a giant kernel lock). One per CPU? One lock total? How many copies of the kernel are running at once? How is caching handled, per cpu, or globally (forcing the two cpu's to look like one?) When a system call executes, what handles it, one cpu? ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | 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 Apr 27 20:52:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18759 for smp-outgoing; Sun, 27 Apr 1997 20:52:03 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18753 for ; Sun, 27 Apr 1997 20:52:01 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id XAA08681; Sun, 27 Apr 1997 23:49:51 -0400 Date: Sun, 27 Apr 1997 23:49:50 -0400 (EDT) From: Ben Black To: Chuck Robey cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 27 Apr 1997, Chuck Robey wrote: > On Sun, 27 Apr 1997, Ben Black wrote: > > > freebsd-smp is not the best example of how to do SMP. it uses the > > simplest method: one giant kernel lock. i don't know that it is > > particularly representative of advanced SMP operating systems (though > > linux also uses a giant kernel lock). > > One per CPU? One lock total? How many copies of the kernel are running one lock. one kernel. > at once? How is caching handled, per cpu, or globally (forcing the two > cpu's to look like one?) When a system call executes, what handles it, > one cpu? caching is handled in hardware with software awareness. system calls are handled by whichever CPU gets scheduled for it.. hence, symmetric. b3n From owner-freebsd-smp Sun Apr 27 20:57:45 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18957 for smp-outgoing; Sun, 27 Apr 1997 20:57:45 -0700 (PDT) Received: from nyx.pr.mcs.net (nyx.pr.mcs.net [204.95.55.81]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18946 for ; Sun, 27 Apr 1997 20:57:39 -0700 (PDT) Received: from nyx.pr.mcs.net (localhost [127.0.0.1]) by nyx.pr.mcs.net (8.8.5/8.8.5) with ESMTP id WAA12920; Sun, 27 Apr 1997 22:57:12 -0500 (CDT) Message-Id: <199704280357.WAA12920@nyx.pr.mcs.net> X-Mailer: exmh version 1.6.9 8/22/96 To: Ben Black cc: Chuck Robey , FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-reply-to: Your message of Sun, 27 Apr 1997 22:53:57 -0400. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 22:57:12 -0500 From: Chris Csanady Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk >freebsd-smp is not the best example of how to do SMP. it uses the >simplest method: one giant kernel lock. i don't know that it is >particularly representative of advanced SMP operating systems (though >linux also uses a giant kernel lock). Actually, linux has moved to a slightly finer grain system. Now they have seperate locks for the run queues, scheduler, and some other things.. --Chris Csanady From owner-freebsd-smp Sun Apr 27 20:58:56 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18995 for smp-outgoing; Sun, 27 Apr 1997 20:58:56 -0700 (PDT) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18990 for ; Sun, 27 Apr 1997 20:58:53 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id VAA09968; Sun, 27 Apr 1997 21:58:42 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id VAA10841; Sun, 27 Apr 1997 21:58:41 -0600 Date: Sun, 27 Apr 1997 21:58:41 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199704280358.VAA10841@fast.cs.utah.edu> To: black@zen.cypher.net, brownie@earthling.net Subject: Re: Quad Pro 150 motherboard? Cc: csubl@csv.warwick.ac.uk, smp@freebsd.org Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > >i think a good rul eof thumb is about 256K per 64MB RAM (except for >servers or where bus speeds far exceed RAM speeds). so, 4 CPUs with 256K >cache should be fine in non-server configs. the 512K P6 chips are just >not cost effective. last week i ordered a P6-150 for $165. tough to >beat that. > My rule of thumb was 256k/16MB ram. But that was before memory got soooo cheap. Actually, there have been numerous benchmarks that have shown that 3 166/512k chips blow away 4 200/256k chips. It was too bad that Intel discontinued the Pro/166. Kevin From owner-freebsd-smp Sun Apr 27 20:59:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA19060 for smp-outgoing; Sun, 27 Apr 1997 20:59:26 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA19055 for ; Sun, 27 Apr 1997 20:59:24 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id XAA07695; Sun, 27 Apr 1997 23:59:16 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id XAA14227; Sun, 27 Apr 1997 23:59:15 -0400 Date: Sun, 27 Apr 1997 23:58:37 -0400 (EDT) From: Chuck Robey To: Ben Black cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 27 Apr 1997, Ben Black wrote: > On Sun, 27 Apr 1997, Chuck Robey wrote: > > > On Sun, 27 Apr 1997, Ben Black wrote: > > > > > freebsd-smp is not the best example of how to do SMP. it uses the > > > simplest method: one giant kernel lock. i don't know that it is > > > particularly representative of advanced SMP operating systems (though > > > linux also uses a giant kernel lock). > > > > One per CPU? One lock total? How many copies of the kernel are running > one lock. one kernel. > > > at once? How is caching handled, per cpu, or globally (forcing the two > > cpu's to look like one?) When a system call executes, what handles it, > > one cpu? > > caching is handled in hardware with software awareness. system calls are > handled by whichever CPU gets scheduled for it.. hence, symmetric. I'm confused, then. If there's only one kernel, then only one cpu can run it, so only one cpu can field the system calls. If both cpu's can field system calls, then unless they contact the other one to get the work done, then there must be two copies of the kernel ruinning, right? I'm probably misunderstanding something. Maybe you meant only one piece of software called "kernel" but two cpus running it? > > > b3n > > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | 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 Apr 27 21:00:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA19150 for smp-outgoing; Sun, 27 Apr 1997 21:00:27 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA19139 for ; Sun, 27 Apr 1997 21:00:22 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id XAA08873; Sun, 27 Apr 1997 23:58:03 -0400 Date: Sun, 27 Apr 1997 23:58:02 -0400 (EDT) From: Ben Black To: Chris Csanady cc: Chuck Robey , FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: <199704280357.WAA12920@nyx.pr.mcs.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk i sit corrected. i expect they will eventually migrate to a fully threaded kernel. On Sun, 27 Apr 1997, Chris Csanady wrote: > > >freebsd-smp is not the best example of how to do SMP. it uses the > >simplest method: one giant kernel lock. i don't know that it is > >particularly representative of advanced SMP operating systems (though > >linux also uses a giant kernel lock). > > Actually, linux has moved to a slightly finer grain system. Now they > have seperate locks for the run queues, scheduler, and some other > things.. > > --Chris Csanady > From owner-freebsd-smp Sun Apr 27 21:16:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA20027 for smp-outgoing; Sun, 27 Apr 1997 21:16:18 -0700 (PDT) Received: from nyx.pr.mcs.net (nyx.pr.mcs.net [204.95.55.81]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA20019 for ; Sun, 27 Apr 1997 21:16:15 -0700 (PDT) Received: from nyx.pr.mcs.net (localhost [127.0.0.1]) by nyx.pr.mcs.net (8.8.5/8.8.5) with ESMTP id XAA12986; Sun, 27 Apr 1997 23:16:06 -0500 (CDT) Message-Id: <199704280416.XAA12986@nyx.pr.mcs.net> X-Mailer: exmh version 1.6.9 8/22/96 To: Ben Black cc: Chuck Robey , FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-reply-to: Your message of Sun, 27 Apr 1997 23:58:02 -0400. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 23:16:05 -0500 From: Chris Csanady Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk >i sit corrected. i expect they will eventually migrate to a fully >threaded kernel. As will we I hope. I was hoping to work on pushing the locks down into the syscalls earlier, but I ran into some trouble. I really knew very little about assembly, and our locks really are not up to it yet. :( Besides, the general concencus was that we didn't want to deal with it now.. --Chris Csanady >On Sun, 27 Apr 1997, Chris Csanady wrote: > >> >> >freebsd-smp is not the best example of how to do SMP. it uses the >> >simplest method: one giant kernel lock. i don't know that it is >> >particularly representative of advanced SMP operating systems (though >> >linux also uses a giant kernel lock). >> >> Actually, linux has moved to a slightly finer grain system. Now they >> have seperate locks for the run queues, scheduler, and some other >> things.. >> >> --Chris Csanady >> From owner-freebsd-smp Sun Apr 27 22:14:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA22724 for smp-outgoing; Sun, 27 Apr 1997 22:14:18 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA22718 for ; Sun, 27 Apr 1997 22:14:16 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id XAA23937 for ; Sun, 27 Apr 1997 23:14:15 -0600 (MDT) Message-Id: <199704280514.XAA23937@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Sun, 27 Apr 1997 21:58:41 MDT." <199704280358.VAA10841@fast.cs.utah.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 23:14:15 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > My rule of thumb was 256k/16MB ram. But that was before memory got > soooo cheap. Actually, there have been numerous benchmarks that have > shown that 3 166/512k chips blow away 4 200/256k chips. It was > too bad that Intel discontinued the Pro/166. but you can still get them. I bought two several weeks ago, and they are very happy running at 200mHz. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Apr 27 22:30:39 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA23439 for smp-outgoing; Sun, 27 Apr 1997 22:30:39 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA23434 for ; Sun, 27 Apr 1997 22:30:37 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id XAA24002; Sun, 27 Apr 1997 23:30:32 -0600 (MDT) Message-Id: <199704280530.XAA24002@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Chuck Robey cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-reply-to: Your message of "Sun, 27 Apr 1997 23:03:38 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 23:30:32 -0600 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Hi, > One per CPU? One lock total? How many copies of the kernel are running > at once? How is caching handled, per cpu, or globally (forcing the two > cpu's to look like one?) When a system call executes, what handles it, > one cpu? We use the giant lock model at present. This is our biggest performance bottleneck, and it will soon change. More specifically there is one lock for the entire (SMP aspect of the) system. There is one copy of the kernel, which both CPUs can execute. The lock prevents more than one from running it at the same time. It is a spin lock, and is a counting lock, ie the possessor can get it mutiple times. cacheing is done by hardware enforcing the MESI model. I believe a system call will be handled by the CPU on which the calling process is running. An interrupt could be handled by any available CPU (simple answer, details are more complex). Caching is almost invisible to the software, we still have to handle TLB invalidations with inter-CPU interrupts. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Apr 27 22:42:06 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA23886 for smp-outgoing; Sun, 27 Apr 1997 22:42:06 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA23878 for ; Sun, 27 Apr 1997 22:42:01 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id XAA24080; Sun, 27 Apr 1997 23:41:56 -0600 (MDT) Message-Id: <199704280541.XAA24080@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Chuck Robey cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-reply-to: Your message of "Sun, 27 Apr 1997 22:53:18 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 27 Apr 1997 23:41:56 -0600 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Hi, > I am looking for FreeBSD info, not Sun info. Heck, the Frontiers book > has a sum total of 4 pages on architectures, none of which is about > FreeBSD. I don't want all that much, a parapgraph or two on what I asked > would do. Could Steve or Peter do that? > > I don't need to know how semaphores work. I want to know how FreeBSD SMP > semaphores differ from the FreeBSD non-SMP semaphores (for instance). We realize that general documents are in order on a lot of issues. Right now, with the merge into -current and all, we are just too busy writing the code! I have setup a page on the website to start collecting whitepapers on these subjects, now we just need to find time to write them. My first paper will probably be a discussion of the APIC I/O system, from the point of view of its actual implimentation, and the WHYs of how it is written. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Sun Apr 27 22:53:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA24711 for smp-outgoing; Sun, 27 Apr 1997 22:53:34 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA24706 for ; Sun, 27 Apr 1997 22:53:32 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id BAA10889; Mon, 28 Apr 1997 01:51:21 -0400 Date: Mon, 28 Apr 1997 01:51:20 -0400 (EDT) From: Ben Black To: Chuck Robey cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > I'm confused, then. If there's only one kernel, then only one cpu can > run it, so only one cpu can field the system calls. If both cpu's can > field system calls, then unless they contact the other one to get the > work done, then there must be two copies of the kernel ruinning, right? > > I'm probably misunderstanding something. Maybe you meant only one piece > of software called "kernel" but two cpus running it? > hence my saying one kernel. there is a single kernel image in memory and both CPUs execute different (or the same) parts of it at the same time. the single lock is to keep the CPUs from stepping on each other. when one CPU wants to access a shared resource (and pretty much everything in the kernel is considered shared) then it acquires the mplock, does its business, releases the lock and continues. if the other CPU has the lock when one wants to acquire it, the second CPU waits. one kernel, one lock. b3n From owner-freebsd-smp Sun Apr 27 22:58:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA24830 for smp-outgoing; Sun, 27 Apr 1997 22:58:18 -0700 (PDT) Received: from root.com (implode.root.com [198.145.90.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA24824 for ; Sun, 27 Apr 1997 22:58:16 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by root.com (8.8.5/8.6.5) with SMTP id WAA02904; Sun, 27 Apr 1997 22:59:55 -0700 (PDT) Message-Id: <199704280559.WAA02904@root.com> X-Authentication-Warning: implode.root.com: localhost [127.0.0.1] didn't use HELO protocol To: Ben Black cc: Chuck Robey , FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-reply-to: Your message of "Mon, 28 Apr 1997 01:51:20 EDT." From: David Greenman Reply-To: dg@root.com Date: Sun, 27 Apr 1997 22:59:55 -0700 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk >> >> I'm confused, then. If there's only one kernel, then only one cpu can >> run it, so only one cpu can field the system calls. If both cpu's can >> field system calls, then unless they contact the other one to get the >> work done, then there must be two copies of the kernel ruinning, right? >> >> I'm probably misunderstanding something. Maybe you meant only one piece >> of software called "kernel" but two cpus running it? >> > >hence my saying one kernel. there is a single kernel image in memory and >both CPUs execute different (or the same) parts of it at the same time. >the single lock is to keep the CPUs from stepping on each other. when >one CPU wants to access a shared resource (and pretty much everything in >the kernel is considered shared) then it acquires the mplock, does its >business, releases the lock and continues. if the other CPU has the lock >when one wants to acquire it, the second CPU waits. > >one kernel, one lock. Uh, that's not what happens. There's a single copy of the kernel, but only one CPU may execute the kernel at a time. The lock is at the outer-most boundry of the kernel. -DG David Greenman Core-team/Principal Architect, The FreeBSD Project From owner-freebsd-smp Sun Apr 27 23:05:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA25137 for smp-outgoing; Sun, 27 Apr 1997 23:05:09 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA25132 for ; Sun, 27 Apr 1997 23:05:07 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id CAA11085; Mon, 28 Apr 1997 02:02:52 -0400 Date: Mon, 28 Apr 1997 02:02:51 -0400 (EDT) From: Ben Black To: David Greenman cc: Chuck Robey , FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: <199704280559.WAA02904@root.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > Uh, that's not what happens. There's a single copy of the kernel, but only > one CPU may execute the kernel at a time. The lock is at the outer-most > boundry of the kernel. > there i go wishing for finer grained locking. someday. b3n From owner-freebsd-smp Mon Apr 28 00:45:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA29322 for smp-outgoing; Mon, 28 Apr 1997 00:45:17 -0700 (PDT) Received: from mx.serv.net (mx.serv.net [199.201.191.10]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA29317 for ; Mon, 28 Apr 1997 00:45:14 -0700 (PDT) Received: from MindBender.serv.net by mx.serv.net (8.7.5/SERV Revision: 2.30) id AAA27324; Mon, 28 Apr 1997 00:45:16 -0700 (PDT) Received: from localhost.HeadCandy.com (michaelv@localhost.HeadCandy.com [127.0.0.1]) by MindBender.serv.net (8.7.5/8.7.3) with SMTP id AAA26201; Mon, 28 Apr 1997 00:43:32 -0700 (PDT) Message-Id: <199704280743.AAA26201@MindBender.serv.net> X-Authentication-Warning: MindBender.serv.net: Host michaelv@localhost.HeadCandy.com [127.0.0.1] didn't use HELO protocol To: Ben Black cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of Sun, 27 Apr 97 20:38:47 -0400. Date: Mon, 28 Apr 1997 00:43:16 -0700 From: "Michael L. VanLoon -- HeadCandy.com" Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >a 4 or 6 CPU P6 board for *other* than a large server...think about that >for a few minutes. > >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most >popular commercial SMP-capable OS) Bullshit. >On Mon, 28 Apr 1997, Mr M P Searle wrote: > >> Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking >> for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards >> around, and those that I could find were for large servers (eg Intel Alder, >> Goliath, etc.) >> >> Thanks, Michael. >> >> ----------------------------------------------------------------------------- Michael L. VanLoon michaelv@MindBender.serv.net --< Free your mind and your machine -- NetBSD free un*x >-- NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... NetBSD ports in progress: PICA, others... ----------------------------------------------------------------------------- From owner-freebsd-smp Mon Apr 28 01:04:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id BAA00210 for smp-outgoing; Mon, 28 Apr 1997 01:04:15 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA00205 for ; Mon, 28 Apr 1997 01:04:13 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id DAA13068; Mon, 28 Apr 1997 03:48:50 -0400 Date: Mon, 28 Apr 1997 03:48:49 -0400 (EDT) From: Ben Black To: "Michael L. VanLoon -- HeadCandy.com" cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704280743.AAA26201@MindBender.serv.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Michael L. VanLoon -- HeadCandy.com wrote: > > >a 4 or 6 CPU P6 board for *other* than a large server...think about that > >for a few minutes. > > > >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most > >popular commercial SMP-capable OS) > > Bullshit. > perhaps you'd like to clarify. NT *server* can handle more CPUs, but NT *workstation* is limited to 2. if you have a microsoft URL i can peruse to dispute that, i'd be happy to admit my mistake. but i doubt you can provide such a thing. b3n From owner-freebsd-smp Mon Apr 28 01:49:52 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id BAA02103 for smp-outgoing; Mon, 28 Apr 1997 01:49:52 -0700 (PDT) Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA02071 for ; Mon, 28 Apr 1997 01:49:27 -0700 (PDT) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.8.4/8.8.4) with SMTP id LAA15562; Mon, 28 Apr 1997 11:53:49 +0300 (EEST) Date: Mon, 28 Apr 1997 11:53:46 +0300 (EEST) From: Narvi To: Steve Passe cc: Ben Black , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704280200.UAA23188@Ilsa.StevesCafe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 27 Apr 1997, Steve Passe wrote: > Hi, > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > for a few minutes. > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > popular commercial SMP-capable OS) > > Actually we have several users reporting success with 4-CPU freeBSD SMP > systems! A 4-CPU systems makes sense for more than just servers. A > development system for one. With our parallel make you can > keep 4 CPUs busy with a group of parallel compiles. Just what I was going to say. :-) It may be even worse, depending on what you are developing - I currently tend to hog with compiles not only my machine but also at least one of our servers as testing will have to happen on my box no matter what. They all are 1 processor only on the moment. A 4 proc. one really wouldn't hurt - I have got to profile it somewhere as well. Sander > > -- > Steve Passe | powered by > smp@csn.net | Symmetric MultiProcessor FreeBSD > > > From owner-freebsd-smp Mon Apr 28 02:03:58 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA02768 for smp-outgoing; Mon, 28 Apr 1997 02:03:58 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id CAA02756 for ; Mon, 28 Apr 1997 02:03:55 -0700 (PDT) Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by who.cdrom.com (8.8.5/8.6.11) with ESMTP id CAA00806 for ; Mon, 28 Apr 1997 02:03:23 -0700 (PDT) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.8.4/8.8.4) with SMTP id MAA15686; Mon, 28 Apr 1997 12:04:36 +0300 (EEST) Date: Mon, 28 Apr 1997 12:04:34 +0300 (EEST) From: Narvi To: "Michael L. VanLoon -- HeadCandy.com" cc: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704280743.AAA26201@MindBender.serv.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Michael L. VanLoon -- HeadCandy.com wrote: > > >a 4 or 6 CPU P6 board for *other* than a large server...think about that > >for a few minutes. > > > >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most > >popular commercial SMP-capable OS) > > Bullshit. Well, at least 2 of the 2 NT workstations installed from 2 different CD-s claim they are capable of upto 2 proc-s. I haven't also seen in any place a reference that says otherwise. Sander > [snip] > > > ----------------------------------------------------------------------------- > Michael L. VanLoon michaelv@MindBender.serv.net > --< Free your mind and your machine -- NetBSD free un*x >-- > NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, > Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... > NetBSD ports in progress: PICA, others... > ----------------------------------------------------------------------------- > From owner-freebsd-smp Mon Apr 28 02:04:43 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA02803 for smp-outgoing; Mon, 28 Apr 1997 02:04:43 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id CAA02798 for ; Mon, 28 Apr 1997 02:04:41 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id EAA14143; Mon, 28 Apr 1997 04:49:18 -0400 Date: Mon, 28 Apr 1997 04:49:18 -0400 (EDT) From: Ben Black To: "Michael L. VanLoon -- HeadCandy.com" cc: Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704280743.AAA26201@MindBender.serv.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk from http://www.microsoft.com/syspro/technet/boes/bo/winntas/technote/nt101.htm Windows NT Workstation supports up to two processors in a symmetric multiprocessing environment. have a nice day. b3n On Mon, 28 Apr 1997, Michael L. VanLoon -- HeadCandy.com wrote: > > >a 4 or 6 CPU P6 board for *other* than a large server...think about that > >for a few minutes. > > > >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most > >popular commercial SMP-capable OS) > > Bullshit. > > >On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > >> Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > >> for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > >> around, and those that I could find were for large servers (eg Intel Alder, > >> Goliath, etc.) > >> > >> Thanks, Michael. > >> > >> > > > ----------------------------------------------------------------------------- > Michael L. VanLoon michaelv@MindBender.serv.net > --< Free your mind and your machine -- NetBSD free un*x >-- > NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, > Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... > NetBSD ports in progress: PICA, others... > ----------------------------------------------------------------------------- > From owner-freebsd-smp Mon Apr 28 02:06:19 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA02852 for smp-outgoing; Mon, 28 Apr 1997 02:06:19 -0700 (PDT) Received: from freefall.freebsd.org (freefall.cdrom.com [204.216.27.21]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id CAA02847 for ; Mon, 28 Apr 1997 02:06:18 -0700 (PDT) From: Peter Wemm Received: (from peter@localhost) by freefall.freebsd.org (8.8.5/8.8.5) id CAA03008 for freebsd-smp; Mon, 28 Apr 1997 02:05:58 -0700 (PDT) Date: Mon, 28 Apr 1997 02:05:58 -0700 (PDT) Message-Id: <199704280905.CAA03008@freefall.freebsd.org> To: freebsd-smp@FreeBSD.org Subject: cvs commit: sys/sys types.h Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk peter 97/04/28 02:05:58 Modified: sys types.h Log: This source tree is now "for reference", trivially prevent it from being accidently compiled. The the 3.0-current version now. Revision Changes Path 1.2 +37 -23 sys/sys/types.h From owner-freebsd-smp Mon Apr 28 02:12:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA03085 for smp-outgoing; Mon, 28 Apr 1997 02:12:08 -0700 (PDT) Received: from spinner.DIALix.COM (root@spinner.dialix.com [192.203.228.67]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id CAA03054; Mon, 28 Apr 1997 02:11:34 -0700 (PDT) Received: from spinner.DIALix.COM (peter@localhost.DIALix.oz.au [127.0.0.1]) by spinner.DIALix.COM with ESMTP id RAA18382; Mon, 28 Apr 1997 17:11:19 +0800 (WST) Message-Id: <199704280911.RAA18382@spinner.DIALix.COM> X-Mailer: exmh version 2.0gamma 1/27/96 To: Peter Wemm cc: freebsd-smp@FreeBSD.org Subject: Re: cvs commit: sys/sys types.h In-reply-to: Your message of "Mon, 28 Apr 1997 02:05:58 MST." <199704280905.CAA03008@freefall.freebsd.org> Date: Mon, 28 Apr 1997 17:11:18 +0800 From: Peter Wemm Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Peter Wemm wrote: peter 97/04/28 02:05:58 Modified: sys types.h Log: This source tree is now "for reference", trivially prevent it from being accidently compiled. The the 3.0-current version now. Revision Changes Path 1.2 +37 -23 sys/sys/types.h ^^^^^^^^^^^^^^^^^^ Don't believe the size of this, it's a 1-line addition. The large delta is because of the way rcs stores it's deltas for 1.2 relative to 1.1.1.1 even though the current version was 1.1.1.4. Cheers, -Peter From owner-freebsd-smp Mon Apr 28 02:19:58 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id CAA03351 for smp-outgoing; Mon, 28 Apr 1997 02:19:58 -0700 (PDT) Received: from nyx.pr.mcs.net (nyx.pr.mcs.net [204.95.55.81]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id CAA03346 for ; Mon, 28 Apr 1997 02:19:54 -0700 (PDT) Received: from nyx.pr.mcs.net (localhost [127.0.0.1]) by nyx.pr.mcs.net (8.8.5/8.8.5) with ESMTP id EAA14037; Mon, 28 Apr 1997 04:13:28 -0500 (CDT) Message-Id: <199704280913.EAA14037@nyx.pr.mcs.net> X-Mailer: exmh version 1.6.9 8/22/96 To: "Michael L. VanLoon -- HeadCandy.com" cc: Ben Black , Mr M P Searle , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of Mon, 28 Apr 1997 00:43:16 -0700. <199704280743.AAA26201@MindBender.serv.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 28 Apr 1997 04:13:28 -0500 From: Chris Csanady Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > >>a 4 or 6 CPU P6 board for *other* than a large server...think about that >>for a few minutes. >> >>(hint: NT4 Workstation can't handle more than 2CPUs and that is the most >>popular commercial SMP-capable OS) > >Bullshit. Hey, NT4 falls over with 1 cpu.. So, you must be talking about it being the most popular SMP OS, correct? ;-) --Chris Csanady >>On Mon, 28 Apr 1997, Mr M P Searle wrote: >> >>> Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking >>> for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards >>> around, and those that I could find were for large servers (eg Intel Alder, >>> Goliath, etc.) >>> >>> Thanks, Michael. >>> >>> > > >----------------------------------------------------------------------------- > Michael L. VanLoon michaelv@MindBender.serv.net > --< Free your mind and your machine -- NetBSD free un*x >-- > NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, > Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... > NetBSD ports in progress: PICA, others... >----------------------------------------------------------------------------- From owner-freebsd-smp Mon Apr 28 04:56:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id EAA10456 for smp-outgoing; Mon, 28 Apr 1997 04:56:49 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id EAA10451 for ; Mon, 28 Apr 1997 04:56:47 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id HAA28774; Mon, 28 Apr 1997 07:56:45 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id HAA10812; Mon, 28 Apr 1997 07:56:44 -0400 Date: Mon, 28 Apr 1997 07:56:05 -0400 (EDT) From: Chuck Robey To: Steve Passe cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: <199704280541.XAA24080@Ilsa.StevesCafe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Sun, 27 Apr 1997, Steve Passe wrote: > Hi, > > > I am looking for FreeBSD info, not Sun info. Heck, the Frontiers book > > has a sum total of 4 pages on architectures, none of which is about > > FreeBSD. I don't want all that much, a parapgraph or two on what I asked > > would do. Could Steve or Peter do that? > > > > I don't need to know how semaphores work. I want to know how FreeBSD SMP > > semaphores differ from the FreeBSD non-SMP semaphores (for instance). > > We realize that general documents are in order on a lot of issues. Right now, > with the merge into -current and all, we are just too busy writing the code! > I have setup a page on the website to start collecting whitepapers on these > subjects, now we just need to find time to write them. My first paper will > probably be a discussion of the APIC I/O system, from the point of view of > its actual implimentation, and the WHYs of how it is written. Actually, your previous comment was what I was looking for. I figure when I get time to march thru the code, I'll figure the rest out, but I want to understand who's executing what to be able to understand any failures. > > -- > Steve Passe | powered by > smp@csn.net | Symmetric MultiProcessor FreeBSD > > > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+----------------------------------------------- From owner-freebsd-smp Mon Apr 28 05:00:19 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id FAA10579 for smp-outgoing; Mon, 28 Apr 1997 05:00:19 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id FAA10574 for ; Mon, 28 Apr 1997 05:00:17 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id IAA29046; Mon, 28 Apr 1997 08:00:16 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id IAA11041; Mon, 28 Apr 1997 08:00:14 -0400 Date: Mon, 28 Apr 1997 07:59:35 -0400 (EDT) From: Chuck Robey To: Ben Black cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Ben Black wrote: > > > > I'm confused, then. If there's only one kernel, then only one cpu can > > run it, so only one cpu can field the system calls. If both cpu's can > > field system calls, then unless they contact the other one to get the > > work done, then there must be two copies of the kernel ruinning, right? > > > > I'm probably misunderstanding something. Maybe you meant only one piece > > of software called "kernel" but two cpus running it? > > > > hence my saying one kernel. there is a single kernel image in memory and > both CPUs execute different (or the same) parts of it at the same time. > the single lock is to keep the CPUs from stepping on each other. when > one CPU wants to access a shared resource (and pretty much everything in > the kernel is considered shared) then it acquires the mplock, does its > business, releases the lock and continues. if the other CPU has the lock > when one wants to acquire it, the second CPU waits. > > one kernel, one lock. Ahh, ok, I'm clear on that now. > > > b3n > > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+----------------------------------------------- From owner-freebsd-smp Mon Apr 28 09:26:00 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA22641 for smp-outgoing; Mon, 28 Apr 1997 09:26:00 -0700 (PDT) Received: from wile-e-coyote.cs.washington.edu (wile-e-coyote.cs.washington.edu [128.95.2.83]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA22636 for ; Mon, 28 Apr 1997 09:25:57 -0700 (PDT) From: mef@cs.washington.edu Received: (mef@localhost) by wile-e-coyote.cs.washington.edu (8.7.2/7.2ws+) id JAA17644; Mon, 28 Apr 1997 09:25:44 -0700 (PDT) Date: Mon, 28 Apr 1997 09:25:44 -0700 (PDT) Message-Id: <199704281625.JAA17644@wile-e-coyote.cs.washington.edu> To: smp@freebsd.org In-reply-to: (message from Ben Black on Mon, 28 Apr 1997 04:49:18 -0400 (EDT)) Subject: Re: Quad Pro 150 motherboard? Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Date: Mon, 28 Apr 1997 04:49:18 -0400 (EDT) From: Ben Black from http://www.microsoft.com/syspro/technet/boes/bo/winntas/technote/nt101.htm Windows NT Workstation supports up to two processors in a symmetric multiprocessing environment. According to that document NT Workstation can only support two processors, but NT Server can support four processors. I wouldn't be surprised if embedded in the code there was some sort of check: if (ms->makemoremoney->is_nt_server == TRUE) { /* support four processors */ } :) M From owner-freebsd-smp Mon Apr 28 10:00:31 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA24425 for smp-outgoing; Mon, 28 Apr 1997 10:00:31 -0700 (PDT) Received: from spinner.DIALix.COM (root@spinner.dialix.com [192.203.228.67]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA24399 for ; Mon, 28 Apr 1997 10:00:22 -0700 (PDT) Received: from spinner.DIALix.COM (peter@localhost.DIALix.oz.au [127.0.0.1]) by spinner.DIALix.COM with ESMTP id AAA21063; Tue, 29 Apr 1997 00:59:07 +0800 (WST) Message-Id: <199704281659.AAA21063@spinner.DIALix.COM> X-Mailer: exmh version 2.0gamma 1/27/96 To: mef@cs.washington.edu cc: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Mon, 28 Apr 1997 09:25:44 MST." <199704281625.JAA17644@wile-e-coyote.cs.washington.edu> Date: Tue, 29 Apr 1997 00:59:06 +0800 From: Peter Wemm Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk mef@cs.washington.edu wrote: > Date: Mon, 28 Apr 1997 04:49:18 -0400 (EDT) > From: Ben Black > from > http://www.microsoft.com/syspro/technet/boes/bo/winntas/technote/nt101.htm > Windows NT Workstation supports up to two processors in a symmetric > multiprocessing environment. > > According to that document NT Workstation can only support two > processors, but NT Server can support four processors. I wouldn't be > surprised if embedded in the code there was some sort of check: > if (ms->makemoremoney->is_nt_server == TRUE) { > /* support four processors */ > } > > :) Yes, that's been fairly well established already. There have been several technical articles about how the NT4 Workstation kernel is identical to the Server version. The article that I'm thinking of described how a certain registry entry controlled what mode the machine ran in. NT3.5 allowed this to change, but NT4 has some process running to reset it back. The article described how they patched something or other to stop the watcher process/thread/whatever and change it, and reboot, and a NT4 Workstation becomes a server, and benchmarks just like a server, etc. Hmm.. Lemmesee... The tech details: http://software.ora.com/news/ms_internet_andrews.html The collection of information about other anti-competitive things M$ have done with NT4: http://software.ora.com/news/ms_internet_frame.html BTW, There are apparently three values.. WinNT, ServerNT and ``LanmanNT''.. Hmm.. The last one, I wonder what that is? > M Cheers, -Peter From owner-freebsd-smp Mon Apr 28 10:20:06 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA25840 for smp-outgoing; Mon, 28 Apr 1997 10:20:06 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id KAA25773 for ; Mon, 28 Apr 1997 10:20:01 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02093; Mon, 28 Apr 1997 10:17:05 -0700 From: Terry Lambert Message-Id: <199704281717.KAA02093@phaeton.artisoft.com> Subject: Re: Quad Pro 150 motherboard? To: black@zen.cypher.net (Ben Black) Date: Mon, 28 Apr 1997 10:17:05 -0700 (MST) Cc: kory@avatar.com, brownie@earthling.net, csubl@csv.warwick.ac.uk, smp@freebsd.org In-Reply-To: from "Ben Black" at Apr 27, 97 10:40:57 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-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > and the point is that quad CPU boards are expensive because the demand > for them as anything but servers (and as servers too, really) IS ALMOST > NONEXISTENT. look at all the extra costs of going from 2 to 4 CPUs on a > board: you can't fit all 4 on the mainboard, so dauthercards are > required, all schemes for this are proprietary. > > so whether or not *you* can use it is not the issue. the fact is you are > greatly in the minority and you know it. think of yourself as the > exception that proves the rule. The PPro's are stackable without additional electronics external to the processor. This makes it a *lot* easier to do multiple CPU's up to large numbers. Personally, I'd like a 32 processor machine; of course, that means fixing the kernel so that it doesn't bog down on interprocessor synchronization, starting with the allocation mechanism and a hierarchical lock manager. My goal would be to have incremental compiles going on in edit sessions so that when I was done editing, the code was read to run. I hate this "wait for the compile" BS. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 10:26:19 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA26139 for smp-outgoing; Mon, 28 Apr 1997 10:26:19 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id KAA26134 for ; Mon, 28 Apr 1997 10:26:16 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02110; Mon, 28 Apr 1997 10:24:57 -0700 From: Terry Lambert Message-Id: <199704281724.KAA02110@phaeton.artisoft.com> Subject: Re: SMP To: chuckr@mat.net (Chuck Robey) Date: Mon, 28 Apr 1997 10:24:57 -0700 (MST) Cc: black@zen.cypher.net, FreeBSD-SMP@FreeBSD.org In-Reply-To: from "Chuck Robey" at Apr 27, 97 11:03:38 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-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > freebsd-smp is not the best example of how to do SMP. it uses the > > simplest method: one giant kernel lock. i don't know that it is > > particularly representative of advanced SMP operating systems (though > > linux also uses a giant kernel lock). > > One per CPU? One lock total? How many copies of the kernel are running > at once? How is caching handled, per cpu, or globally (forcing the two > cpu's to look like one?) When a system call executes, what handles it, > one cpu? Think of processors as schedulable resources. Only one processor is allowed to be in the kernel at a time. It's SMP in that: o Any CPU can service an interrupt, fault, or system call o All CPUs can run at the same time as long only one of them is in the kernel So it's symmetric, but it is far from the fine grain parallelism that would be necessary to get the most out of additional processors. It is probably reaching it's break point at 4 (or even 3) processors, where it's no longer valuable to add more processors. Compare this to Sequent's Dynix, which scales well to 32 processors, or Thinking Machines, which scales into the thousands of processors. Heck, the FS code is basically non-reentrant; UnixWare 2.x got a 60% improvement in the the UFS (FFS) performance just by allowing kernel preemption and FS reentrancy. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 10:37:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA26672 for smp-outgoing; Mon, 28 Apr 1997 10:37:15 -0700 (PDT) Received: from spinner.DIALix.COM (root@spinner.dialix.com [192.203.228.67]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA26662 for ; Mon, 28 Apr 1997 10:36:55 -0700 (PDT) Received: from spinner.DIALix.COM (peter@localhost.DIALix.oz.au [127.0.0.1]) by spinner.DIALix.COM with ESMTP id BAA21307; Tue, 29 Apr 1997 01:32:14 +0800 (WST) Message-Id: <199704281732.BAA21307@spinner.DIALix.COM> X-Mailer: exmh version 2.0gamma 1/27/96 To: Terry Lambert cc: black@zen.cypher.net (Ben Black), kory@avatar.com, brownie@earthling.net, csubl@csv.warwick.ac.uk, smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Mon, 28 Apr 1997 10:17:05 MST." <199704281717.KAA02093@phaeton.artisoft.com> Date: Tue, 29 Apr 1997 01:32:13 +0800 From: Peter Wemm Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Terry Lambert wrote: [..] > Personally, I'd like a 32 processor machine; of course, that means > fixing the kernel so that it doesn't bog down on interprocessor > synchronization, starting with the allocation mechanism and a > hierarchical lock manager. > > My goal would be to have incremental compiles going on in edit sessions > so that when I was done editing, the code was read to run. I hate this > "wait for the compile" BS. Hmm... marry emacs and gcc together, so that c-mode has some real meaning.. :-) The editor/compiler could keep a nice big parse tree in memory, being updated as you change lines. The editor could flag truely invalid lines because the incremental parser would see it. It'd be a "simple" matter of telling it when you wanted the output produced since the majority of the compile is already done. Argh, emacs, cpp, gcc and binutils merged into a monolith.. what a horrid thought... :-] > Terry Lambert > terry@lambert.org Cheers, -Peter From owner-freebsd-smp Mon Apr 28 10:44:05 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27098 for smp-outgoing; Mon, 28 Apr 1997 10:44:05 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id KAA27031 for ; Mon, 28 Apr 1997 10:43:57 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02151; Mon, 28 Apr 1997 10:41:47 -0700 From: Terry Lambert Message-Id: <199704281741.KAA02151@phaeton.artisoft.com> Subject: Re: SMP To: ccsanady@nyx.pr.mcs.net (Chris Csanady) Date: Mon, 28 Apr 1997 10:41:47 -0700 (MST) Cc: black@zen.cypher.net, chuckr@mat.net, FreeBSD-SMP@FreeBSD.org In-Reply-To: <199704280357.WAA12920@nyx.pr.mcs.net> from "Chris Csanady" at Apr 27, 97 10:57:12 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-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > >freebsd-smp is not the best example of how to do SMP. it uses the > >simplest method: one giant kernel lock. i don't know that it is > >particularly representative of advanced SMP operating systems (though > >linux also uses a giant kernel lock). > > Actually, linux has moved to a slightly finer grain system. Now they > have seperate locks for the run queues, scheduler, and some other > things.. These are isolated subsystems. They don't ever reenter code on multiple CPU's simultaneously. This is not much of a symmetry win; it also isn't a scalable win if they don't place the locks in a hierarchical relationship. Without that change, they are subject to deadly embrace deadlocks if they get any more complex in their locking structure. The correct approach to incremental improvement is "push down". You define locking primitives that assume transitive closure over all the other locks in the kernel (ie: that there is a hierarchical relationship and intention modes), and initially run them all off the single lock. Then you "push them down" into the kernel subsystems, "through" the system call interface. The important thing to note is that you lock *subsystems*, not *datum* at this point. Every time you descend the hierarchy completely, you free the top level lock to be nothing but an intention mode holder, and you prerun Warshal's to the n-1 level in the tree to make conflict calculation trivial and nearly instantaneous. Eventually, the terminal locks all lock datum, not subsystems, and you can use non-SMP aware subsystem components (like protocol stacks or FS modules) by locking the non-terminal locks... but you want to use SMP aware code where possible. This is different from the Solaris approach, where they lock subsystems nearly exclusively, and functionally encapsulate the datum. For one thing, the Solaris stuff is only scalable to 4-8 CPU's before the interprocessor contention makes it worthless to add more CPU's. All you have to do is look at the VFS code to see the scalability problems in Solaris. SVR4 is not any better (exception: the Unisys 6000/50 SMP SVR4.0.2, independently developed by Unisys, did FS locking the right way). Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 10:44:37 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27163 for smp-outgoing; Mon, 28 Apr 1997 10:44:37 -0700 (PDT) Received: from uhf.wdc.net (uhf.wdc.net [198.147.74.44]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA27149 for ; Mon, 28 Apr 1997 10:44:33 -0700 (PDT) Received: from localhost (bad@localhost) by uhf.wdc.net (8.8.5/8.6.12) with SMTP id NAA01788; Mon, 28 Apr 1997 13:44:13 -0400 (EDT) Date: Mon, 28 Apr 1997 13:44:12 -0400 (EDT) From: Bernie Doehner To: Terry Lambert cc: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704281717.KAA02093@phaeton.artisoft.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > The PPro's are stackable without additional electronics external > to the processor. This makes it a *lot* easier to do multiple > CPU's up to large numbers. Didn't anyone ever notice those neat traces between the CPUs and other devices on the MB? The higher and higher you go in frequency, the shorter and shorter a wavelength becomes and the more and more these traces start to look like antennas. There's a physical limit as to how long these traces can be and this limit also affects how closely you can space CPUs, without setting up a CPU hierarchy. > Personally, I'd like a 32 processor machine; of course, that means > fixing the kernel so that it doesn't bog down on interprocessor > synchronization, starting with the allocation mechanism and a > hierarchical lock manager. And finding a way to build a 3-D motherboard. > My goal would be to have incremental compiles going on in edit sessions > so that when I was done editing, the code was read to run. I hate this > "wait for the compile" BS. > Would that be nice? :) Bernie From owner-freebsd-smp Mon Apr 28 10:45:43 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27213 for smp-outgoing; Mon, 28 Apr 1997 10:45:43 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id KAA27208 for ; Mon, 28 Apr 1997 10:45:40 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02167; Mon, 28 Apr 1997 10:44:21 -0700 From: Terry Lambert Message-Id: <199704281744.KAA02167@phaeton.artisoft.com> Subject: Re: SMP To: chuckr@mat.net (Chuck Robey) Date: Mon, 28 Apr 1997 10:44:21 -0700 (MST) Cc: black@zen.cypher.net, FreeBSD-SMP@FreeBSD.org In-Reply-To: from "Chuck Robey" at Apr 27, 97 11:58:37 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-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > I'm confused, then. If there's only one kernel, then only one cpu can > run it, so only one cpu can field the system calls. If both cpu's can > field system calls, then unless they contact the other one to get the > work done, then there must be two copies of the kernel ruinning, right? > > I'm probably misunderstanding something. Maybe you meant only one piece > of software called "kernel" but two cpus running it? The lock is acquired at system call time. Only one CPU can acquire the lock at a time, so the kernel is not reentered. But any CPU may acquire the lock. Effectively, there is a single kernel which can be run by only one CPU at a time. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 10:52:19 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27588 for smp-outgoing; Mon, 28 Apr 1997 10:52:19 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id KAA27581 for ; Mon, 28 Apr 1997 10:52:15 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02220; Mon, 28 Apr 1997 10:50:21 -0700 From: Terry Lambert Message-Id: <199704281750.KAA02220@phaeton.artisoft.com> Subject: Re: SMP To: ccsanady@nyx.pr.mcs.net (Chris Csanady) Date: Mon, 28 Apr 1997 10:50:21 -0700 (MST) Cc: black@zen.cypher.net, chuckr@mat.net, FreeBSD-SMP@FreeBSD.org In-Reply-To: <199704280416.XAA12986@nyx.pr.mcs.net> from "Chris Csanady" at Apr 27, 97 11:16:05 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-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > >i sit corrected. i expect they will eventually migrate to a fully > >threaded kernel. > > As will we I hope. I was hoping to work on pushing the locks down > into the syscalls earlier, but I ran into some trouble. I really > knew very little about assembly, and our locks really are not up > to it yet. :( > > Besides, the general concencus was that we didn't want to deal with > it now.. So fake it. Define per subsystem locks that, in implementation, access the global lock. The only time you have to worry about reentrancy is when you are accessing a common object. The first stage of the push-down is to access the global lock at multiple choke-points in each subsystem rather than at the single choke-point in the trap code. Things like "getpid" which can't sleep pending a resource don't need locks at all, at that point. The getppid is a bit more of a problem, if one thread of a process is disassociating (becoing a child of init) at the time another thread is calling; so you need to be careful with your assumptions to not screw future multithreading. Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 10:54:32 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27660 for smp-outgoing; Mon, 28 Apr 1997 10:54:32 -0700 (PDT) Received: from news.quick.net (donegan@news.quick.net [207.212.170.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA27654 for ; Mon, 28 Apr 1997 10:54:30 -0700 (PDT) Received: (from donegan@localhost) by news.quick.net (8.8.5/8.6.9) id KAA17544; Mon, 28 Apr 1997 10:54:19 -0700 (PDT) Date: Mon, 28 Apr 1997 10:54:19 -0700 (PDT) From: "Steven P. Donegan" To: freebsd-smp@freebsd.org Subject: CVS Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk After upgrading to 3.0-CURRENT yesterday assuming (incorrectly) that SMP had been merged into the 3.0-CURRENT dist I am again in need of the CVSup code - ftp to freefall refuses to accept anon and the CVS files put in place on the standard archive also refuse access. I promise I'll place a copy where I can access it so I don't bother folks with this simple problem again :-) If there is a 3.0-CURRENT with integrated SMP please point the way. If there is an accessable CVSup.14.whatever tar available somewhere please also point the way. Thanks... Steven P. Donegan donegan@quick.net From owner-freebsd-smp Mon Apr 28 10:55:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27699 for smp-outgoing; Mon, 28 Apr 1997 10:55:55 -0700 (PDT) Received: from mx.serv.net (mx.serv.net [199.201.191.10]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA27690 for ; Mon, 28 Apr 1997 10:55:52 -0700 (PDT) Received: from MindBender.serv.net by mx.serv.net (8.7.5/SERV Revision: 2.30) id KAA13927; Mon, 28 Apr 1997 10:55:31 -0700 (PDT) Received: from localhost.HeadCandy.com (michaelv@localhost.HeadCandy.com [127.0.0.1]) by MindBender.serv.net (8.7.5/8.7.3) with SMTP id KAA28971; Mon, 28 Apr 1997 10:54:07 -0700 (PDT) Message-Id: <199704281754.KAA28971@MindBender.serv.net> X-Authentication-Warning: MindBender.serv.net: Host michaelv@localhost.HeadCandy.com [127.0.0.1] didn't use HELO protocol To: Narvi cc: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of Mon, 28 Apr 97 12:04:34 +0300. Date: Mon, 28 Apr 1997 10:53:55 -0700 From: "Michael L. VanLoon -- HeadCandy.com" Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >> >a 4 or 6 CPU P6 board for *other* than a large server...think about that >> >for a few minutes. >> >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most >> >popular commercial SMP-capable OS) >> Bullshit. >Well, at least 2 of the 2 NT workstations installed from 2 different CD-s >claim they are capable of upto 2 proc-s. I haven't also seen in any place >a reference that says otherwise. That's because your _license_ restricts you to no more than two processors. I've seen NT run on up to 8-way Pentium and Pentium Pro, and DEC Alpha, boxes. We use 4-way Alphas and Compaq's all over the place at work. There's nothing inherent in the design of NT that restricts it even to 8-way. It's just that's the most I've seen it run on, and NT isn't really designed to run optimally on massively parallel designs. Sequent, for example, has some proprietary patches to the NT kernel which enable it to run efficiently on their up-to-32 processor (as far as I know up to 32 -- I'm not a Sequent expert) Intel-based "mini-main-frames". This coupled with the fact that, as far as I know (and it's not very far), there isn't a standard for multi-processor design, for Intel chips, that goes beyond four processors. So anything bigger would be somewhat proprietary and require specific support code. ----------------------------------------------------------------------------- Michael L. VanLoon michaelv@MindBender.serv.net --< Free your mind and your machine -- NetBSD free un*x >-- NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... NetBSD ports in progress: PICA, others... ----------------------------------------------------------------------------- From owner-freebsd-smp Mon Apr 28 10:56:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA27770 for smp-outgoing; Mon, 28 Apr 1997 10:56:29 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id KAA27761 for ; Mon, 28 Apr 1997 10:56:25 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02242; Mon, 28 Apr 1997 10:53:58 -0700 From: Terry Lambert Message-Id: <199704281753.KAA02242@phaeton.artisoft.com> Subject: Re: Quad Pro 150 motherboard? To: black@zen.cypher.net (Ben Black) Date: Mon, 28 Apr 1997 10:53:58 -0700 (MST) Cc: michaelv@MindBender.serv.net, csubl@csv.warwick.ac.uk, smp@freebsd.org In-Reply-To: from "Ben Black" at Apr 28, 97 03:48:49 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > >popular commercial SMP-capable OS) > > > > Bullshit. > > > > perhaps you'd like to clarify. NT *server* can handle more CPUs, but NT > *workstation* is limited to 2. if you have a microsoft URL i can peruse > to dispute that, i'd be happy to admit my mistake. but i doubt you can > provide such a thing. Maybe he was disputing the claim that it's the "most popular commercial SMP-capable OS". 8-) 8-). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 11:03:31 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28141 for smp-outgoing; Mon, 28 Apr 1997 11:03:31 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28131 for ; Mon, 28 Apr 1997 11:03:26 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id OAA23966; Mon, 28 Apr 1997 14:01:02 -0400 Date: Mon, 28 Apr 1997 14:01:00 -0400 (EDT) From: Ben Black To: Terry Lambert cc: michaelv@MindBender.serv.net, csubl@csv.warwick.ac.uk, smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704281753.KAA02242@phaeton.artisoft.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Terry Lambert wrote: > > > >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > >popular commercial SMP-capable OS) > > > > > > Bullshit. > > > > > > > perhaps you'd like to clarify. NT *server* can handle more CPUs, but NT > > *workstation* is limited to 2. if you have a microsoft URL i can peruse > > to dispute that, i'd be happy to admit my mistake. but i doubt you can > > provide such a thing. > > Maybe he was disputing the claim that it's the "most popular commercial > SMP-capable OS". > > 8-) 8-). > then perhaps i should clarify. we are on a freebsd list discussing SMP x86 boards. i was referring to operating systems that run on those boards. certainly there are more Solaris/SPARC installs out there than NT, but last i checked Solaris/SPARc didn't run very well on my P6. b3n From owner-freebsd-smp Mon Apr 28 11:04:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28272 for smp-outgoing; Mon, 28 Apr 1997 11:04:41 -0700 (PDT) Received: from critter.dk.tfs.com (phk.cybercity.dk [195.8.129.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28243; Mon, 28 Apr 1997 11:04:18 -0700 (PDT) Received: from critter (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.5/8.8.5) with ESMTP id UAA00293; Mon, 28 Apr 1997 20:03:12 +0200 (CEST) To: Terry Lambert cc: black@zen.cypher.net (Ben Black), kory@avatar.com, brownie@earthling.net, csubl@csv.warwick.ac.uk, smp@freebsd.org From: Poul-Henning Kamp Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Mon, 28 Apr 1997 10:17:05 PDT." <199704281717.KAA02093@phaeton.artisoft.com> Date: Mon, 28 Apr 1997 20:03:12 +0200 Message-ID: <291.862250592@critter> Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In message <199704281717.KAA02093@phaeton.artisoft.com>, Terry Lambert writes: >The PPro's are stackable without additional electronics external >to the processor. This makes it a *lot* easier to do multiple >CPU's up to large numbers. The original design was for up to four PP's to be stackable. Unfortunately it transpired to not be the case (for reasons for signal integrity I belive, and therefore 4*PP cards have a number of buffers more than 2*PP cards. >My goal would be to have incremental compiles going on in edit sessions >so that when I was done editing, the code was read to run. I hate this >"wait for the compile" BS. Ever seen or heard about the TeX WYSIWYG computer built with InMos transputers ? :-) -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail. From owner-freebsd-smp Mon Apr 28 11:05:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28374 for smp-outgoing; Mon, 28 Apr 1997 11:05:49 -0700 (PDT) Received: from critter.dk.tfs.com (phk.cybercity.dk [195.8.129.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28354; Mon, 28 Apr 1997 11:05:37 -0700 (PDT) Received: from critter (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.5/8.8.5) with ESMTP id UAA00310; Mon, 28 Apr 1997 20:04:34 +0200 (CEST) To: Bernie Doehner cc: Terry Lambert , smp@freebsd.org From: Poul-Henning Kamp Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Mon, 28 Apr 1997 13:44:12 EDT." Date: Mon, 28 Apr 1997 20:04:34 +0200 Message-ID: <308.862250674@critter> Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In message , Bernie Doehne r writes: >Didn't anyone ever notice those neat traces between the CPUs and other >devices on the MB? The higher and higher you go in frequency, the shorter >and shorter a wavelength becomes and the more and more these traces start >to look like antennas. There's a physical limit as to how long these >traces can be and this limit also affects how closely you can space CPUs, >without setting up a CPU hierarchy. Uhm, we're about an order of magnitude from that point yet... -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail. From owner-freebsd-smp Mon Apr 28 11:08:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28610 for smp-outgoing; Mon, 28 Apr 1997 11:08:48 -0700 (PDT) Received: from critter.dk.tfs.com (phk.cybercity.dk [195.8.129.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28594; Mon, 28 Apr 1997 11:08:30 -0700 (PDT) Received: from critter (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.5/8.8.5) with ESMTP id UAA00342; Mon, 28 Apr 1997 20:07:21 +0200 (CEST) To: "Michael L. VanLoon -- HeadCandy.com" cc: Narvi , smp@freebsd.org From: Poul-Henning Kamp Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Mon, 28 Apr 1997 10:53:55 PDT." <199704281754.KAA28971@MindBender.serv.net> Date: Mon, 28 Apr 1997 20:07:20 +0200 Message-ID: <340.862250840@critter> Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In message <199704281754.KAA28971@MindBender.serv.net>, "Michael L. VanLoon -- HeadCandy.com" writes: >There's nothing inherent in the design of NT that restricts it even to >8-way. Uhm, no. The limit is 16. This is because certain stuff is kept as two bits per CPU in one word of memory. Minor recompiling would fix this limit though. Anybody know what Intel/DOE will run on the 9144*PPro machine they're building ? -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail. From owner-freebsd-smp Mon Apr 28 11:10:50 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28755 for smp-outgoing; Mon, 28 Apr 1997 11:10:50 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA28747 for ; Mon, 28 Apr 1997 11:10:46 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA02308; Mon, 28 Apr 1997 11:05:51 -0700 From: Terry Lambert Message-Id: <199704281805.LAA02308@phaeton.artisoft.com> Subject: Re: Quad Pro 150 motherboard? To: peter@spinner.dialix.com (Peter Wemm) Date: Mon, 28 Apr 1997 11:05:51 -0700 (MST) Cc: terry@lambert.org, black@zen.cypher.net, kory@avatar.com, brownie@earthling.net, csubl@csv.warwick.ac.uk, smp@freebsd.org In-Reply-To: <199704281732.BAA21307@spinner.DIALix.COM> from "Peter Wemm" at Apr 29, 97 01:32:13 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > My goal would be to have incremental compiles going on in edit sessions > > so that when I was done editing, the code was read to run. I hate this > > "wait for the compile" BS. > > Hmm... marry emacs and gcc together, so that c-mode has some real > meaning.. :-) The editor/compiler could keep a nice big parse tree in > memory, being updated as you change lines. The editor could flag truely > invalid lines because the incremental parser would see it. It'd be a > "simple" matter of telling it when you wanted the output produced since > the majority of the compile is already done. > > Argh, emacs, cpp, gcc and binutils merged into a monolith.. what a horrid > thought... :-] How about using 'vi' instead? Actually, I was thinking merging cscope with a compiler would be the more useful way to do it... Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 11:11:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28813 for smp-outgoing; Mon, 28 Apr 1997 11:11:33 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA28801 for ; Mon, 28 Apr 1997 11:11:29 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA02356; Mon, 28 Apr 1997 11:10:08 -0700 From: Terry Lambert Message-Id: <199704281810.LAA02356@phaeton.artisoft.com> Subject: Re: Quad Pro 150 motherboard? To: phk@dk.tfs.com (Poul-Henning Kamp) Date: Mon, 28 Apr 1997 11:10:08 -0700 (MST) Cc: bad@uhf.wireless.net, terry@lambert.org, smp@freebsd.org In-Reply-To: <308.862250674@critter> from "Poul-Henning Kamp" at Apr 28, 97 08:04:34 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-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > >Didn't anyone ever notice those neat traces between the CPUs and other > >devices on the MB? The higher and higher you go in frequency, the shorter > >and shorter a wavelength becomes and the more and more these traces start > >to look like antennas. There's a physical limit as to how long these > >traces can be and this limit also affects how closely you can space CPUs, > >without setting up a CPU hierarchy. > > Uhm, we're about an order of magnitude from that point yet... Weber State University has a 32 processor Sequent box that would love to be able to run a free OS... Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Mon Apr 28 11:13:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA28867 for smp-outgoing; Mon, 28 Apr 1997 11:13:08 -0700 (PDT) Received: from zibbi.mikom.csir.co.za (zibbi.mikom.csir.co.za [146.64.24.58]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28862 for ; Mon, 28 Apr 1997 11:13:04 -0700 (PDT) Received: (from jhay@localhost) by zibbi.mikom.csir.co.za (8.8.5/8.8.5) id UAA00694; Mon, 28 Apr 1997 20:12:06 +0200 (SAT) From: John Hay Message-Id: <199704281812.UAA00694@zibbi.mikom.csir.co.za> Subject: Re: CVS In-Reply-To: from "Steven P. Donegan" at "Apr 28, 97 10:54:19 am" To: donegan@quick.net (Steven P. Donegan) Date: Mon, 28 Apr 1997 20:12:06 +0200 (SAT) Cc: 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-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > After upgrading to 3.0-CURRENT yesterday assuming (incorrectly) that SMP > had been merged into the 3.0-CURRENT dist I am again in need of the CVSup > code - ftp to freefall refuses to accept anon and the CVS files put in > place on the standard archive also refuse access. I promise I'll place a > copy where I can access it so I don't bother folks with this simple > problem again :-) > > If there is a 3.0-CURRENT with integrated SMP please point the way. If > there is an accessable CVSup.14.whatever tar available somewhere please > also point the way. > Look on hub.freebsd.org/pub/CVSup (it's the old freefall). The SMP code is in -CURRENT. Here is a small piece of sys/i386/conf/LINT ===== # SMP OPTIONS: # # SMP enables building of a Symmetric MultiProcessor Kernel. # APIC_IO enables the use of the IO APIC for Symmetric I/O. # NCPU sets the number of CPUs, defaults to 2. # NBUS sets the number of busses, defaults to 4. # NAPIC sets the number of IO APICs on the motherboard, defaults to 1. # NINTR sets the total number of INTs provided by the motherboard. ===== John -- John Hay -- John.Hay@mikom.csir.co.za From owner-freebsd-smp Mon Apr 28 11:18:14 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA29071 for smp-outgoing; Mon, 28 Apr 1997 11:18:14 -0700 (PDT) Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA28977 for ; Mon, 28 Apr 1997 11:16:52 -0700 (PDT) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.8.4/8.8.4) with SMTP id VAA23262; Mon, 28 Apr 1997 21:21:08 +0300 (EEST) Date: Mon, 28 Apr 1997 21:21:06 +0300 (EEST) From: Narvi To: Terry Lambert cc: Chuck Robey , black@zen.cypher.net, FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: <199704281744.KAA02167@phaeton.artisoft.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Terry Lambert wrote: > > I'm confused, then. If there's only one kernel, then only one cpu can > > run it, so only one cpu can field the system calls. If both cpu's can > > field system calls, then unless they contact the other one to get the > > work done, then there must be two copies of the kernel ruinning, right? > > > > I'm probably misunderstanding something. Maybe you meant only one piece > > of software called "kernel" but two cpus running it? > > The lock is acquired at system call time. Only one CPU can acquire the > lock at a time, so the kernel is not reentered. But any CPU may > acquire the lock. > > Effectively, there is a single kernel which can be run by only one > CPU at a time. > NB! I am not a SMP guru (and have never claimed to be one). Just an idea, maybe just another wheel(TM). Couldn't we divide the kernel (and anything "provided" by it into) two (at least initially): 1) client - accepts syscalls, translates, etc. and passes the result on to 2 2) kernel server - does all the low level stuff We could then thread 1) and 2). Each thread in 2 has it's own lock (initally, there is one lock on the whole of 2). If one thread in 2 needs to use a service provided by another, it blocks on it. There is two types of blocking, soft and hard - on soft blocking, we mark the thread as soft-blocked and allocated new of it's type as needed. Hard blocking is blocking we can't help :-( like hardware driver related blocking. There is a fixed number and when it blocks things just qeue up. We each entry to 2 is accociated a request qeue - actually a graph, so that if we discover there were two writes to the same disk block we write only the last, saving one disk access. So we could have a processor for each disk interface card (scsi, ide, etc), each network driver, and some more. Well, in practice it moight not work out like that :-( Sander > > Terry Lambert > terry@lambert.org > --- > Any opinions in this posting are my own and not those of my present > or previous employers. > From owner-freebsd-smp Mon Apr 28 11:21:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA29219 for smp-outgoing; Mon, 28 Apr 1997 11:21:28 -0700 (PDT) Received: from caipfs.rutgers.edu (root@caipfs.rutgers.edu [128.6.155.100]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA29141 for ; Mon, 28 Apr 1997 11:19:50 -0700 (PDT) Received: from jenolan.caipgeneral (jenolan.rutgers.edu [128.6.111.5]) by caipfs.rutgers.edu (8.8.5/8.8.5) with SMTP id OAA25339; Mon, 28 Apr 1997 14:16:46 -0400 (EDT) Received: by jenolan.caipgeneral (SMI-8.6/SMI-SVR4) id OAA00308; Mon, 28 Apr 1997 14:15:28 -0400 Date: Mon, 28 Apr 1997 14:15:28 -0400 Message-Id: <199704281815.OAA00308@jenolan.caipgeneral> From: "David S. Miller" To: terry@lambert.org CC: ccsanady@nyx.pr.mcs.net, black@zen.cypher.net, chuckr@mat.net, FreeBSD-SMP@FreeBSD.org In-reply-to: <199704281741.KAA02151@phaeton.artisoft.com> (message from Terry Lambert on Mon, 28 Apr 1997 10:41:47 -0700 (MST)) Subject: Re: SMP Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk From: Terry Lambert Date: Mon, 28 Apr 1997 10:41:47 -0700 (MST) These are isolated subsystems. They don't ever reenter code on multiple CPU's simultaneously. Yes we most certainly do re-enter code for numerous things, currently: 1) System calls which modify process signal disposition 2) Signal delivery itself 3) sys_wait4() 4) A huge slew of "easy" system calls (getpid(), sys_times(), these sorts of things) Are all multi-threaded and can all run on all processors simultaneously. Just as a quick example on my 3 processor development Sparc machine. ? ./lat_sig lat_sig Signal handler installation: 4.993 microseconds Signal handler overhead: 37.701 microseconds Protection fault: 59.407 microseconds Ok, now this: ? Signal handler installation: 3.933 microseconds Signal handler installation: 3.835 microseconds Signal handler installation: 3.730 microseconds Signal handler overhead: 46.994 microseconds Signal handler overhead: 44.269 microseconds Signal handler overhead: 46.942 microseconds Protection fault: 78.210 microseconds Protection fault: 96.422 microseconds Protection fault: 74.614 microseconds While not entirely obvious, it does at least appear that the signal specific parts of this benchmark run about the same with one invocation as 3 do in parallel. The fault cost is noticably more for the multiple invocation, showing how the master lock is still necessary for the mm/vfs subsystems. We are working on threading those as well as the networking right now. Also since the scheduler is completely SMP safe, heavily user bound processes can run in and out of the scheduler without the master lock. We also have SMP safe wait queues already, and have solved the lost wakeup problem even for SMP already... Furthermore our interrupt handling subsystem is completely SMP safe, interrupt handlers can run in parallel on all processors at once. The correct approach to incremental improvement is "push down". You define locking primitives that assume transitive closure over all the other locks in the kernel (ie: that there is a hierarchical relationship and intention modes), and initially run them all off the single lock. Yes Terry, we all know about your transitive closures... ---------------------------------------------//// Yow! 11.26 MB/s remote host TCP bandwidth & //// 199 usec remote TCP latency over 100Mb/s //// ethernet. Beat that! //// -----------------------------------------////__________ o David S. Miller, davem@caip.rutgers.edu /_____________/ / // /_/ >< From owner-freebsd-smp Mon Apr 28 11:25:01 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA29385 for smp-outgoing; Mon, 28 Apr 1997 11:25:01 -0700 (PDT) Received: from cypher.net (black@zen.pratt.edu [205.232.115.155]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA29375 for ; Mon, 28 Apr 1997 11:24:58 -0700 (PDT) Received: (from black@localhost) by cypher.net (8.8.5/8.7.1) id OAA24443; Mon, 28 Apr 1997 14:22:12 -0400 Date: Mon, 28 Apr 1997 14:22:12 -0400 (EDT) From: Ben Black To: "Michael L. VanLoon -- HeadCandy.com" cc: Narvi , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-Reply-To: <199704281754.KAA28971@MindBender.serv.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk yup, the license restricts you. NT4WKS refuses to run on more than 2 CPUs. does that change a damn thing? nope. note that i didn't say anything about NT4SRVR which *can* handle more than 2 because M$ wills it so. the fact is, when a machine is running NT on more than 4 CPUs it is not running a stock NT. my point remains valid: the most popular commercial, SMP, x86 OS is NT Workstation and you cannot make that run on 4 CPUs, hence another reason there is little demand for more than dual CPU boards. b3n On Mon, 28 Apr 1997, Michael L. VanLoon -- HeadCandy.com wrote: > > >> >a 4 or 6 CPU P6 board for *other* than a large server...think about that > >> >for a few minutes. > >> >(hint: NT4 Workstation can't handle more than 2CPUs and that is the most > >> >popular commercial SMP-capable OS) > > >> Bullshit. > > >Well, at least 2 of the 2 NT workstations installed from 2 different CD-s > >claim they are capable of upto 2 proc-s. I haven't also seen in any place > >a reference that says otherwise. > > That's because your _license_ restricts you to no more than two > processors. > > I've seen NT run on up to 8-way Pentium and Pentium Pro, and DEC > Alpha, boxes. We use 4-way Alphas and Compaq's all over the place at > work. > > There's nothing inherent in the design of NT that restricts it even to > 8-way. It's just that's the most I've seen it run on, and NT isn't > really designed to run optimally on massively parallel designs. > Sequent, for example, has some proprietary patches to the NT kernel > which enable it to run efficiently on their up-to-32 processor (as far > as I know up to 32 -- I'm not a Sequent expert) Intel-based > "mini-main-frames". > > This coupled with the fact that, as far as I know (and it's not very > far), there isn't a standard for multi-processor design, for Intel > chips, that goes beyond four processors. So anything bigger would be > somewhat proprietary and require specific support code. > > ----------------------------------------------------------------------------- > Michael L. VanLoon michaelv@MindBender.serv.net > --< Free your mind and your machine -- NetBSD free un*x >-- > NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, > Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... > NetBSD ports in progress: PICA, others... > ----------------------------------------------------------------------------- > From owner-freebsd-smp Mon Apr 28 11:30:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA29742 for smp-outgoing; Mon, 28 Apr 1997 11:30:15 -0700 (PDT) Received: from caipfs.rutgers.edu (root@caipfs.rutgers.edu [128.6.155.100]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA29735 for ; Mon, 28 Apr 1997 11:30:13 -0700 (PDT) Received: from jenolan.caipgeneral (jenolan.rutgers.edu [128.6.111.5]) by caipfs.rutgers.edu (8.8.5/8.8.5) with SMTP id OAA25600; Mon, 28 Apr 1997 14:23:51 -0400 (EDT) Received: by jenolan.caipgeneral (SMI-8.6/SMI-SVR4) id OAA00328; Mon, 28 Apr 1997 14:22:33 -0400 Date: Mon, 28 Apr 1997 14:22:33 -0400 Message-Id: <199704281822.OAA00328@jenolan.caipgeneral> From: "David S. Miller" To: terry@lambert.org CC: ccsanady@nyx.pr.mcs.net, black@zen.cypher.net, chuckr@mat.net, FreeBSD-SMP@FreeBSD.org In-reply-to: <199704281750.KAA02220@phaeton.artisoft.com> (message from Terry Lambert on Mon, 28 Apr 1997 10:50:21 -0700 (MST)) Subject: Re: SMP Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk From: Terry Lambert Date: Mon, 28 Apr 1997 10:50:21 -0700 (MST) The getppid is a bit more of a problem, if one thread of a process is disassociating (becoing a child of init) at the time another thread is calling; so you need to be careful with your assumptions to not screw future multithreading. getppid() can be done with no locking... From owner-freebsd-smp Mon Apr 28 11:32:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA29888 for smp-outgoing; Mon, 28 Apr 1997 11:32:18 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA29879 for ; Mon, 28 Apr 1997 11:32:14 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id MAA26677; Mon, 28 Apr 1997 12:30:28 -0600 (MDT) Message-Id: <199704281830.MAA26677@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Terry Lambert cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-reply-to: Your message of "Mon, 28 Apr 1997 10:50:21 PDT." <199704281750.KAA02220@phaeton.artisoft.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 28 Apr 1997 12:30:28 -0600 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Hi, > So fake it. Define per subsystem locks that, in implementation, access > the global lock. > > The only time you have to worry about reentrancy is when you are accessing > a common object. > > The first stage of the push-down is to access the global lock at multiple > choke-points in each subsystem rather than at the single choke-point in > the trap code. Things like "getpid" which can't sleep pending a resource > don't need locks at all, at that point. The getppid is a bit more of a > problem, if one thread of a process is disassociating (becoing a child of > init) at the time another thread is calling; so you need to be careful > with your assumptions to not screw future multithreading. it is definately time to start designing this area for real. we will undoubtedly have many lively discussions about lock models and strategies, but one thing that should preceed this is a "survey" of the current kernel to identify the critical regions. Once (and IMHO only once) we have a complete "map" that we can stand back and look at, will we be in a position to start design. If we just jump in and say "separate locks for a, b, & c" we are doomed to waste valuable time debuging dead-locks and crashes. such a strategy might work for the first 50% of improvement, but that last 50%... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Mon Apr 28 11:35:44 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA00270 for smp-outgoing; Mon, 28 Apr 1997 11:35:44 -0700 (PDT) Received: from mx.serv.net (mx.serv.net [199.201.191.10]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA00265 for ; Mon, 28 Apr 1997 11:35:41 -0700 (PDT) Received: from MindBender.serv.net by mx.serv.net (8.7.5/SERV Revision: 2.30) id LAA15819; Mon, 28 Apr 1997 11:35:41 -0700 (PDT) Received: from localhost.HeadCandy.com (michaelv@localhost.HeadCandy.com [127.0.0.1]) by MindBender.serv.net (8.7.5/8.7.3) with SMTP id LAA29188; Mon, 28 Apr 1997 11:34:12 -0700 (PDT) Message-Id: <199704281834.LAA29188@MindBender.serv.net> X-Authentication-Warning: MindBender.serv.net: Host michaelv@localhost.HeadCandy.com [127.0.0.1] didn't use HELO protocol To: Poul-Henning Kamp cc: Narvi , smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of Mon, 28 Apr 97 20:07:20 +0200. <340.862250840@critter> Date: Mon, 28 Apr 1997 11:33:56 -0700 From: "Michael L. VanLoon -- HeadCandy.com" Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >>There's nothing inherent in the design of NT that restricts it even to >>8-way. >Uhm, no. The limit is 16. This is because certain stuff is kept as Actually, the limit is 32, for NT Server 4.0 (some people are quoting NT 3.5 documentation...). But, that's just minor quibling. Since this isn't a FreeBSD issue, this thread should end... (And yes, I wasn't differentiating between NT Workstation and Server. Obviously Microsoft wants you to run Server for large SMP machines.) ----------------------------------------------------------------------------- Michael L. VanLoon michaelv@MindBender.serv.net --< Free your mind and your machine -- NetBSD free un*x >-- NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... NetBSD ports in progress: PICA, others... ----------------------------------------------------------------------------- From owner-freebsd-smp Mon Apr 28 12:05:59 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA02004 for smp-outgoing; Mon, 28 Apr 1997 12:05:59 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA01999 for ; Mon, 28 Apr 1997 12:05:56 -0700 (PDT) Received: from zwei.siemens.at (zwei.siemens.at [193.81.246.12]) by who.cdrom.com (8.8.5/8.6.11) with ESMTP id MAA02001 for ; Mon, 28 Apr 1997 12:05:52 -0700 (PDT) Received: from pc5829.hil.siemens.at (root@firix [10.1.143.100]) by zwei.siemens.at with ESMTP id VAA12178; Mon, 28 Apr 1997 21:03:09 +0200 (MET DST) Received: from localhost (mingo@localhost) by pc5829.hil.siemens.at (8.6.12/8.6.9) with SMTP id VAA16263; Mon, 28 Apr 1997 21:08:15 +0200 Date: Mon, 28 Apr 1997 21:08:15 +0200 (MET DST) From: Ingo Molnar To: Terry Lambert cc: Chris Csanady , black@zen.cypher.net, chuckr@mat.net, FreeBSD-SMP@freebsd.org Subject: Re: SMP In-Reply-To: <199704281741.KAA02151@phaeton.artisoft.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Terry Lambert wrote: > > Actually, linux has moved to a slightly finer grain system. Now they > > have seperate locks for the run queues, scheduler, and some other > > things.. > This is not much of a symmetry win; it also isn't a scalable win if > they don't place the locks in a hierarchical relationship. Without > that change, they are subject to deadly embrace deadlocks if they get > any more complex in their locking structure. what kind of deadlocks do you picture. We only have spinlocks in Linux and we disallow scheduling with held spinlocks. [thus you can think of those locks like atomic operations]. There isnt much to be done wrong there. Anything that gets out of the big kernel lock is done via totally parallel code and spinlocks after that. [ and resource locking is just done the same way like on uniprocessor Linux, on a per-object basis ] -- mingo From owner-freebsd-smp Mon Apr 28 12:43:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA04003 for smp-outgoing; Mon, 28 Apr 1997 12:43:17 -0700 (PDT) Received: from INET-05-IMC.microsoft.com (mail5.microsoft.com [131.107.3.31]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA03997 for ; Mon, 28 Apr 1997 12:43:15 -0700 (PDT) Received: by INET-05-IMC with Internet Mail Service (5.0.1458.14) id ; Mon, 28 Apr 1997 12:39:04 -0700 Message-ID: <7D06B4AA8B39D011A64900805F682CDA014AAB4E@RED-09-MSG.dns.microsoft.com> From: Arlie Davis To: "'mef@cs.washington.edu'" , smp@freebsd.org Subject: RE: Quad Pro 150 motherboard? Date: Mon, 28 Apr 1997 12:38:06 -0700 X-Priority: 3 X-Mailer: Internet Mail Service (5.0.1458.14) Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk As I said in a previous message, NT "itself" supports up to 32 processors. In other words, it was designed to scale to 32 processors. The retail version of NT Server is only _licensed_ for machines with up to four processors. The retail version of NT Workstation is only _licensed_ for machines with up to two processors. Yes, basically, it is just a software check to see if your machine / OS is licensed to use as many processors as you are offering. Requiring more licensing for more hardware (or users, or networks, or interfaces, or whatever) is a very common, accepted practice in this industry. And if you think it is simply gouging the poor, unwitting public, then you are not taking into account the very different development and support requirements for machines on different sides of the multiprocessor scale. Again, NTS and NTW both support up to 32 processors. You have to go through an OEM to do so. -- arlie > -----Original Message----- > From: mef@cs.washington.edu [SMTP:mef@cs.washington.edu] > Sent: Monday, April 28, 1997 9:26 AM > To: smp@freebsd.org > Subject: Re: Quad Pro 150 motherboard? > > Date: Mon, 28 Apr 1997 04:49:18 -0400 (EDT) > From: Ben Black > from > > http://www.microsoft.com/syspro/technet/boes/bo/winntas/technote/nt101 > .htm > Windows NT Workstation supports up to two processors in a symmetric > > multiprocessing environment. > > According to that document NT Workstation can only support two > processors, but NT Server can support four processors. I wouldn't be > surprised if embedded in the code there was some sort of check: > if (ms->makemoremoney->is_nt_server == TRUE) { > /* support four processors */ > } > > :) > > M From owner-freebsd-smp Mon Apr 28 12:44:39 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA04061 for smp-outgoing; Mon, 28 Apr 1997 12:44:39 -0700 (PDT) Received: from INET-01-IMC.microsoft.com (mail1.microsoft.com [131.107.3.41]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA04056 for ; Mon, 28 Apr 1997 12:44:37 -0700 (PDT) Received: by mail1.microsoft.com with Internet Mail Service (5.0.1458.14) id ; Mon, 28 Apr 1997 12:44:06 -0700 Message-ID: <7D06B4AA8B39D011A64900805F682CDA014AAB4F@RED-09-MSG.dns.microsoft.com> From: Arlie Davis To: "'Terry Lambert'" Cc: smp@freebsd.org Subject: RE: Quad Pro 150 motherboard? Date: Mon, 28 Apr 1997 12:40:59 -0700 X-Priority: 3 X-Mailer: Internet Mail Service (5.0.1458.14) Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Well, the PPros are stackable up to four processors without additional glue. I think Intel calls groups of four processors "clusters". SMP with more than one cluster requires a lot of additional glue. -- arlie > -----Original Message----- > From: Terry Lambert [SMTP:terry@lambert.org] > Sent: Monday, April 28, 1997 10:17 AM > To: black@zen.cypher.net > Cc: kory@avatar.com; brownie@earthling.net; csubl@csv.warwick.ac.uk; > smp@freebsd.org > Subject: Re: Quad Pro 150 motherboard? > > The PPro's are stackable without additional electronics external > to the processor. This makes it a *lot* easier to do multiple > CPU's up to large numbers. > > From owner-freebsd-smp Mon Apr 28 12:44:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA04067 for smp-outgoing; Mon, 28 Apr 1997 12:44:48 -0700 (PDT) Received: from INET-04-IMC.microsoft.com (mail4.microsoft.com [131.107.3.29]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA04062 for ; Mon, 28 Apr 1997 12:44:44 -0700 (PDT) Received: by INET-04-IMC with Internet Mail Service (5.0.1458.14) id ; Mon, 28 Apr 1997 12:35:28 -0700 Message-ID: <7D06B4AA8B39D011A64900805F682CDA014AAB4D@RED-09-MSG.dns.microsoft.com> From: Arlie Davis To: "'Narvi'" , "Michael L. VanLoon -- HeadCandy.com" Cc: smp@freebsd.org Subject: RE: Quad Pro 150 motherboard? Date: Mon, 28 Apr 1997 12:30:43 -0700 X-Priority: 3 X-Mailer: Internet Mail Service (5.0.1458.14) Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Any processor limitation in NT, under 32 processors, is purely a licensing issue. I believe the retail versions of NT Server and NT Workstation are both licensed up to some number of processors, and acquiring a license for a higher number of processors requires purchasing the software through an OEM vendor. The idea behind this is that machines with a large number of processors would serve customers best by having the OEM support the machine as a whole -- hardware and OS. This is NOT the official word Microsoft -- I'm just an employee. Here is a clip from http://www.microsoft.com/syspro/technet/boes/winnt/nt351/tech3000.htm : "Multiprocessor Versions Available from Hardware Vendors" "As mentioned, Windows NT Server will run on hardware with as many as 32 processors. It is important to note that the retail version of Windows NT Server is licensed to run on machines with up to four processors. This is not a limitation of the scalability of Windows NT Server-it is a licensing restriction of the retail package. Versions of Windows NT Server that run on more than four processors are available from OEMs who offer multiprocessor hardware systems (up to 32 processors). Windows NT Server is included as part of the base system on these larger machines and is fully supported by OEMs who work with Microsoft." I can't find the similar text for NT Workstation, but it's there. -- arlie > -----Original Message----- > From: Narvi [SMTP:narvi@haldjas.folklore.ee] > Sent: Monday, April 28, 1997 2:05 AM > To: Michael L. VanLoon -- HeadCandy.com > Cc: smp@freebsd.org > Subject: Re: Quad Pro 150 motherboard? > > Well, at least 2 of the 2 NT workstations installed from 2 different > CD-s > claim they are capable of upto 2 proc-s. I haven't also seen in any > place > a reference that says otherwise. > From owner-freebsd-smp Mon Apr 28 19:56:16 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA18512 for smp-outgoing; Mon, 28 Apr 1997 19:56:16 -0700 (PDT) Received: from lundin.abq.nm.us (lundin.abq.nm.us [198.59.115.228]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA18507 for ; Mon, 28 Apr 1997 19:56:12 -0700 (PDT) Received: (from aflundi@localhost) by lundin.abq.nm.us. (8.8.5/8.8.5) id UAA04324 for smp@freebsd.org; Mon, 28 Apr 1997 20:54:45 -0600 (MDT) Date: Mon, 28 Apr 1997 20:54:45 -0600 (MDT) From: Alan Lundin Message-Id: <199704290254.UAA04324@lundin.abq.nm.us.> In-Reply-To: Poul-Henning Kamp "Re: Quad Pro 150 motherboard?" (Apr 28, 8:07pm) X-Mailer: Mail User's Shell (7.2.5 10/14/92) To: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Apr 28, 8:07pm, Poul-Henning Kamp wrote: > Subject: Re: Quad Pro 150 motherboard? > > Anybody know what Intel/DOE will run on the 9144*PPro machine they're > building ? The machine is up and running (more or less) at Sandia National Labs with an OS by Barney Maccabe and his students, which is called PUMA, I think. Unfortunately, Barney is a Linux type, but I've been working on him! :-) --alan From owner-freebsd-smp Mon Apr 28 20:02:04 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id UAA18809 for smp-outgoing; Mon, 28 Apr 1997 20:02:04 -0700 (PDT) Received: from parkplace.cet.co.jp (parkplace.cet.co.jp [202.32.64.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA18801 for ; Mon, 28 Apr 1997 20:02:02 -0700 (PDT) Received: from localhost (michaelh@localhost) by parkplace.cet.co.jp (8.8.5/CET-v2.1) with SMTP id DAA05849; Tue, 29 Apr 1997 03:00:50 GMT Date: Tue, 29 Apr 1997 12:00:49 +0900 (JST) From: Michael Hancock To: Narvi cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Mon, 28 Apr 1997, Narvi wrote: > Couldn't we divide the kernel (and anything "provided" by it into) two (at > least initially): > > 1) client - accepts syscalls, translates, etc. and passes the result on to > 2 > 2) kernel server - does all the low level stuff We want SMP. Mike Hancock From owner-freebsd-smp Tue Apr 29 00:02:13 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA00581 for smp-outgoing; Tue, 29 Apr 1997 00:02:13 -0700 (PDT) Received: from haldjas.folklore.ee (Haldjas.folklore.ee [193.40.6.121]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA00336 for ; Mon, 28 Apr 1997 23:57:39 -0700 (PDT) Received: from localhost (narvi@localhost) by haldjas.folklore.ee (8.8.4/8.8.4) with SMTP id KAA00696; Tue, 29 Apr 1997 10:03:09 +0300 (EEST) Date: Tue, 29 Apr 1997 10:03:07 +0300 (EEST) From: Narvi To: Michael Hancock cc: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Tue, 29 Apr 1997, Michael Hancock wrote: > On Mon, 28 Apr 1997, Narvi wrote: > > > Couldn't we divide the kernel (and anything "provided" by it into) two (at > > least initially): > > > > 1) client - accepts syscalls, translates, etc. and passes the result on to > > 2 > > 2) kernel server - does all the low level stuff > > We want SMP. I take you didn't read it all carefully enough (especially the part you snipped). It would be SMP. Initially there would be two locks (= at most two processors in the kernel). In the progress of further threading the kernel the number would grow. As already said, both of the "layers" of the kernel would be threaded and there could be n (def. n=number of processors in the system) processors in the kernel, distributed between the two in whatever manner is needed on the moment. The "upper layer" of kernel also would not care in most cases what it was that the lower did. Lets take something simple, like fork: 1) process calls fork 2) we wait until we can get the syscall_entry lock (the process just sleeps). 3) we are in the syscall now: a) check whetever know syscall - return error if not b) call make_syscall_worker_thread(syscall_nr, params) c) release lock, sleep until someone wants to make another syscall 4) the syscall_worker thread (the current one, at any time there may be several) does what is needed (acquires locks, calls the "lower layer") and then returns, unblocking the process. Obviously eventually there could even be more than one syscall thread creating syscall_worker threads. I don't see how any part of it is not SMP. Sander > > Mike Hancock > > From owner-freebsd-smp Tue Apr 29 00:04:45 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA00753 for smp-outgoing; Tue, 29 Apr 1997 00:04:45 -0700 (PDT) Received: from jumping-spider.aracnet.com (root@jumping-spider.aracnet.com [204.188.47.14]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA00743 for ; Tue, 29 Apr 1997 00:04:40 -0700 (PDT) Received: from cbrowni2-home (ppp-r21.aracnet.com [205.238.13.23]) by jumping-spider.aracnet.com (8.8.5/8.8.5) with SMTP id AAA10734; Tue, 29 Apr 1997 00:05:14 -0700 Message-ID: <33658F3C.86D@earthling.net> Date: Mon, 28 Apr 1997 23:03:40 -0700 From: Chris Browning Reply-To: brownie@earthling.net X-Mailer: Mozilla 3.01 (WinNT; U) MIME-Version: 1.0 To: Ben Black CC: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Ben, > i think a good rul eof thumb is about 256K per 64MB RAM (except for > servers or where bus speeds far exceed RAM speeds). Hmm, I don't quite buy this. First off, what does cache size have to do with memory size? There may be some "general" corrilation in a uniprocessor system, but that is less of a concern in MP environments. The important factor is bus contention, since you have 2-4 (in PPP's case) high speed processors all fighting over the same bus. I believe it is more the case that you want your cache size to grow as a function of the # of processers and with a bus utilization factor thrown in there. Also, the bus speed in servers is typically the same as all other systems, so I don't get the comparison to RAM speeds. Commodity servers use the same RAM speed and bus speed as desktop machines. > so, 4 CPUs with 256K > cache should be fine in non-server configs. the 512K P6 chips are just Well, 4 CPUs IS currently a server configuration. As I mentioned before, all the >2way systems I know of are servers. Granted, these can also make kick-butt workstations, that is not what they were designed for. In addition, 256k is not enough cache for a PPP system for a 4 way system no matter what "configuration" it is. You will get a performance hit because of the bus contention. > not cost effective. last week i ordered a P6-150 for $165. tough to > beat that. That is a good deal. Of course, you loose about 10% of your bus bandwidth running the bus at 60MHz instead of 66MHz (which is what the 150 MHz runs at). > > On Sun, 27 Apr 1997, Chris Browning wrote: > > > Michael, > > I concur with Ben here. Typically in the PC world, anything above > > 2 procs is geared towards servers. I can't think of many non-server > > > 2way systems out there. So, if you want >2way, you will have > > to go with one of the server class machines, which are typically > > expensive. In addition, I would not bother with the 150MHz PPP. If > > I remember correctly, the 150 only comes in the 256k cache size. If > > you are going to do 4way, do yourself a favor and get the 512k cache > > PPP. 4way PPP will saturate the processor bus quite quickly, so the > > more cache the better. I believe they make a 166/512k PPP, so for > > cost effectiveness, that is what I would recommend. > > > > Chris > > Not speaking for Intel. > > Ben Black wrote: > > > > > > a 4 or 6 CPU P6 board for *other* than a large server...think about that > > > for a few minutes. > > > > > > (hint: NT4 Workstation can't handle more than 2CPUs and that is the most > > > popular commercial SMP-capable OS) > > > > > > On Mon, 28 Apr 1997, Mr M P Searle wrote: > > > > > > > Is there such a thing as a cheap quad Pentium Pro motherboard? I'm looking > > > > for an SMP Pro 150, but there aren't many 4 or 6 Pro motherboards > > > > around, and those that I could find were for large servers (eg Intel Alder, > > > > Goliath, etc.) > > > > > > > > Thanks, Michael. From owner-freebsd-smp Tue Apr 29 00:04:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA00764 for smp-outgoing; Tue, 29 Apr 1997 00:04:48 -0700 (PDT) Received: from jumping-spider.aracnet.com (root@jumping-spider.aracnet.com [204.188.47.14]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA00746 for ; Tue, 29 Apr 1997 00:04:43 -0700 (PDT) Received: from cbrowni2-home (ppp-r21.aracnet.com [205.238.13.23]) by jumping-spider.aracnet.com (8.8.5/8.8.5) with SMTP id AAA10745; Tue, 29 Apr 1997 00:05:20 -0700 Message-ID: <33659138.D27@earthling.net> Date: Mon, 28 Apr 1997 23:12:08 -0700 From: Chris Browning Reply-To: brownie@earthling.net X-Mailer: Mozilla 3.01 (WinNT; U) MIME-Version: 1.0 To: Poul-Henning Kamp CC: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? References: <291.862250592@critter> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > >The PPro's are stackable without additional electronics external > >to the processor. This makes it a *lot* easier to do multiple > >CPU's up to large numbers. > > The original design was for up to four PP's to be stackable. > Unfortunately it transpired to not be the case (for reasons > for signal integrity I belive, and therefore 4*PP cards have > a number of buffers more than 2*PP cards. What are you talking about here? I have never heard anything about having to add buffers for > 2 PPP designs. Can you point to a PPP errata for this? I think someone has misinformed you. Chris Not speaking for Intel. From owner-freebsd-smp Tue Apr 29 00:04:46 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA00759 for smp-outgoing; Tue, 29 Apr 1997 00:04:46 -0700 (PDT) Received: from jumping-spider.aracnet.com (root@jumping-spider.aracnet.com [204.188.47.14]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA00748 for ; Tue, 29 Apr 1997 00:04:44 -0700 (PDT) Received: from cbrowni2-home (ppp-r21.aracnet.com [205.238.13.23]) by jumping-spider.aracnet.com (8.8.5/8.8.5) with SMTP id AAA10739; Tue, 29 Apr 1997 00:05:18 -0700 Message-ID: <33659096.10DA@earthling.net> Date: Mon, 28 Apr 1997 23:09:26 -0700 From: Chris Browning Reply-To: brownie@earthling.net X-Mailer: Mozilla 3.01 (WinNT; U) MIME-Version: 1.0 To: Terry Lambert CC: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP References: <199704281724.KAA02110@phaeton.artisoft.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > Compare this to Sequent's Dynix, which scales well to 32 processors, > or Thinking Machines, which scales into the thousands of processors. The Sequent S81 at Georgia Tech, which uses Dynix, is not an SMP machine. Some of it's processor us a master-slave relationship. Scheduling occurs on the masters. I believe the CM's also had some dedicated nodes for different purposes, thus not making them SMP. Chris From owner-freebsd-smp Tue Apr 29 00:04:54 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA00774 for smp-outgoing; Tue, 29 Apr 1997 00:04:54 -0700 (PDT) Received: from jumping-spider.aracnet.com (root@jumping-spider.aracnet.com [204.188.47.14]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA00765 for ; Tue, 29 Apr 1997 00:04:51 -0700 (PDT) Received: from cbrowni2-home (ppp-r21.aracnet.com [205.238.13.23]) by jumping-spider.aracnet.com (8.8.5/8.8.5) with SMTP id AAA10756; Tue, 29 Apr 1997 00:05:26 -0700 Message-ID: <336591CD.576@earthling.net> Date: Mon, 28 Apr 1997 23:14:37 -0700 From: Chris Browning Reply-To: brownie@earthling.net X-Mailer: Mozilla 3.01 (WinNT; U) MIME-Version: 1.0 To: Poul-Henning Kamp CC: smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? References: <340.862250840@critter> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Anybody know what Intel/DOE will run on the 9144*PPro machine they're > building ? I believe that there is a "real-time" micro kernel that runs on each processor. I basically sits there waiting for a message to arrive and then executes based on that message. This is pretty much an educated guess. Next time I wonder by the TFLOPS building, I will try to think to ask :-). Chris Not speaking for Intel From owner-freebsd-smp Tue Apr 29 00:13:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA01108 for smp-outgoing; Tue, 29 Apr 1997 00:13:33 -0700 (PDT) Received: from critter.dk.tfs.com (phk.cybercity.dk [195.8.129.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA01103; Tue, 29 Apr 1997 00:13:25 -0700 (PDT) Received: from critter (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.5/8.8.5) with ESMTP id JAA01513; Tue, 29 Apr 1997 09:12:22 +0200 (CEST) To: brownie@earthling.net cc: Poul-Henning Kamp , smp@freebsd.org From: Poul-Henning Kamp Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Mon, 28 Apr 1997 23:12:08 PDT." <33659138.D27@earthling.net> Date: Tue, 29 Apr 1997 09:12:21 +0200 Message-ID: <1511.862297941@critter> Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In message <33659138.D27@earthling.net>, Chris Browning writes: >> >The PPro's are stackable without additional electronics external >> >to the processor. This makes it a *lot* easier to do multiple >> >CPU's up to large numbers. >> >> The original design was for up to four PP's to be stackable. >> Unfortunately it transpired to not be the case (for reasons >> for signal integrity I belive, and therefore 4*PP cards have >> a number of buffers more than 2*PP cards. > >What are you talking about here? I have never heard anything about >having to add buffers for > 2 PPP designs. Can you point to a >PPP errata for this? I think someone has misinformed you. > >Chris Well, I talked to a couple of people who did speak for Intel, they said that it was close to impossible to get four PP to work on a "plain" PC motherboard, you would have to do the daughterboard+buffer thing... The reasons stated was "termal, electrical & practical"... -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail. From owner-freebsd-smp Tue Apr 29 08:33:50 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA21285 for smp-outgoing; Tue, 29 Apr 1997 08:33:50 -0700 (PDT) Received: from mercury.marcftp.com (mercury.marcftp.com [208.137.183.2]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA21279 for ; Tue, 29 Apr 1997 08:33:46 -0700 (PDT) Received: from mercury.marcftp.com ([127.0.0.1]) by mercury.marcftp.com (Netscape Mail Server v2.0) with SMTP id AAA26353 for ; Tue, 29 Apr 1997 10:33:13 -0500 From: bjack@mercury.marcftp.com (Bruce Jackson) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: FreeBSD-SMP@FreeBSD.org Subject: Re: SMP In-Reply-To: <33659096.10DA@earthling.net> References: <199704281724.KAA02110@phaeton.artisoft.com> <33659096.10DA@earthling.net> X-Actually-From: "Bruce Jackson" Date: Tue, 29 Apr 1997 10:33:13 -0500 Message-ID: <19970429153313.AAA26353@mercury.marcftp.com> Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Terry Lambert Terry@lambert.org: >> Compare this to Sequent's Dynix, which scales well to 32 processors, >> or Thinking Machines, which scales into the thousands of processors. Actually, Sequent's large SMP machines like the S2000/700 and S81 systems only went to 30 processors. These machines shared a single pool of system memory and relied on processor caches to reduce memory contention. Sequent's latest technology called NUMA-Q (Non Uniform Memory Architecture-Quad (?)) organizes processors in groups of four each with its own memory. This technology scales to hundreds or even thousands of processors. For huge servers I expect hierarchical archetectures to be the next big thing. Chris Browning: > The Sequent S81 at Georgia Tech, which uses Dynix, is not an SMP > machine. Some of it's processor us a master-slave relationship. > Scheduling occurs on the masters. I believe the CM's also had > some dedicated nodes for different purposes, thus not making them > SMP. According to Sequent, except during system startup DYNIX is a true SMP OS and all CPU's operate on a peer basis. The DYNIX kernel is not threaded and runs on a single CPU though. The ability to reserve a CPU for a specific job or function does not preclude the system from having SMP capabilities. I've never used DYNIX/ptx so I do not know if it has a threaded kernel. From owner-freebsd-smp Tue Apr 29 09:29:58 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA24280 for smp-outgoing; Tue, 29 Apr 1997 09:29:58 -0700 (PDT) Received: from dn800e0.fingerhut.com (dn800e0-ext.fingerhut.com [204.221.45.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA24264 for ; Tue, 29 Apr 1997 09:29:54 -0700 (PDT) Received: from dn800e0.fingerhut.com (root@localhost) by dn800e0.fingerhut.com (8.7.5/8.7.3) with ESMTP id LAA00726 for ; Tue, 29 Apr 1997 11:31:28 -0500 (CDT) Received: from seag.fingerhut.com (GF007E0.SEAG.fingerhut.com [151.210.140.7]) by dn800e0.fingerhut.com (8.7.5/8.7.3) with SMTP id LAA00720 for ; Tue, 29 Apr 1997 11:31:27 -0500 (CDT) Received: from gf006e0.seag.fingerhut.com by seag.fingerhut.com (SMI-8.6/SMI-SVR4) id LAA07039; Tue, 29 Apr 1997 11:29:54 -0500 Received: by gf006e0.seag.fingerhut.com (5.x/SMI-SVR4) id AA09621; Tue, 29 Apr 1997 11:29:49 -0500 Date: Tue, 29 Apr 1997 11:29:49 -0500 Message-Id: <9704291629.AA09621@gf006e0.seag.fingerhut.com> From: Bruce Albrecht To: freebsd-smp@freebsd.org Subject: Tyan ATX1668 gets kernel panic on boot Mime-Version: 1.0 (generated by tm-edit 7.68) Content-Type: text/plain; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I downloaded -current from Sunday, 4/27, and built world, and attempted to build an SMP kernel. When it boots, it gets a kernel panic in CPU0 when running CPUIDLE1. I've noticed that LINT has changed since Sunday. Should I download -current again and try again? From owner-freebsd-smp Tue Apr 29 11:24:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA01085 for smp-outgoing; Tue, 29 Apr 1997 11:24:29 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA01078 for ; Tue, 29 Apr 1997 11:24:27 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA04826; Tue, 29 Apr 1997 11:23:27 -0700 From: Terry Lambert Message-Id: <199704291823.LAA04826@phaeton.artisoft.com> Subject: Re: SMP To: brownie@earthling.net Date: Tue, 29 Apr 1997 11:23:27 -0700 (MST) Cc: terry@lambert.org, FreeBSD-SMP@FreeBSD.org In-Reply-To: <33659096.10DA@earthling.net> from "Chris Browning" at Apr 28, 97 11:09:26 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-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > Compare this to Sequent's Dynix, which scales well to 32 processors, > > or Thinking Machines, which scales into the thousands of processors. > > The Sequent S81 at Georgia Tech, which uses Dynix, is not an SMP > machine. Some of it's processor us a master-slave relationship. This must be an older machine with the I/O processor relationship. I assume it's a 386 based box? > Scheduling occurs on the masters. I believe the CM's also had > some dedicated nodes for different purposes, thus not making them > SMP. The Connection Machines have processor local memory. You're right that they aren't true SMP, at least at the OS level. The issue was concurrency and scalability, though, not ethnic purity of the implementation. I think the Sandia labs machine isn't very "pure" either (I think it has cluster local memory). For what it's worth, I think that a hybrid system is probably the best trade off, in any case, for doing things which are inherently parallelizable (fluidic modelling, laminar air flow, factoring, etc.). This is getting off topic, but I admit I started it when I whined about the PC architecture. 8-). Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Tue Apr 29 13:21:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA07957 for smp-outgoing; Tue, 29 Apr 1997 13:21:36 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA07950 for ; Tue, 29 Apr 1997 13:21:29 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id OAA00439; Tue, 29 Apr 1997 14:21:15 -0600 (MDT) Message-Id: <199704292021.OAA00439@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Bruce Albrecht cc: freebsd-smp@freebsd.org Subject: Re: Tyan ATX1668 gets kernel panic on boot In-reply-to: Your message of "Tue, 29 Apr 1997 11:29:49 CDT." <9704291629.AA09621@gf006e0.seag.fingerhut.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 29 Apr 1997 14:21:15 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > I downloaded -current from Sunday, 4/27, and built world, and > attempted to build an SMP kernel. When it boots, it gets a kernel > panic in CPU0 when running CPUIDLE1. I've noticed that LINT has > changed since Sunday. Should I download -current again and try again? what messages came out of the panic? have you ever run an SMP kenrel on this hardware before? ie is this a new problem for you, or the first attempt at SMP? definitly cvsup -current and rebuild the kernel. this should only be a few files. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Apr 29 13:55:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA09560 for smp-outgoing; Tue, 29 Apr 1997 13:55:26 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA09542 for ; Tue, 29 Apr 1997 13:55:20 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id OAA00593; Tue, 29 Apr 1997 14:54:53 -0600 (MDT) Message-Id: <199704292054.OAA00593@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Poul-Henning Kamp cc: brownie@earthling.net, smp@freebsd.org Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Tue, 29 Apr 1997 09:12:21 +0200." <1511.862297941@critter> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 29 Apr 1997 14:54:53 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > Well, I talked to a couple of people who did speak for Intel, they said > that it was close to impossible to get four PP to work on a "plain" > PC motherboard, you would have to do the daughterboard+buffer thing... > > The reasons stated was "termal, electrical & practical"... I'm willing to accept this as fact. But I don't think it justifies the $900+ that AMI gets for their CPU daughtercards! Unfortunately the tail wags the dog here. Bean-counters crunch numbers and say: It will cost us $X to engineer/start manufacturing of this product. If we sell 1,000 of them this would be $X1 per unit. If we sell 1,000,000 of them this would be $X2 per unit. We *THINK* demand would be Y units given these 2 prices (and other factors). We will sell the units at price $Z based on the above facts. So in the case of 4-CPU motherboards it appears the bean-counters have decided there isn't enough demand to justify setting up for large volume and lower unit price. So the boards are too expensive for most of us. So most of us can't buy one. I submit that if someone designed a good 4-CPU P6 motherboard that sold in the $600-800 range (dual P6 boards are $260-360) it would become quite popular once the prices of P6 settle a little more (is the klammoth available yet?). -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Tue Apr 29 14:08:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA10997 for smp-outgoing; Tue, 29 Apr 1997 14:08:28 -0700 (PDT) Received: from critter.dk.tfs.com (phk.cybercity.dk [195.8.129.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA10992; Tue, 29 Apr 1997 14:08:12 -0700 (PDT) Received: from critter (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.5/8.8.5) with ESMTP id XAA03039; Tue, 29 Apr 1997 23:07:07 +0200 (CEST) To: Steve Passe cc: Poul-Henning Kamp , brownie@earthling.net, smp@freebsd.org From: Poul-Henning Kamp Subject: Re: Quad Pro 150 motherboard? In-reply-to: Your message of "Tue, 29 Apr 1997 14:54:53 MDT." <199704292054.OAA00593@Ilsa.StevesCafe.com> Date: Tue, 29 Apr 1997 23:07:06 +0200 Message-ID: <3037.862348026@critter> Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >I submit that if someone designed a good 4-CPU P6 motherboard that sold >in the $600-800 range (dual P6 boards are $260-360) it would become >quite popular once the prices of P6 settle a little more (is the klammoth >available yet?). I tend to agree, but the prices for the CPU's will still make it a rather expensive affair... But if you find one such card, let me know :-) -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail. From owner-freebsd-smp Tue Apr 29 19:42:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA27498 for smp-outgoing; Tue, 29 Apr 1997 19:42:23 -0700 (PDT) Received: from meshsv64.os.mesh.ad.jp (meshsv64.os.mesh.ad.jp [133.205.64.132]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA27491 for ; Tue, 29 Apr 1997 19:42:20 -0700 (PDT) Received: from cubic.shisham.co.jp (tksm1DS10.tks.mesh.ad.jp [133.205.221.130]) by meshsv64.os.mesh.ad.jp (8.8.4+2.7Wbeta4/3.5Wpl1-) with ESMTP id LAA04171 for ; Wed, 30 Apr 1997 11:42:07 +0900 (JST) Received: from localhost (benzaiten.shisham.co.jp [172.25.1.2]) by cubic.shisham.co.jp (8.8.4+2.7Wbeta4/3.5Wbeta) with ESMTP id LAA00660 for ; Wed, 30 Apr 1997 11:27:43 +0900 (JST) Message-Id: To: smp@freebsd.org Subject: why general protection fault happen on SMP kernel? From: Shiba In-Reply-To: Your message of "Mon, 21 Apr 1997 23:16:58 -0600" References: <199704220516.XAA14156@Ilsa.StevesCafe.com> X-Mailer: Mew version 1.69 on Emacs 19.34.1 / Mule 2.3 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Date: Wed, 30 Apr 1997 11:27:42 +0900 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >>><199704220516.XAA14156@Ilsa.StevesCafe.com> >>> Steve Passe wrote Sorry,my poor english. Steve> Make/Model of motherboard: Steve> Giga-byte GA686DX Steve> --- Steve> Http links to manufacturer and/or board info: Steve> http://www.giga-byte.com/GA686DX.html Steve> --- Steve> SMP kernel works [yes/no]: Steve> yes NO Steve> --- Steve> APIC_IO works [yes/no]: Steve> yes NO Steve> --- Steve> Special #defines/options needed: Steve> none NONE I cvsupped down the current and compiled it, and ~~~~~~~ 'make world' . 'general protection fault' happen on the 'syscon.c(sc??putc)' config file of 'SINGLE cpu kernel' is machine "i386" cpu "I386_CPU" cpu "I486_CPU" cpu "I586_CPU" cpu "I686_CPU" ident benzaiten maxusers 64 options "MAXMEM=(128*1024)" options SYSVSEM,SYSVSHM,SYSVMSG options "SHMMAXPGS=2048" options "MAXDSIZ=(512*1024*1024)" options "DFLDSIZ=(512*1024*1024)" #options SMP #options APIC_IO #options SMP_INVLTBL #options NCPU=2 #options NBUS=2 #options NAPIC=1 #options NINTR=24 options INET #InterNETworking options FFS #Berkeley Fast Filesystem options NFS,MFS #Network Filesystem options "COMPAT_43" #Compatible with BSD 4.3 [KEEP THIS!] options UCONSOLE #Allow users to grab the console options QUOTA options "NSWAPDEV=4" config kernel root on sd0 swap sd0 and sd1 and sd2 controller pci0 device de0 controller ncr0 #tekram-dc390f controller ncr1 #asus-sc-200 controller scbus0 at ncr0 bus 0 controller scbus1 at ncr1 bus 0 disk sd0 at scbus0 target 0 unit 0 #device od0 #See LINT for possible `od' options. controller isa0 controller fdc0 at isa? port "IO_FD1" bio irq 6 drq 2 vector fdintr disk fd0 at fdc0 drive 0 device sc0 at isa? port "IO_KBD" tty irq 1 vector scintr device npx0 at isa? port "IO_NPX" irq 13 vector npxintr device sio0 at isa? port "IO_COM1" tty irq 4 vector siointr device sio1 at isa? port "IO_COM2" tty irq 3 vector siointr device lpt0 at isa? port? tty irq 7 vector lptintr pseudo-device loop pseudo-device ether pseudo-device bpfilter 8 pseudo-device vn 4 pseudo-device log pseudo-device gzip # Exec gzipped a.out's #options KTRACE #kernel tracing #options DDB OK. but add option under options SMP options APIC_IO options SMP_INVLTBL options NCPU=2 options NBUS=2 options NAPIC=1 options NINTR=24 NG. my hardware list GA-686DX P6 180M * 2 (overclocking now(233M)) tekram dc-390f asus sc-200 dec de500-xa From owner-freebsd-smp Tue Apr 29 19:47:59 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA27674 for smp-outgoing; Tue, 29 Apr 1997 19:47:59 -0700 (PDT) Received: from cs.utah.edu (cs.utah.edu [128.110.4.21]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA27669 for ; Tue, 29 Apr 1997 19:47:57 -0700 (PDT) Received: from fast.cs.utah.edu by cs.utah.edu (8.8.4/utah-2.21-cs) id UAA11892; Tue, 29 Apr 1997 20:47:43 -0600 (MDT) Received: by fast.cs.utah.edu (8.6.10/utah-2.15-leaf) id UAA08040; Tue, 29 Apr 1997 20:47:42 -0600 Date: Tue, 29 Apr 1997 20:47:42 -0600 From: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Message-Id: <199704300247.UAA08040@fast.cs.utah.edu> To: shiba@mxs.meshnet.or.jp, smp@freebsd.org Subject: Re: why general protection fault happen on SMP kernel? Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > GA-686DX > P6 180M * 2 (overclocking now(233M)) > tekram dc-390f > asus sc-200 > dec de500-xa Try running your 180's at 180, and see if you still get the GPF. Kevin From owner-freebsd-smp Tue Apr 29 21:08:51 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA01880 for smp-outgoing; Tue, 29 Apr 1997 21:08:51 -0700 (PDT) Received: from meshsv64.os.mesh.ad.jp (meshsv64.os.mesh.ad.jp [133.205.64.132]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA01870 for ; Tue, 29 Apr 1997 21:08:45 -0700 (PDT) Received: from cubic.shisham.co.jp (tksm1DS01.tks.mesh.ad.jp [133.205.221.121]) by meshsv64.os.mesh.ad.jp (8.8.4+2.7Wbeta4/3.5Wpl1-) with ESMTP id NAA07368; Wed, 30 Apr 1997 13:08:03 +0900 (JST) Received: from localhost (localhost [127.0.0.1]) by cubic.shisham.co.jp (8.8.4+2.7Wbeta4/3.5Wbeta) with ESMTP id NAA00917; Wed, 30 Apr 1997 13:07:35 +0900 (JST) Message-Id: To: vanmaren@fast.cs.utah.edu Cc: smp@freebsd.org Subject: Re: why general protection fault happen on SMP kernel? In-Reply-To: Your message of "Tue, 29 Apr 1997 20:47:42 -0600" References: <199704300247.UAA08040@fast.cs.utah.edu> X-Mailer: Mew version 1.69 on Emacs 19.34.1 / Mule 2.3 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Date: Wed, 30 Apr 1997 13:07:35 +0900 From: SHIBA Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >>><199704300247.UAA08040@fast.cs.utah.edu> >>> vanmaren@fast.cs.utah.edu (Kevin Van Maren) wrote Kevin> > GA-686DX Kevin> > P6 180M * 2 (overclocking now(233M)) Kevin> > tekram dc-390f Kevin> > asus sc-200 Kevin> > dec de500-xa Kevin> Try running your 180's at 180, and see if you still get the GPF. thanks, 180--->200 (233->200 ;_;) OK. but kernel execute on 233MHz before Apr. 20 (about) why? From owner-freebsd-smp Tue Apr 29 23:26:57 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA08580 for smp-outgoing; Tue, 29 Apr 1997 23:26:57 -0700 (PDT) Received: from housing1.stucen.gatech.edu (ken@housing1.stucen.gatech.edu [130.207.52.71]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA08575 for ; Tue, 29 Apr 1997 23:26:55 -0700 (PDT) Received: (from ken@localhost) by housing1.stucen.gatech.edu (8.8.5/8.8.5) id CAA01915; Wed, 30 Apr 1997 02:25:10 -0400 (EDT) From: Kenneth Merry Message-Id: <199704300625.CAA01915@housing1.stucen.gatech.edu> Subject: Re: SMP In-Reply-To: <199704291823.LAA04826@phaeton.artisoft.com> from Terry Lambert at "Apr 29, 97 11:23:27 am" To: terry@lambert.org (Terry Lambert) Date: Wed, 30 Apr 1997 02:25:10 -0400 (EDT) Cc: brownie@earthling.net, 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-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Terry Lambert wrote... > > > Compare this to Sequent's Dynix, which scales well to 32 processors, > > > or Thinking Machines, which scales into the thousands of processors. > > > > The Sequent S81 at Georgia Tech, which uses Dynix, is not an SMP > > machine. Some of it's processor us a master-slave relationship. > > This must be an older machine with the I/O processor relationship. I > assume it's a 386 based box? Yes, it was a 386 machine. (I say was because ol' hydra has gone to the great bit heap in the sky...) I believe in its final configuration it was 16 33MHz 386's, with 387 coprocessors. Dynix had this fun utility called monitor that would show the load on each of the processors... (a little ascii-based thing, looked kinda like this: 0 ---> 8 ---------> 1 ------> 9 ------> ... ... Except it was full-screen. Ken -- Kenneth Merry ken@ulc199.residence.gatech.edu Disclaimer: I don't speak for GTRI, GT, or Elvis. From owner-freebsd-smp Wed Apr 30 06:41:49 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id GAA26653 for smp-outgoing; Wed, 30 Apr 1997 06:41:49 -0700 (PDT) Received: from abby.skypoint.net (abby.skypoint.net [199.86.32.252]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id GAA26646 for ; Wed, 30 Apr 1997 06:41:46 -0700 (PDT) Received: (from uucp@localhost) by abby.skypoint.net (8.8.5/alexis 2.7) with UUCP id IAA03768; Wed, 30 Apr 1997 08:41:39 -0500 (CDT) Received: (from bruce@localhost) by zuhause.mn.org (8.8.5/8.8.5) id IAA06969; Wed, 30 Apr 1997 08:38:47 -0500 (CDT) Date: Wed, 30 Apr 1997 08:38:47 -0500 (CDT) Message-Id: <199704301338.IAA06969@zuhause.mn.org> From: Bruce Albrecht To: Steve Passe Cc: freebsd-smp@FreeBSD.ORG Subject: Re: Tyan ATX1668 gets kernel panic on boot In-Reply-To: <199704292021.OAA00439@Ilsa.StevesCafe.com> References: <9704291629.AA09621@gf006e0.seag.fingerhut.com> <199704292021.OAA00439@Ilsa.StevesCafe.com> X-Mailer: VM 6.22 under 19.15p2 XEmacs Lucid Mime-Version: 1.0 (generated by tm-edit 7.106) Content-Type: text/plain; charset=US-ASCII Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Steve Passe writes: > Hi, > > > I downloaded -current from Sunday, 4/27, and built world, and > > attempted to build an SMP kernel. When it boots, it gets a kernel > > panic in CPU0 when running CPUIDLE1. I've noticed that LINT has > > changed since Sunday. Should I download -current again and try again? > > what messages came out of the panic? > > have you ever run an SMP kenrel on this hardware before? ie is this a new > problem for you, or the first attempt at SMP? > > definitly cvsup -current and rebuild the kernel. this should only be a few > files. I cvsup'd -current last night, did a make world, and rebuilt the kernel. I now have a running SMP system. Thanks everyone! From owner-freebsd-smp Wed Apr 30 13:08:42 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA14748 for smp-outgoing; Wed, 30 Apr 1997 13:08:42 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA14730 for ; Wed, 30 Apr 1997 13:08:37 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by who.cdrom.com (8.8.5/8.6.11) with ESMTP id MAA07023 for ; Wed, 30 Apr 1997 12:00:20 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id NAA02694 for ; Wed, 30 Apr 1997 13:00:17 -0600 (MDT) Message-Id: <199704301900.NAA02694@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: freebsd-smp@freebsd.org Subject: mapping of current mp_lock usage. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 30 Apr 1997 13:00:17 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I have created a document that maps the usage of the "giant lock" (mp_lock) in the current SMP kernel: http://www.freebsd.org/~fsmp/SMP/papers/lockmap.txt this should help those interested to understand what/where we lock in the current implimentation. It is from this point that we will start the "push-down" of the lock into the various sub-systems. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Apr 30 14:26:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA18825 for smp-outgoing; Wed, 30 Apr 1997 14:26:53 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA18820 for ; Wed, 30 Apr 1997 14:26:50 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id PAA03188 for ; Wed, 30 Apr 1997 15:26:41 -0600 (MDT) Message-Id: <199704302126.PAA03188@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@freebsd.org Subject: Re: hardware list Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 30 Apr 1997 15:26:41 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I have update the "known working hardware" page with the info returned to me tha past several weeks. Please review this page and report any hardware that you have working with the current SMP kernel that is NOT LISTED in this page: http://www.freebsd.org/~fsmp/SMP/hardware.html tia, -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Apr 30 19:37:27 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA29664 for smp-outgoing; Wed, 30 Apr 1997 19:37:27 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA29613 for ; Wed, 30 Apr 1997 19:37:21 -0700 (PDT) From: mef@cs.washington.edu Received: from daffy-duck.cs.washington.edu (daffy-duck.cs.washington.edu [128.95.2.11]) by who.cdrom.com (8.8.5/8.6.11) with ESMTP id TAA09359 for ; Wed, 30 Apr 1997 19:02:21 -0700 (PDT) Received: (mef@localhost) by daffy-duck.cs.washington.edu (8.7.2/7.2ws+) id TAA26047; Wed, 30 Apr 1997 19:02:13 -0700 (PDT) Date: Wed, 30 Apr 1997 19:02:13 -0700 (PDT) Message-Id: <199705010202.TAA26047@daffy-duck.cs.washington.edu> To: smp@csn.net CC: smp@freebsd.org, ulbright@cs.washington.edu In-reply-to: <199704240005.SAA24132@Ilsa.StevesCafe.com> (message from Steve Passe on Wed, 23 Apr 1997 18:05:56 -0600) Subject: Re: Intel PR440FX motherboard anybody ? Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Steve, Last week you wanted me to check whether just using the SMP_TIMER_NC option would work. We updated to 3.0-CURRENT 4/29/97 and built the kernel with the SMP_TIMER_NC option. When booting the system we still had the same problem with the APIC no connection stuff. After uncommenting the following defines in smptest.h the system seemed to work just fine. #define FAKE_8254_NC #define IRQ_LO_NC What additional information can we provide to you to help out in fixing this? As mentioned, the system we have is an Intel Buckeye dual processor, which I believe uses a PR440FX motherboard. Any way, I suppose this motherboard should make it onto your rogue board list. Marc From: Steve Passe Date: Wed, 23 Apr 1997 18:05:56 -0600 > 1) needed to turn on "options SMP_TIMER_NC" I would be interested to know if this is still true with the latest code. My last changes should make this only necessary if the motherboard lies about the timer/APIC connection. The MP table I have on file for this board shows it properly declaring the missing connection, and thus the software should work without explicitly declaring it via SMP_TIMER_NC (with the latest code, this is a recent change). -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Wed Apr 30 19:46:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA02105 for smp-outgoing; Wed, 30 Apr 1997 19:46:26 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA02079 for ; Wed, 30 Apr 1997 19:46:18 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by who.cdrom.com (8.8.5/8.6.11) with SMTP id LAA06552 for ; Wed, 30 Apr 1997 11:00:18 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA02337; Wed, 30 Apr 1997 10:58:27 -0700 From: Terry Lambert Message-Id: <199704301758.KAA02337@phaeton.artisoft.com> Subject: Re: why general protection fault happen on SMP kernel? To: shiba@mxs.meshnet.or.jp (SHIBA) Date: Wed, 30 Apr 1997 10:58:27 -0700 (MST) Cc: vanmaren@fast.cs.utah.edu, smp@freebsd.org In-Reply-To: from "SHIBA" at Apr 30, 97 01:07:35 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-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Kevin> Try running your 180's at 180, and see if you still get the GPF. > > thanks, > 180--->200 (233->200 ;_;) > OK. > > but kernel execute on 233MHz before Apr. 20 (about) When's the official start of summer in your neck of the woods? 8-). Could be ambient temperature is up... so the same cooling is less effective. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Wed Apr 30 21:06:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA06312 for smp-outgoing; Wed, 30 Apr 1997 21:06:36 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id VAA06265; Wed, 30 Apr 1997 21:06:26 -0700 (PDT) Received: from lamb.sas.com (daemon@lamb.sas.com [192.35.83.8]) by who.cdrom.com (8.8.5/8.6.11) with SMTP id KAA06062 ; Wed, 30 Apr 1997 10:01:19 -0700 (PDT) Received: from mozart by lamb.sas.com (5.65c/SAS/Gateway/01-23-95) id AA03867; Wed, 30 Apr 1997 13:01:06 -0400 Received: from iluvatar.unx.sas.com by mozart (5.65c/SAS/Domains/5-6-90) id AA13297; Wed, 30 Apr 1997 12:49:00 -0400 Received: by iluvatar.unx.sas.com (5.65c/SAS/Generic 9.01/3-26-93) id AA18235; Wed, 30 Apr 1997 12:47:25 -0400 From: "John W. DeBoskey" Message-Id: <199704301647.AA18235@iluvatar.unx.sas.com> Subject: Adaptec AIC-7880 Ultra/Ultra W problems To: freebsd-current@freebsd.org, freebsd-smp@freebsd.org Date: Wed, 30 Apr 1997 12:46:55 -0400 (EDT) X-Mailer: ELM [version 2.4 PL23] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, As the subject says, I'm having problems with the 7880 onboard scsi controller. I am running 3.0-CURRENT as of Monday, April 28. I am running an SMP kernel, without the additional CPU's activated. When doing heavy i/o the following messages appear on the console, followed by the machine locking up. I've run with the adapter in both wide/ultra wide mode with the same results. The drives are WD 4.3GB, Single-ended Ultra Fast Wide SCSI-3. The disconnnect option is on. Any ideas on how to approach this problem are greatly appreciated. Thanks, John ps: If this is the 2940 problem, I'll be more than happy to test any patches. These are typed in by hand so any typos are mine. ------------------------------------------------------------------ sd1: SCB 0x1 - timed out in message out phase, SCSISIGI == 0xb6 SEQADDR = 0x47 SCSISEQ = 0x12 SSTAT0 = 0x7 SSTAT1 = 0x3 sd1: abort message in buffer sd1: SCB 0 - Abort Completed sd1: no longer in timeout sd1: SCB 0xd - timed out while idle, LASTPHASE == 0x1, SCSISIGI = 0x0 SEQADDR = 0xd SCSISEQ = 0x12 SSTAT0 = 0x5 SSTAT1 = 0x2 sd1: Queueing an Abort SCB sd1: Abort Message Sent sd1: SCB 0x0 - timed out in message out phase, SCSISIGI == 0xb6 SEQADDR = 0x47 SCSISEQ = 0x12 SSTAT0 = 0x7 SSTAT1 = 0x3 The above sequence repeats many times. The dmesg output for my machine follows: Copyright (c) 1992-1997 FreeBSD Inc. Copyright (c) 1982, 1986, 1989, 1991, 1993 The Regents of the University of California. All rights reserved. FreeBSD 3.0-CURRENT #0: Mon Apr 28 12:46:53 EST 1997 root@magenta.pc.sas.com:/usr/src/sys/compile/FOURWAY FreeBSD/SMP: Multiprocessor motherboard cpu0 (BSP): apic id: 0, version: 0x00040011 cpu1 (AP): apic id: 4, version: 0x00040011 cpu2 (AP): apic id: 1, version: 0x00040011 cpu3 (AP): apic id: 2, version: 0x00040011 io0 (APIC): apic id: 14, version: 0x000f0011 CPU: Pentium Pro (686-class CPU) Origin = "GenuineIntel" Id = 0x619 Stepping=9 Features=0xfbff,MTRR,PGE,MCA,CMOV> real memory = 134217728 (131072K bytes) avail memory = 129470464 (126436K bytes) bdevsw_add_generic: adding D_DISK flag for device 16 eisa0: Probing for devices on the EISA bus Probing for devices on PCI bus 0: chip0 rev 5 on pci0:14:0 pci0:15:0: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:1: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:2: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:3: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:4: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:5: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:6: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] pci0:15:7: Intel Corporation, device=0x0008, class=0xff, subclass=0x00 [no driver assigned] chip1 rev 5 on pci0:20:0 chip2 rev 6 on pci0:25:0 chip3 rev 6 on pci0:26:0 Probing for devices on PCI bus 1: fxp0 rev 1 int a irq 10 on pci1:10:0 fxp0: Ethernet address 00:a0:c9:2d:14:34 ahc0 rev 0 int a irq 11 on pci1:11:0 ahc0: Using left over BIOS settings ahc0: aic7880 Wide Channel, SCSI Id=7, 16 SCBs ahc0: waiting for scsi devices to settle scbus0 at ahc0 bus 0 sd0 at scbus0 target 0 lun 0 sd0: type 0 fixed SCSI 2 sd0: Direct-Access 1030MB (2109840 512 byte sectors) cd0 at scbus0 target 5 lun 0 cd0: type 5 removable SCSI 2 cd0: CD-ROM can't get the size ahc1 rev 0 int a irq 10 on pci1:12:0 ahc1: Using left over BIOS settings ahc1: aic7880 Wide Channel, SCSI Id=7, 16 SCBs ahc1: waiting for scsi devices to settle scbus1 at ahc1 bus 0 sd1 at scbus1 target 0 lun 0 sd1: type 0 fixed SCSI 2 sd1: Direct-Access 4095MB (8388314 512 byte sectors) sd2 at scbus1 target 1 lun 0 sd2: type 0 fixed SCSI 2 sd2: Direct-Access 4095MB (8388314 512 byte sectors) sd3 at scbus1 target 2 lun 0 sd3: type 0 fixed SCSI 2 sd3: Direct-Access 4095MB (8388314 512 byte sectors) sd4 at scbus1 target 3 lun 0 sd4: type 0 fixed SCSI 2 sd4: Direct-Access 4095MB (8388314 512 byte sectors) sd5 at scbus1 target 4 lun 0 sd5: type 0 fixed SCSI 2 sd5: Direct-Access 4095MB (8388314 512 byte sectors) sd6 at scbus1 target 5 lun 0 sd6: type 0 fixed SCSI 2 sd6: Direct-Access 4095MB (8388314 512 byte sectors) uk0 at scbus1 target 6 lun 0 uk0: type 3 fixed SCSI 2 uk0: Unknown Probing for devices on the ISA bus: sc0 at 0x60-0x6f irq 1 on motherboard sc0: VGA color <16 virtual consoles, flags=0x0> sio0 at 0x3f8-0x3ff irq 4 on isa sio0: type 16550A sio1 at 0x2f8-0x2ff irq 3 on isa sio1: type 16550A lpt0 at 0x378-0x37f irq 7 on isa lpt0: Interrupt-driven port lp0: TCP/IP capable interface lpt1 not found mse0 not found at 0x23c psm0 at 0x60-0x64 irq 12 on motherboard psm0: device ID 0, 2 buttons fdc0 at 0x3f0-0x3f7 irq 6 drq 2 on isa fdc0: NEC 72065B fd0: 1.44MB 3.5in scd0 not found at 0x230 npx0 on motherboard npx0: INT 16 interface apm0: disabled, not probed. changing root device to sd0a Enabled INTs: 1, 2, 3, 4, 6, 7, 8, 10, 11, 12, imen: 0x00ffe221 WARNING: / was not properly dismounted. SMP: All idle procs online. -- jwd@unx.sas.com (w) John W. De Boskey (919) 677-8000 x6915 From owner-freebsd-smp Wed Apr 30 22:46:29 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id WAA11431 for smp-outgoing; Wed, 30 Apr 1997 22:46:29 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id WAA11426; Wed, 30 Apr 1997 22:46:27 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id XAA05180; Wed, 30 Apr 1997 23:46:22 -0600 (MDT) Message-Id: <199705010546.XAA05180@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: "John W. DeBoskey" cc: freebsd-current@freebsd.org, freebsd-smp@freebsd.org Subject: Re: Adaptec AIC-7880 Ultra/Ultra W problems In-reply-to: Your message of "Wed, 30 Apr 1997 12:46:55 EDT." <199704301647.AA18235@iluvatar.unx.sas.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 30 Apr 1997 23:46:22 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > As the subject says, I'm having problems with the 7880 onboard >scsi controller. I am running 3.0-CURRENT as of Monday, April 28. >I am running an SMP kernel, without the additional CPU's >activated. what motherboard is this? have you previously sent me the output of mptable? it looks like 'options APIC_IO' is NOT enabled, is this true? is the standard UP kernel working OK? -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 04:06:30 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id EAA17359 for smp-outgoing; Thu, 1 May 1997 04:06:30 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id EAA17341 for ; Thu, 1 May 1997 04:06:27 -0700 (PDT) Received: from cenotaph.snafu.de (gw-deadnet.snafu.de [194.121.229.33]) by who.cdrom.com (8.8.5/8.6.11) with SMTP id DAA11507 for ; Thu, 1 May 1997 03:40:12 -0700 (PDT) Received: by cenotaph.snafu.de from deadline.snafu.de using smtp id m0wMtHE-000KAzC; Thu, 1 May 1997 12:39:52 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Received: by deadline.snafu.de id m0wMtHD-000433C; Thu, 1 May 1997 12:39:51 +0200 (CEST) (Smail-3.2 1996-Jul-4 #1) Message-Id: Date: Thu, 1 May 1997 12:39:51 +0200 (CEST) Mime-Version: 1.0 X-Newsreader: knews 0.9.8 References: In-Reply-To: From: root@deadline.snafu.de (Andreas S. Wetzel) Subject: Re: hardware list X-Original-Newsgroups: lists.freebsd-smp To: Steve Passe Cc: smp@freebsd.org Content-Type: text/plain; charset=us-ascii Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk In article , Steve Passe writes: --- Make/Model of motherboard: ASUS P/I-P65UP5 with C-P55T2D cpu card --- Http links to manufacturer and/or board info: http://www.asus.com/products/Specs/MB/p65up5-Spec.asp --- SMP kernel works [yes/no]: yes --- APIC_IO works [yes/no]: yes --- Special #defines/options needed: # SMP options options SMP # SMP kernel options APIC_IO # Symmetric (APIC) I/O options NCPU=2 # number of CPUs options NBUS=2 # number of busses options NAPIC=1 # number of IO APICs options NINTR=24 # number of INTs options SMP_INVLTLB # --- Comments: Regards, Mickey -- (__) (@@) Andreas S. Wetzel Mail: mickey@deadline.snafu.de /-------\/ Utrechter Strasse 41 Web: http://cenotaph.snafu.de/mickey/ / | || 13347 Berlin Fon: <+4930> 456 066 90 * ||----|| Germany Fax: <+4930> 456 066 91/92 ~~ ~~ From owner-freebsd-smp Thu May 1 06:57:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id GAA24918 for smp-outgoing; Thu, 1 May 1997 06:57:03 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id GAA24911 for ; Thu, 1 May 1997 06:57:00 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id PAA07481; Thu, 1 May 1997 15:45:19 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id MAA08496; Thu, 1 May 1997 12:42:18 +0200 (CEST) Message-ID: <19970501124217.57137@klemm.gtn.com> Date: Thu, 1 May 1997 12:42:17 +0200 From: Andreas Klemm To: Steve Passe Cc: freebsd-smp@FreeBSD.org Subject: Re: mapping of current mp_lock usage. References: <199704301900.NAA02694@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 In-Reply-To: <199704301900.NAA02694@Ilsa.StevesCafe.com>; from Steve Passe on Wed, Apr 30, 1997 at 01:00:17PM -0600 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Wed, Apr 30, 1997 at 01:00:17PM -0600, Steve Passe wrote: > Hi, > > I have created a document that maps the usage of the "giant lock" (mp_lock) > in the current SMP kernel: > > http://www.freebsd.org/~fsmp/SMP/papers/lockmap.txt File Not found The requested URL /~fsmp/SMP/papers/lockmap.txt was not found on this server. The same for any fbsd SMP directory ... ??!! -- powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html From owner-freebsd-smp Thu May 1 07:01:13 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA25288 for smp-outgoing; Thu, 1 May 1997 07:01:13 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA25281 for ; Thu, 1 May 1997 07:01:04 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id PAA07489; Thu, 1 May 1997 15:45:22 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id PAA02810; Thu, 1 May 1997 15:39:29 +0200 (CEST) Date: Thu, 1 May 1997 15:39:29 +0200 (CEST) Message-Id: <199705011339.PAA02810@klemm.gtn.com> Mime-Version: 1.0 X-Newsreader: knews 0.9.8 References: <6VA7t7pLSfB@surfer.ods.de> <6VbVZdMLSfB@surfer.ods.de> <5k1utd$dj8$1@hrz-ws11.hrz.uni-kassel.de> <6VndaKB5SfB@surfer.ods.de> From: andreas@klemm.gtn.com (Andreas Klemm) Subject: Re: Dual PPro Mainboard f. SCO SMP Newsgroups: de.comp.os.unix To: WH@ODS.de (Walter Haslbeck), smp@FreeBSD.org Cc: everybodyunix@wup.de, jk@ct.heise.de Content-Type: text/plain; charset=us-ascii Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk [Posted and mailed] In article <6VndaKB5SfB@surfer.ods.de>, WH@ODS.de (Walter Haslbeck) writes: [snip] > BTW: Wie siehts eigentlich z.Z. an der FreeUnix Front aus in Sachen SMP? > Gibts irgendwo im Web Benchmarkergebnisse, die SCO SMP mit Linux/SMP > bzw. FreeBSD/SMP vergleichen? I'm writing this in english, so I can write this also to the FreeBSD SMP mailinglist. I suppose the people there are interested in numbers and figures as well ;-) I'll put together some nice gif files later, that are suitable to be put onto a Web Server. I made something like an application level benchmark by compiling my FreeBSD custom kernel with different kernels and job parameters for make (-j). You can see this kind of tests already in PC magazines like C't, where they bench the compile time for a Linux kernel. For reference: Tyan Titan Pro ATX, 64 MB RAM, 2 x 200 MHz Pro >>> normal -current kernel make -j 1 262.09 real 189.47 user 15.21 sys make -j 2 215.58 real 191.09 user 16.24 sys make -j 4 213.17 real 191.39 user 19.49 sys make -j 8 213.24 real 191.91 user 19.40 sys make -j 16 215.19 real 192.10 user 19.31 sys >>> -current SMP kernel 1 CPU make -j 1 269.95 real 189.22 user 23.39 sys make -j 2 222.68 real 191.38 user 23.03 sys make -j 4 218.24 real 192.00 user 23.47 sys make -j 8 217.93 real 191.35 user 24.29 sys make -j 16 220.15 real 191.58 user 24.26 sys >>> SMP kernel 2 CPUs (after sysctl -w kern.smp_active=2) make -j 1 252.35 real 144.69 user 75.06 sys make -j 2 137.06 real 169.23 user 56.46 sys make -j 4 119.95 real 176.02 user 53.25 sys make -j 8 119.08 real 175.47 user 54.97 sys make -j 16 120.03 real 178.42 user 53.32 sys observations: 1) comparing FreeBSD Uniprocessor vs. SMP kernel (both with 1 CPU) Using the SMP kernel the system spent a little more time in kernel mode. 2) Using the SMP kernel with 2 CPU's and make -j 4 and -j 8 is really fast The performance boost you get now with FreeBSD SMP _for_this_kind_of_application is factor 1.9 !!! To sum up: you nearly double the performance of your system !!! I think important here is, that every PPro CPU has it's own 2nd level cache. Andreas /// -- powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html From owner-freebsd-smp Thu May 1 07:36:37 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA26967 for smp-outgoing; Thu, 1 May 1997 07:36:37 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA26953 for ; Thu, 1 May 1997 07:36:34 -0700 (PDT) Received: from abby.skypoint.net (abby.skypoint.net [199.86.32.252]) by who.cdrom.com (8.8.5/8.6.11) with ESMTP id HAA12007 for ; Thu, 1 May 1997 07:06:43 -0700 (PDT) Received: (from uucp@localhost) by abby.skypoint.net (8.8.5/alexis 2.7) with UUCP id JAA20341 for freebsd-smp@freebsd.org; Thu, 1 May 1997 09:06:41 -0500 (CDT) Received: (from bruce@localhost) by zuhause.mn.org (8.8.5/8.8.5) id IAA01945; Thu, 1 May 1997 08:56:17 -0500 (CDT) Date: Thu, 1 May 1997 08:56:17 -0500 (CDT) Message-Id: <199705011356.IAA01945@zuhause.mn.org> From: Bruce Albrecht To: freebsd-smp@freebsd.org Subject: Where to start SMP? X-Mailer: VM 6.22 under 19.15p2 XEmacs Lucid Mime-Version: 1.0 (generated by tm-edit 7.106) Content-Type: text/plain; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Where's the recommended location for starting SMP? I'm doing it in /usr/local/etc/rc.d, but I was wondering if I was missing something. I think it should be added to /etc/rc and /etc/rc.conf now that the SMP code is in -current. From owner-freebsd-smp Thu May 1 10:49:32 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA05615 for smp-outgoing; Thu, 1 May 1997 10:49:32 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA05610 for ; Thu, 1 May 1997 10:49:30 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id LAA07241; Thu, 1 May 1997 11:49:15 -0600 (MDT) Message-Id: <199705011749.LAA07241@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Bruce Albrecht cc: freebsd-smp@freebsd.org Subject: Re: Where to start SMP? In-reply-to: Your message of "Thu, 01 May 1997 08:56:17 CDT." <199705011356.IAA01945@zuhause.mn.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 11:49:15 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > Where's the recommended location for starting SMP? I'm doing it in > /usr/local/etc/rc.d, but I was wondering if I was missing something. > I think it should be added to /etc/rc and /etc/rc.conf now that the > SMP code is in -current. -- there is no 'recommended' place. experiment, let us know what works well... Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 10:54:24 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA05946 for smp-outgoing; Thu, 1 May 1997 10:54:24 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA05940 for ; Thu, 1 May 1997 10:54:19 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id LAA07275; Thu, 1 May 1997 11:54:08 -0600 (MDT) Message-Id: <199705011754.LAA07275@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Andreas Klemm cc: freebsd-smp@FreeBSD.org Subject: Re: mapping of current mp_lock usage. In-reply-to: Your message of "Thu, 01 May 1997 12:42:17 +0200." <19970501124217.57137@klemm.gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 11:54:08 -0600 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Hi, > > I have created a document that maps the usage of the "giant lock" (mp_lock) > > in the current SMP kernel: > > > > http://www.freebsd.org/~fsmp/SMP/papers/lockmap.txt > > File Not found > > The requested URL /~fsmp/SMP/papers/lockmap.txt was not found on this server. > > The same for any fbsd SMP directory ... ??!! try again, there were network problems most of the nite. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 10:57:40 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id KAA06184 for smp-outgoing; Thu, 1 May 1997 10:57:40 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [194.77.0.15]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id KAA06132; Thu, 1 May 1997 10:56:45 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id TAA10171; Thu, 1 May 1997 19:45:17 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id TAA11398; Thu, 1 May 1997 19:44:12 +0200 (CEST) Message-ID: <19970501194412.06645@klemm.gtn.com> Date: Thu, 1 May 1997 19:44:12 +0200 From: Andreas Klemm To: smp@freebsd.org Cc: smp@csn.net, peter@freebsd.org, jkh@freebsd.org, reny@klemm.gtn.com Subject: Made nice HTML page about FreeBSD SMP performance Mime-Version: 1.0 Content-Type: multipart/mixed; boundary=2oS5YaxWCcQjTEyO X-Mailer: Mutt 0.69 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk --2oS5YaxWCcQjTEyO Content-Type: text/plain; charset=us-ascii Hi ! Wouldn't it be a good idea to put something like this onto the FreeBSD WWW server and of course onto other servers as you like ?! In usenet I saw today questions about FreeBSD SMP performance. So I did the little work. If you come to other conclusions than I after seeing the figures, then please tell me what I should change and why. I'f finish the document then and would like to give it somebody, who maintains the FreeBSD WWW Server. BTW, who is it currently ? BTW: thanks to my wife Reinhild who did the nice jpeg figures ;-) Andreas /// -- powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html --2oS5YaxWCcQjTEyO Content-Type: text/html Content-Disposition: attachment; filename="kbench.html" FreeBSD SMP - kernel compilation bench

FreeBSD SMP - kernel compilation bench

I made something like an application level benchmark to bench the performance of the FreeBSD SMP (Symmetric Multi Processing) kernel against the normal one.

My application was compiling my FreeBSD custom kernel. I tried several -j options for make to find out optimal values for that kind of equipment und to make best use out of FreeBSD-SMP.

I got interesting results by doing that under a normal FreeBSD-current kernel environment, a SMP kernel running only one CPU and doing that on a SMP kernel and using the two CPU's.

You can see this kind of tests already in PC magazines like C't, where they bench the compile time for a Linux kernel.

All test were performed in single user mode using FreeBSD-current's kernel sources of May, 1st 1997.

My test equipment

  • Tyan Titan Pro ATX (S1668ATX)
  • 2 x 200 MHz Pentium Pro CPU's
  • 64 MB RAM, 60ns, PS/2 (2x32)
  • AHA 2940, IBM DORS 32160

FreeBSD-current uni-processor kernel
jobs (make -j) time real [s] time user [s] time sys [s]
1 262.09 189.47 15.21
2 215.58 191.09 16.24
4 213.17 191.39 19.49
8 213.24 191.91 19.40
16 215.19 192.10 19.31

FreeBSD-current custom SMP kernel 1 CPU
jobs (make -j) time real [s] time user [s] time sys [s]
1 269.95 189.22 23.39
2 222.68 191.38 23.03
4 218.24 192.00 23.47
8 217.93 191.35 24.29
16 220.15 191.58 24.26

FreeBSD-current custom SMP kernel 2 CPUs
jobs (make -j) time real [s] time user [s] time sys [s]
1 252.35 144.69 75.06
2 137.06 169.23 56.46
4 119.95 176.02 53.25
8 119.08 175.47 54.97
16 120.03 178.42 53.32

Some performance figures


Figure 1: real time in seconds

This figure shows two things:

  • The FreeBSD SMP kernel with 2 CPU's is about 90% faster than a normal uni-processor system. Yeah ! That's factor 1.9 ;-)
  • The benchmark results of the SMP kernel running with only one CPU are a little slower than running on a normal kernel.


Figure 2: user time in seconds

Although the SMP kernel with one CPU is a little slower, you see, that user time is about the same !


Figure 3: sys time (kernel mode) in seconds

  • Here you see the little system overhead of the SMP kernel comparing normal kernel and one CPU SMP kernel.
  • If both CPU's are running, the system time increases about factor 3-4. This must be the about 10% you are loosing if you compare the real times.
  • But no matter, 90% faster is definitively faster ;-)

Andreas Klemm
Last modified: Thu May 1 18:53:11 CEST 1997 --2oS5YaxWCcQjTEyO Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="smpbench-real.gif" R0lGODdhuwEOAfcAAAAAAAAAewAAvQAA/wB9AAC+AAD/AEpNSmNlY3sAAHt5e4SChIyKjJSa lKWmpa2yrb0AAL26vcbHxs7PztbX1t7f3ufj5+fn5+/v7/8AAP///wAACIBg3ADwzgAXvwAI 74AAAvCwABERADAAlLmw8BEREQGUG+zwthARAAgAAC4bpLm2zhEAvwAA7+YboBa2HwMAAwIA OACwzAARv24EswAACsAAJAIACOY1CBYuzwMPv+kA3BawzgMRvwIEYAAAUAAAJYA1BvAuABEP AAAIAEgA8M2wzg0RvwwEigAALAAAJDAACLmwzxERv0gE3LkAzuE1YAsuUA8PJQgICPTIbsC4 ABIRwAgAAgNoXgDNAQC/AADvAAAEALgAABEAAExgeALwzmAADPCwzxcRvwgA73wAA86wAL8R AO8AABSUzs7wdxIRJAgACAAbDNC2AACIALjNABG/AEwAFAKwAAARAGAEGvAAABcAAAA1ELgu oBEPEcj8lM64zL8Rv+8A77eYzi3NdwS/JADvCFYEYC0AUAQAJQBg/wDw/wAX/wAI/zwAAM+w AAAAKJCwABIRADyUAM/wAAAAOJAAzhIAvwBvGQAxJAAPA/gAAECAGVbwJAMRAwBsYJC68BIR FwBgLgDwPAAXDwAICLsuUgE8AQAPAA5SKQEBAAApAACA3AHwzkDwcM+6zzwBAs/sAL8QAO8I APjmXM66zXxgAlbOAAO/AAAK/5AA/xIA/7tuXAEAzQDAvwAC7w5gZwHOzQC/vwDv7wBrDADO AOgAvtHOOb+/D+/vCABvXAAxzQAPvwDOsAC/EQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACwAAAAAuwEOAQAI/gA1CBxIsKDBgwgTKlzIsKHD hxAjSpxIsaLFixgzatzIsaPHjyBDihxJsqTJkyhTqlzJsqXLlzBjypxJs6bNmzhz6tzJs6fP n0CDCh1KtKjRo0iTKl3KtKnTp1CjSp1KtarVq1izat3KtavXr2DDih1LtqzZs2jTql3Ltq3b t3Djyp1Lt65drRUYCAQA4OIEBRP5alBAYeaCw4gT312MVbDGAxEC941wIOZhCwssaN6sObHn z4gZiw7KV7BpAAoAQG6geoJAB3wbFIwAoIIGvgoAa4ANQLaGCwha3+5bAYCElog3Z+bMvLlz 5aCjLxhNvaXj0xQw8M2uejeACxcA/ugdmHpv7dffw+tFcOF25eu6U15mvvy5/fsA7kOXDrq6 /4vX9RWgecOV1tdAA5pnoAYYPJCagAc6dtJ8zdWn34Wb5Yfhhhbkx19//4U4HIEJnoZQgiMS FFxxEBJYUnL2WcghfjNiqOGMH34m4lsDltgXaxKoR1B5KWoApJB8hdfibfGFROF9Mtbo3I1S PkdllVNylqOOO241wQHGmZiiY7yNNxBthUnonXgCTQAAbyNSYJyTT+oXJZYZ4mmlnnviKV2X bx3wgEUPVPZRnRjeqeeVfHbYaHOMNjodoG39ZZECrnWE6IaK4hnpoo8y96mek1JqqkYwYtkp lqOyGipn/q2qeuqsFh0Wa6KvanZrjbvymmtntAb7EIy9QvlrsRsim+yvmAnrbEJ1KpuhhqtW Ka1+12LLbKnPdruptHwJIABf1UqZrX3norttt8+mCumM4Yorr4H0enoss+k+xy27pm4qarIA yCvwvJDSa3Bp7+aab5a/7suviO7SiG3AA1c87rUHZ2zvvQ0/3KW/6qJLscUk88XnpxkfnLDC 63r8H8gSF0zyzAKb7OqiKRu8ca4Ou2xXxPDKTPPQBFvL8ZQ5G7jyqz37LBfMy05L9NTzgnu0 p0kj7KfTdQHNoc2OTjsy1USDbePVoaKctdnKcf10ufgBUEABJlMJwAADxEt2/tlfo/1ovgY3 7TZa82Es99yIm3033nmPvXfJuy78r998Cj74WO4qyxfinCeeH1+MM67345BryzLlpF6e1rdf H97563SDHrrojpM+MNuwog4qz6qX5bWuNroO+/CLzy667TSjrPvOoVreO1ZQhy3y8NRzLrvx jSNf+uSvSs59889/FT3wVgpf/fnXGz+69lXnfnr3LYef1e99imr++fjHjj3e67N/sdLwex/T 5Ac9uC2tQ/fLnwKLh73+ka5/OTPX8mRFwKnQ70JXKo0CN2g9BqpPew4cWgQZFkDeVTAmpUGQ hFI4LAPGTFcGMMDmOMjB9M0uhDPDIQiTlrb4ndA6/gi6TRAPJESGjO9ssIphDGdIwwV6MHQ6 rJr/RFg7ca1Na/Ub4A9lIqAhBhFaR4xahpSoxAQ2kXo2hGIVrbjGKbLRjf9b26Oct8WSXMeL e0FIGIOWRDIu0YxnhN0T1Xi7NroxiuzD4RVx17Y6wqSLeSQiEQvkQgn20Y9/DGT+0khIRPrP kFME5d40RkdHkgSSQpSkQbyXxQ5h0o9M1CQaB0k7A8GxZqJM5C0tZrJSmtIjqgzmFwnCShKW 5pWYBKQsPbc/9aXsgbnc4S4LGQBy/XIlLCyQCieJIGbBSmsAQGYylbnM2HGymWrMWMU8qctp 1iwA8PTQNYFSTPvZTZyv/oxlOecWS4Oh85+1VKc7o4k8AMAzns2ap0+KWTcN3ROf+dwnP8lp TgMB9KIBPRg03fnOg1YzoQrliff4AgEI2O2SEIUlRdG3UuLRC6MwTadAa8bRjnpUniEVaaNI WtKS3uihKc1nSzunz3L6M6ZIlWnO4MgXjx4UpznVSbp42lOf5mmMQRVnUas3VFlmLalgzWjS cujUm4I0qjjJFlWr2lOHojSrEcXfViX6urlW9KthBWvKympWX6KVJeACAFsHa1WgwhWZdrUr XT1H10WeM6/84+tTz/rXmiBrrYRlq+IOm1J9Knax/ARtXQHpWFqqT7II9WtlVRI5wWb2tSZ9 /itnESu3z4LWtsvsqlwzdkPUflS1q0VJrDAL28Eadrb4NJBoibdcoja3g71FLVSDS5NREbe4 hP0pcjl7sNvqtom4zS0BPGhQ6VKWuihEGnbXa1VXbve9MjRYIMPr1ed6jgDjja5kp4vejWRz Qdpcpf3YS+DYhhO+CCajfOVq38Q1mG74xS95fcvf/mJElUXMo4aJOeACs5cvCQ6xUOnVwQfT V5MAiLCE9VtWalkYJKjcMDdvI7YMuNbDxf2ciHes1Yw197tnTLGK8wtF6QIAuC9+CITuuOFu 2gwANsbxekHM4yoHNWtBfjCEh0xk0ZkXyUlmyIxROUkA6ioDUZZy/o6tzGb4Jm20JhYylye8 3/OG2SIzTuUwuwkrNKdZza89cJsHLeJFLlbOXO5yZOsM5jsfpMx4zDCfM+TnPwM6u4TOtKb/ 6Fg0JlrFdOZrhR0tEXopaJuPFlWlLX3pqgp607CONW0P9mlQs9ipoyb1STK46hu3WrOyDraw tVrrCIe6xXbWtUl43etfY3rY0B42oj99bFwnW9mnVPWqWe1sk0b727GedqKr3VdsY1Pb2/Z1 t18N7nazWdxzLvKXzc1aSG2b285mt7v3HWJ4D5mB5WU0vVPCKCjfW93r5rfC+13sFXt53gPf tb3vje9f63vhGM+qv239cIFHfNkTpzjC/vOd8ZLDdeMO77ior/3xjhSc4n4eeasvbvKaK7jh 5C53y7ONbpF3G9g2DzoscX5razd65xFhNswrPnOhO12GRJf3fo+MdIf8d4VqmvSZl45mmTf9 6TZHuaLzRmGWV13GGSazgKfF9Zj/3NVgr7nYcz7Zo0dc7RjmMAI71Ha3v93bcS/53Iuu87M3 BO97VhCl+870S9M88OAevNRXbnd6RyjSZWabwRnvdcdDfuGSVzmyK4/tvKu9IErn/N99+nl+ h37RHjd8QiAdycTTuM+M9/vfH996Wb+e7BCX/aMBfPU8p171q+d972H9+4BTXvgfOT7yd7/8 bze/7KSH/toX/p973b9d+dUf9PWDr32NSD/3nQc0+MNv5fHHvvwZOT/6Vw949jM/6qI3Ovz9 2/Pud53+62d/hYZ/sPd8+xd//ed/jad+Arhp7meABwggCaiA6admVNaAbfaALUZ1ESiBuKeA zfZzPGUgGFhlGohr2ReBLweCIWhxmXVMJZhgJ1h3HXhhIceC/zdzFQiDMYhcM5haNeiBH4iD OWiBHkaCPXhyBAh8EFiDmRcgY3aDRLiAL1iBsMWDSUhsxUZ3vxWERSJpebcX9jOFLfhhIoiF WVhGS+h8+ueEaYd6ifdTn0OGlWaFPmWHOIaGSfiDXeiFksRCYah4W0eHRXiF9Gdc/hfYg3yY a/BnfG+4TRM4hZ13XYd4h4kogItodtDniIF4e9xHiIXoapX4YfTCfpmYgspmeo+od58IiqxG iaOoWeqWMnF3il4IhlhnfJFIhr6Gh5XoixljcrZ4ixQhf3QoWLAYi6JoccEYeWtIfsQIEcZ4 jL4IgNWYhzmjaWKnaGzYhtGYdLsIigaijMZFjm2VXUkjgw3Hjdj3jaUWjq7YdaVBjtfoeBaY jhC1jQBHYQuQTe4oZvAYj3WYjI5Xj+r3iwYzdATYjTeVZ/+oENMokAc3j+tmjndokZeoj5PX Yg8JjkMokd1HkQeJkRZpUgnAbhqZf0/VkR7ZiiDpf+NY/mAEuXslCQAJcJIKWWt0l3UsuRAR +ZKcJ5KBVpImVZM3iZM3t4WEx5MiQXsXdk0/CZQhSYkGOZLmaJNHiZLPyJHCRUzm5zJPiGrb N4hSKY4UOZM0iZFHeZNaqZQbWU1iVny1t2R/eHlfFCF3RJdi6ZDBskKYN5Z8UZYvSS8kqZZr iZRq6Jaix5deWXuRhJdFhHh4FEx2qWeOyS6nZ5mSJoZ7x3eCKZUZs3tVKWVYeZhtqZMsxpiN qZmnZ5cLknnbZBrDRJcxhpmzuYqQ6JKfKZgHY5WGaZo5SW36pZqrKZmReZkxBkncpIrLCZa3 GYZYpJu7OZ3ySJjYNZoWeJhZ/hmc4xZdTHkif4lhlweZkFaZmWmc/MKctreC1NmeSxea50iU pamdpymc8kacxXlqj+iaXUR7eGmed8lkD6OeTcaZH+meCMp18KmM80mf3BlvRYafQRiW+ol6 UpigGIp+KSOC2gmcSamTACehPemTAZmhJhqUCzplHeqhiamT/QhgIyqEZHmiNOqKOaNZK8qi S4RziSGiMQqRJVqjQkqNpZGjOgp1W9ijP2qDBzqkTgqSDZqj9TluiOGjS3oiQfqkWsqCUbqi UzpnpnalFRGVW1qmMGmka/ml/9aPYsqk0mmmcEqBaLqdH0ptVtqmqdakcbqn6DenbPmg/4an bjqj/nxaqH3XpQ4KqMYmqIPqmYb6qG2HqEeKpON2pyxJfExJppBaqJKapsmkkxAJiHAIowU6 pqt0IlBYmZs5oPlpoboSmJsaqwfnp/rmb4xZnn+Im5b6aKjniMgJhk7TnKXqVo4qq8ZqY7T6 qXPmk6uUq9CpYaKql/pZnqcKh5e5qpgpqsMKNpt3rLLaqXRapxJ2eNTan3umnEMEoMJkrav5 nO0arLqqed56rOD6p9zpo3mpZ6RaJKbGnK9Zre9aIKq6q5SSmVpXrPMKqfWKmC06rkmnr5o5 qgVKoOsJsHEYsLb5q676pgkbpwtbq6BmdV7pmthqmev6hidreyjLriXb/i1yOZaE2rF7+rHK Ol7SqK3jCZ6RKaDiKaAqG2Bj5o9bxJ4yy6nJqpAEy6hAmqVFW6M0O3RJq7Szd6FNa6ZPe3NR K7V5qqdVq6Ufm4jfqbUyirBdW6Yfayv7Krbmx7Rli6Fnq6RqC0xs27YI+ra2ErfRN7d0S50L e5JVird5y7V7e6J9azCA63J6O7ifWbhserhy+02wqrhO66cnmbV4K7QGihk9ajCSm6CFa7lq 24k3siUZ07llWbiOa6rD2hlbEh0HY7o2Squpq7qremSte7svSi+wy6VzCrqXe670grvCi7a6 u7sKiqa++7vDGpsGMrzSYbubW7zG23XIG6o+/ptKQgu6y5mpqdpkyeszouuTByO85EK6ryu5 nXqrI+us5/qVt5mfKfu94Bu2pRY4ruu8nnG+VSupqhm0KqSr0JqvMKqtFjtmkbaz8ktdKYO/ OaK/9Oql5DqyEOu9x/mYaJey7WrA12qwHZgxDMwfpauwHYqvOduf2VuXsnnB/2qxtUvAOxuN xme/DAy9+cu5cBqllmqu5sqrqrqKtanBASusFCymHky6uBvCk2uaIjuZ1yrBK3vBB1yyGEyt pYqnRVylH1ylNly3SqxkONvEdzmt7+vCYUpM1yuwYpm6V5zF0WsguzmfCTy7OksUKFwabPwZ Dhy7bCnHyxbHy2a9/sE7w3i8xUSIlX7Mx058FWv8vK2LxJx3koi8a2Exvj2axXnca5AcyXYU Epi7UJR8x7nrxvKoyX8cfT/7E3Vsx6CctqQMY6d0FP0rw857yK2MsS6nFYvsurWsFLRMT5/c uLsMyyPRy6dUxsE8FLRMv6RxzEZhzP7VskJBzMxMx9A8zfPUw9YcVc7scsqczV1yyJ3ozaYC zqcszqbUyeb8S+Gczo60zk+ZxuwcIu6MZ6oYzyKCzgCCm/bczrG5z/yMdv78Q/Uc0BU00AQt PxQKFto7wKwMyCisIA5ZfGqCzwe9E1mbddisZInMmgA7sABd0SjUvSRisrmYxuQZsSRd/pcU DMTAutE/DNKPFJ4koq4y/cLSyrMoG6YTncGjSnwwbRMrzNF1DNEuwtEoPdT9msEefbEAWqE/ jU01bdTvC637uZmBKMS129ITW9W2/M0R7aOWknQ53BeEwSMj7Zgp/MRpvcMpLMQnm5zAGr5U /dHSDBfdTBCQUb8WQRl2vcM2jYvsC7QzfbEmnaoPjcaz6ZeJXddmQSZ84QDm8QAAwB4aQAEP wgDtcSbngRuqEQGscQCuARzCIRhuAtlr4hvDpwHFcRxc46vaJxiwMQGlPRwVUBx6ASYYEB5N QiTboR0AwB2VQdndwRez7R3gwSapzSSDczD7JybOTdZ8wQCs/v2/YwKZDfIgAsvahpvcd/3U TPHc1S0QFfAAwYEAZuwiA7Iip+EAB2DeX2jGS9Ld3s3LfRHbs80XElAcspEa7dEd5AGZ4Z0k JgIbtnEkyN2se9Ek8z0Vjv0mkT3Z7YEBrIEamS0QaBLeI1LaJiInpl0ms9cXcjLdC87ghoIR grJrlVEoI14VDAAAmYIRYU0SLe4amOI2/hjDh72XDb3ibtGzHV3Fz8rjdSFMQFybU13NQq4W Bq3CGH3kSR4XS67P26vYTy4XUX7VOsvYVf7dTv7EWp3RW94WN96YL4vA5RzmaJ7mar7mbN7m bv7mcB7ncj7ndF7ndn7neJ7ner7nQXze537+54Ae6II+6IRe6IZ+6Iie6Iq+6Ize6I7+6JAe 6ZI+6ZRe6ZZ+6Zie6Zq+6Zze6Z7+6aAe6qI+6qQOFQEBADs= --2oS5YaxWCcQjTEyO Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="smpbench-sys.gif" R0lGODdhuwEOAfcAAAAAAAAAewAAvQAA/wB9AAC+AAD/AEpNSmNlY3sAAHt5e4SChIyKjJSa lKWmpa2yrb0AAL26vcbHxs7PztbX1t7f3ufj5+fn5+/v7/8AAP///wAACIBg3ADwzgAXvwAI 74AAAvCwABERADAAlLmw8BEREQGUG+zwthARAAgAAC4bpLm2zhEAvwAA7+YboBa2HwMAAwIA OACwzAARv24EswAACsAAJAIACOY1CBYuzwMPv+kA3BawzgMRvwIEYAAAUAAAJYA1BvAuABEP AAAIAEgA8M2wzg0RvwwEigAALAAAJDAACLmwzxERv0gE3LkAzuE1YAsuUA8PJQgICPTIbsC4 ABIRwAgAAgNoXgDNAQC/AADvAAAEALgAABEAAExgeALwzmAADPCwzxcRvwgA73wAA86wAL8R AO8AABSUzs7wdxIRJAgACAAbDNC2AACMALi6AEwAFAKwAAARAGAENfAALhcADwA19LguwBEP EgAICMj8A864ALeYAC3NuAS/EVYE/C0ABwQAAABg/wDw/wAX/wAI/zwAAM+wAAAAKCCwABYR ADyUAM/wAAAAOCAAzhYAvwBvGQAxJAAPA/gAAECAGVbwJAMRAwA8YCC68BYRFwBgLgDwPAAX D7suUgE8AQAPAA5SKQEBAAApAACA3AHwzkDAcM+6z78Rv+8A7zwBAs/sAL8QAO8IAPi2XM66 zXxgAlbOAAO/AAAK/yAA/xYA/7tuXAEAzQDAvwAC7w5gZwHOzQC/vwDv7wBrDgDOAOgAvtHO Ob+/D+/vCABvXAAxzQAPvwDOsAC/EQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACwAAAAAuwEOAQAI/gA1CBxIsKDBgwgTKlzIsKHD hxAjSpxIsaLFixgzatzIsaPHjyBDihxJsqTJkyhTqlzJsqXLlzBjypxJs6bNmzhz6tzJs6fP n0CDCh1KtKjRo0iTKl3KtKnTp1CjSp1KtarVq1izat3KtavXr2DDih1LtqzZs2jTql3Ltq3b t3DjyoVbgYFAAAAuTlBQUQGFuYADp8S78UCEihEOyFzAuLFjwZC34iVMGYACAIYbYJ4g0AHe BgUjAKiggcJlAAwuXL6g4YJl03hTa6gAQMJLxhYWWNjNe7fj38AbRx5ulLCGyhQw4E2OWYPn C67tDrws8AAADK4VTADwQMMDABOs/mO3fJfvysa8dfdez759+uDwFxCfH9N45bt5j+edPJmg 8dMM2KbBAQhogIBiAAqon0q4raeeexBGCECE78UXHH0YbmTffvltuOBBxs32AAIAFCjadg4I VMGIJeKHUoPsPUjhjLxNSOONFkxoIXwZ9liQhx4uqJkE0RFEnQarHaeYcnhhgCQArDV3nHkj oQehjDhKmCWNNm65I3A+xhVebfd9aJxnqIUGwF8YaGYZa86lqUGbeCnAGgW1VYmle3tuyV6X froHaKDtDfolmGESd0B3CDEwWkIPKAaSlRDi5RuhlWIqqKabahpfomJS6V+JnCWk3UcwVgoA BBDoyOmf/q+uN2iss8YqH6i4hpRqoXix6qulsdYYrLDD7lbrq7fmqmxGlAq6qq/QsnosodNS W6yx1/q27LYVMVZtjs9GK+63WZK7pbnnZpsbt+w6ZOWxk4kr76/Booujvfeqm2y7/Ba066y9 ziuwtLRei++NB9+4b7/87optjeEOPDCwmCZMocUX68twww4Ta2zEEk+sKcaZGqzxxtt2LGtv AGQQcMgiW1ssySVfuzDKPjarZY0Z9AwyzOPKPCzNndqMc6Iq15xjzz4DHXO6M2dLNJ9H95j0 zsYy3bTT81KMsMlgG101cTqXy7LWPv/MNcH5Rh12sTePLdfVX/OM9tZrR2vv/tQrvz1s3HK3 VbbQS9+Nd970cul3wdkCHnhadPsJaMuGH454q/zBu/irfDv4+FsNdv5w1pWnffmvXWauutcV S93452s1K3qOZ5fOtNprVzv56qpnvDmnjsP+VcezT2671rhzPW3CvGf+u6fCkzX46IyTfrzl iGsedfP8mS129MP32Xe9tV+Pfd61zk47l9xPRn2wwYNPVeTvc2q8+bcnDzTr6hffPsbxk99T ple06hUOf7c7XeLqZ7/nfex/MRLgVegHq6GVD4Hny52l+ufAe2UugBI8CgFnVLwLYtBlp3uZ 8wzIQviF8Cb9wU9+3CU+75EPXJQ7Yf40OLHuEa6F/rZ6YU1muJ+BzFAhFCThzICVQx0mEGYv 258P62bB1wlxJkQ8jhGRmETF1YtiTXRiBlGnQGlNsYBARNYVaRLDLB6ki22zn41SJ8a74S6K ZYTWClnWQUKBcI0lyaIbZVhDzsmRj9ar4w7JmEco/u+R7TsZIF8iyC3+SF0ekxzA7KZItK0K j43sYSgZacYzYuqPkxRJJe9iEPWNr1zaO2AnkQfKUY7LltIK2er8hMpUhiSG+jmiETHJQAnp jpOztGO8cEnKUeqvh6ZsTy99SUliFvNP5KJjMo+XuVDWsozfzN4e00NNn7jylc6iYhi3yc1l Zo+ZrYKnHt23rnLy5JyI/kxnHNfJTvPxx2nhVOAznWmjadpzJfjMJMuYh8x+6rCbXYNnQAma gIIedCcJrR/r9unQbf7TjBKV57gSQFKLXjQnGV3fx0bW0I7OEpIw5Z5IpUXSkmrrpDhJ6Ryn pk2X+vSnqrMlAGpq03ri1CY63ajQ+PnTpnZ0dZcbKlEretOjIpWYSm1dIp3KVa5CtYdTLapB rRpIdW3QbbLsqlrXGtRfhbWmJiUrGw2WOrQyda147Sp/3grXqspVJqLjn13zStjCuoyvYv3r XDmXvsEa9rF6RSxVjapYmBAtqyqt3l0hy9lkSpWvca1sNan1rRJutbOo9axk8TJW0X7EYpjN /qdmU0tbj66Wsq5tycFii84Gnra2wMXgZ9+qo9xa1oNQI99kgsvchyI2tMZFKPsC5b+WNve6 hhvuVKEb3cFcjGSBvR92x5vd5/q1u97Vp1YPKV7yutdn5sUtei8CTGD6h1dpRC71Nvve4GqX qNydL0VWqcVLLrSK1B1Ue/s73v/2Vb4CnogwBznMlaI1ubLlL4Np6+DERpi+bbTkfXn7w+kW yoQbbm6HJ9vaD4u4iKw0Ij3NCkuspTXFzF1xgF38EAIL8zjWRBeJF4xj/8a3xTz2cSuD7EUl WrfIqV0xi3lskfqG6L5Y/W4cbwzlKN8WwlT2SEaXt1sUdxm1Ut5x/pg7ktD0wdbMZ+Zsms+7 5tdmubdb1nCc8zpnMNdZQ3e2MOH0vGe2fhnJf2ZImwW9Xi4X2rB9RnSiFTLmzhH50ZA+9KRB 0ubAwhnTfNb0psVs1gl5+smgxmukR03qJSqUpahOtVpXzWqOhBfPg5Y1ZNNc3FoDGoinBpeu H0trX2eEp7E0JO0IPeynitrYGKFZbC/bO0c326nFhnaVG41GWBbg2+Dm3rW9+mxtVwS25sIY AMDN7na7W9zjfmm5zS3hc+1Nk+7Ot7713b54+3Pe9I7I3srswX0b/ODsxsu74R3vbAdc4FS0 IQkRTnGDr7viCW8eqOcsaXNnU93sw7jI/hM+copDsr8cf/i5jQlrCZV85Ap/OcYvnu+Yyvm2 Hdc2vHhqTJnP3OcVjznQv/3IpuqYziovcDCXXEFlY3PoCBc61Pk9dYu3m3e2BS3SH24fEd9F VsFmWdWtPnaql/3dQV+dc7XuZ48XmMJAZjQLpX72cNcd7XcnOs1ljvXSHb3tOn+712VIY4hN Ju90r3viy773s2fOk0dOuiU75PUZu65GBsi8AXg39Mbf3fOOz/vVw/j3nH84c0r/8aIxr/nW t57zJhe93WUPesUTgPSRl7x/Yqz0ChfeWK4PvvAzD3u9y57ox6+94wlwe8iz3fRJlvGP4/57 vAz/+tjffPFD/k/74xOd+c1HXu51D5HVmzr76E+/9rlX8sUz3vsFAAD4w3+78ZPfIZ2eo/r3 z//Xs//qyad8jDd/9Adfz3d/5RdoOdJ/DNiAwQdJYyeAyzd/TeRwCEhppTY5DriBHJh9MaV8 EjiABFiBq+UtV3aBCNE/c8QyHdiCLtiAj5R8BMh8JPhcCzB9KAgiF6aBL9iDPsiBzTN18jeD BchrOYh/X/RKAPCDTNiEHRiEQUeENCh+z3WESOh0GeaEWriFT7h98SeFU1h/oIWDVmhgFZNs S8iFariGQMgfYBiGBhhWJ/gjIbZ7/PEjGjFhc7h0wUR5g8cw0mZjaciGhFiI/DeE/mBYg9tF hr3HITKWert3bJHYe5BIYJQIiKTlZMBniJzYicKHiFKoiHDFiJcYIpRReTEWYhxiil0HIpP4 Yo0oQ5hobxE3iJ54i5wIikQoihVFii6yRUUEY7Hoh1rUisUIjCn4irwnjMfIe+xCcM6ySwuI i9RYiLo4g6K4hwnRdcHYh3Qofae4jH3IjIOnenWYir4IKgOnKgLQju7ojjFVjfLYgtc4glRY UhJ2jN3oisQojqw0Ya0UkHj4ikeUjmEiZBLyjgq5kAypkB/4kP8zj/T4hkU4ivhnh6l4if8o joMkjOSIg5Y4jH+4LB/Hjg15kii5kHiRkiyJkgDgkLwj/pHqV48USIXamIJ1SHkgiY7E6Ebn OI6tZIzgKH2zqDRg15JI2ZAvmZRMKQAriZQxKZOvR5EkeJNluBBkVilNuZVOyZVQ6ZUwuTrV SJPgV4FWeZVYeWLOApZQuZRseZJu+ZZwGZVrSJY0qDpoSYdEtIcKtjxyyZJx+ZcOKZgtGZjN w4R2eXuOYZD3p2RmiFlPSZgqGZiSGZmSqZRJyT0OmJgA0BiMmYMeyXS6c5kMSZmVSZouKZiH iX2c6Zl5uY2QqJcmiZrwSJvtaJq0aZm5yR8PSJV3+JoCSWGWd5S2eZu4+Ze6WZzHuZu9+Yad +Zko2JFM9yfFWZu2uZykiZ3Z/hkAtrh5FHl70HmBBfmHAKOdcpmcp1md72ielRkA3Ol6iXl7 wAkiv2lfvvdAk5Gb1cmeqqmeDume7zmVzjmf0SYrA3CgB/o/XMmfC8qg5+mf6wmgAap5dnmW BKpoBoqgGrqhGto+k6mcEGqdIQoAEjqhxJeIZFifePibI5mPAxmUHfKT4Tk2CsahNnqjNgqR Oro66omeylmiJuqdoQibwLiXsemMVfaNAtmMTKqRElSjOBqlUjqlVFqlHco7/RmitwmkQUqT jGiOjyicPLmRLCqjpbikzDijVQOlVtqmbvqmVso9mamlW8ql3VmPvnhl3RiayFiJPVmJS0qe fYqk/i/EpnB6qIiaqDfaPINJpyTKpV2KjT2mk/pon6zYH9KJeqoXqDLUj2p6NIaqqKI6qqSK pRD6qJB6pyMoEXvapBhJiZn6ooIqq7EISKFKqriaq6Jqqu0Jqe6pqmWZjuPph+TYp47JpMU6 fSE5nraaobr6rNBaqjzqlaiaqs0JnpPaipTqim9njD4plIQqi6Von0J0q9F6ruiKqNOaktVq rfB5lxfaaiyTrvRar7u6rk7pqxLanUL6qfG6jX9irwI7sIrKH/q6r73pr/+aggFLsA77sG3a rr46Gf63sHY2rxCbsRqbowcLoJ15gyxqsbbmrBtbshkrsakKHAorsjqI/rEm+7IOi7J2+hsr y7LTaSwwm7MDK7NA+rHeYrPyirM6O7TpyrMlujpAO7IuS7RMq6tGu68/m7RsRrJNW7W72rG/ eoNSO2B82RsgixdWG7bqirXcWbMiC3fUp7KrI7Zsi6NP67Fma7FoW1CH8rX80bZs+7Zlu7UD Rp51ayG8g7c6+7YWyrfJ6Iyq87d1G7iC+7BPW7iGy61n2qmTobg78rFr27hFe7Bxa7NzS2mr Y7msFR+Mq7mjarSdq2jKeG5rRq5c+0Hw4bOiqzqmC6c8m7qq26KsSqChK7uWGxylW7sca6ek WKbcWIxW5qljWqmXGq68G7q/e7mZW7soa6FK/gYkG5m92StIfxqS/2qOsBu9i0mztNu2Elu4 1zt5PJmme4mpsKipkUuf4bu4hxK8OXu+F6m9fLqtTQpjAMmRtBq/OJm4wCu+Jpg5JtuujJm+ GdmM4bi9g8rAAsyq0GvAwDu99lqtn/mtfyp4C5KmyyudfIi7VkXCJ1HBFnzBCHyuqAq5Eyxh JmwS4MsfKTy+nonBifqoLvzCAuekTME7v+u7sYvDVEqiO8zDuysZBBy7FkzEG2rESKxKnHbE lETAQlzDpUvFUVx+r6W7PQHENTzEWrzFPTYSMXzCeknDYRy1ZMxpZlwVKCy6bUwUZ4xRcQy8 czwUdezDObXEWpvH/kGxx2PMRqgHyOYUsrbGx4FsyGVRrEWxx4z8E/0YyeVUyFMMyZTcyFLs vJnsI2eMtp1sq4McynIDyqRcqF58yvJjytE2KpisylnhuscmwrC8ykday6jcqbgcQrS8y9HT y74MO8D8FWZryZaMf8a7IdqYvHo6ym5HlGSxsic4yVw8qMiKpojrj8HsEm1kZfoornwYzjH6 za+avNm8qaj4qtrLydtMEgxcJhLsvy6yiuMMqP9YyHpKkCuqou0MWO6rv6lXn90cwesc0Jh6 h8z6iwkNqx5s0P2MUAR9rNoseD7J0Ej6v4QKwuUYmxidK8v8mXtBwTCMJH8RF+O8Sg+M/tKh 6b/yPIkS3L+37I+0/MpsscOGIdKIISkmvY/d2r+s2K3E6ogX7crQbMzdnM/gnNCynChnghcp oh/fgQB3AiBwIhCiQRp1ghkRoBkHwBkXQCJdvSAo0hmfsY15QRsK8jjKymqE4RkTMNZ4UQG0 YRfikR1GEqPMsSbKoRhSrST6MdZxAh1yEpTl8ct4ydZCDc95kSBKGiSEgQEPcBp9KCCHTdgf 8tBUodhmkh8rQiIF8oib7SIkQhsc4gAE8oswGtqYndl54dZwXRu0ARpJIiUCcSSOzSGu4Yie QRpDUiT0WdirPRVNDQBPjRdRzRp08iYEIRp/cdsagCL3gSdPzY0m0vHbeJLWwf0UtH0Ri3LC ihEp2S0VjlIqGBHSJTHeSELeao3UejmURA2U4S0X34rNGimm8S3fyIjOrgqLqXzfZzHM3DvN /O3fcAHgG+3KrEzgaGHgQ82tNK3gScHg18zfDw7hSGFfzGrOy6vIFt7hHv7hIB7iIj7iJF7i Jn7iKJ7iKr7iLN7iLv7iMB7jMj7jNF7jNn7jOJ7jOr7jPN7jPv7jQB7kQj7kRF7kRn7kSJ7k Sr7kTN7kTv7kUB7lUj7lVF7lVn7lWG64AQEAOw== --2oS5YaxWCcQjTEyO Content-Type: image/gif Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="smpbench-user.gif" R0lGODdhuwEOAfcAAAAAAAAAewAAvQAA/wB9AAC+AAD/AEpNSmNlY3sAAHt5e4SChIyKjJSa lKWmpa2yrb0AAL26vcbHxs7PztbX1t7f3ufj5+fn5+/v7/8AAP///wAACIBg3ADwzgAXvwAI 74AAAvCwABERADAAlLmw8BEREQGUG+zwthARAAgAAC4bpLm2zhEAvwAA7+YboBa2HwMAAwIA OACwzAARv24EswAACsAAJAIACOY1CBYuzwMPv+kA3BawzgMRvwIEYAAAUAAAJYA1BvAuABEP AAAIAEgA8M2wzg0RvwwEigAALAAAJDAACLmwzxERv0gE3LkAzuE1YAsuUA8PJQgICPTIbsC4 ABIRwAgAAgNoXgDNAQC/AADvAAAEALgAABEAAExgeALwzmAADPCwzxcRvwgA73wAA86wAL8R AO8AABSUzs7wdxIRJAgACAAbDNC2AACMALi6AEwAFAKwAAARAGAE9PAAwBcAEgA1A7guAMj8 AM64uL8REbeYAC3NCAS/AFYEYC0A8AQAFwBg/wDw/wAX/wAI/zwAAM+wAAAAKMCwABwRADyU AM/wAAAAOMAAzhwAvwBvGQAxJAAPA/gAAECAGVbwJAMRAwA8YMC68BwRFwBgLgDwPAAXDwAI CLsuUgE8AQAPAA5SKQEBAAApAACA3AHwzkDAcM+6z78Rv+8A7zwBAs/sAL8QAO8IAPi2XM66 zXxgAlbOAAO/AAAK/8AA/xwA/7tuXAEAzQDAvwAC7w5gZwHOzQC/vwDv7wBrDQDOAOgAvtHO Ob+/D+/vCABvXAAxzQAPvwDOsAC/EQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACwAAAAAuwEOAQAI/gA1CBxIsKDBgwgTKlzIsKHD hxAjSpxIsaLFixgzatzIsaPHjyBDihxJsqTJkyhTqlzJsqXLlzBjypxJs6bNmzhz6tzJs6fP n0CDCh1KtKjRo0iTKl3KtKnTp1CjSp1KtarVq1izat3KtavXr2DDih1LtqzZs2jTql3Ltq3b twQrMBAIAMDFCQo2KqAAt69flnU3HoiwMcKBv4gTi6wbuDEABQAGN4g8QaCDug0KRgBQQUNg z3YpQAbA4IJlzHQf560AQILi17AzfnZMAUPd2pE1XL5wgTRByKmDHwCAoXfe3b3n1u1MN2/s 59AfzrY7PThjxgSrg9YwmoFr0Ne3/g/83HKB+fPoo6tnWl27Y4Tu7Wqo8AABAATix8sPztK8 hQUWBChggOgVaKB56yV4U3vyVTeZBMn9Jt9wF1AQGGSm5fZghOR55lxK/hE44IgklijggSgi qOCKIU0wnATviffZZb4RtBlfLgLwQGAYTPaYabrVNVd+Fn5n0nkjAmjikkwCwOSJKR7I4pQq HfDARg8cdqSSJHL55JcBOgnmmE5GKSWVaFKEl16VjYTkkl6O2aScYIpJp5kGpqlnUCGSeN2f gNopp6B0mkhooSUSimeBezaqUp8j1iXApJRWaqkAgWYaKKKJcjqnp/+h6OioGr3pJwCXpqrq qpKmqimg/p+C6qesJi5A6q0UQTogY6v26murvgb76p+0hllsl7gm25CpkaIa7LOWAgsttNJG OyxjsR5LoLLcHuTfoRZUO+2vzo77rLjmUnottsfa2u27fRKKbrqXzktvpfbei2+51q4L7onv KstsuAJmqu++B7PKb8L1LqxvptsGTKquBRc8wMUYD+Cvw5Pm+zDHDHcMcsiYiumuxHoOHKnF GbfssssbGzytxwmPHHJdAZiMcpoUdxrmy0AHLfTQMcvMMM0HAxDA0jrvzGLPBcc89NRUVz11 zKoinfTSTEfstHoq7wpAAWSXbfbZZBcNqNVst/3yuiRHy3XX/30dHdRRo633/t58F1AX31K7 LXjVw6aL89w5e223YniH+3ffkAM+duSU+33t4JgHXbjciNN98uJ+hZ135aSnPXnppT9u9rCZ t94yoJ1z3TTocDXOGOqkq4475brnrqnrmCsde+J1097Wef86vnvlvS8vufN9+ws8zMMn/rnx aMWb6OnQ791892d/D3743JO/MeHVl4l9WrrKm+n4lsMPuPzR0/97xsLHrv76ZIUtrwEADCAA Y+a98tHPdAfUm/jkx5jXpa94/AOL7SIlwApa8IID3BgDDZjABTKQAAQAgAOHt78IckV0u6Ig BlfIwhYagIAFTKACZUg+EIZwhJ2bnQmx0riV7cqF/kAMohAzeC0aehB+ALAhCEWIPxIqbodU 6eGsfjjEKlrRikXEHQc7qMQl4hBxOoQiVFCYrXBd8YxoTOOw0HbE8dWli15sov7qcj0xtgQ7 qdkPHh0ixTICII2ADGQg/XXAJMIxjhjLHxifaMeV6NEz44EkXRhCRjKpUJCYzGQm1Sa5Q9qQ iYl0IgQbCRPqRPKRCemjJamoyVa68pUV1JQnlQjKiylybmEk5R0lCUlUGkSVg7okLIdJzFca cpY3lGMOGalLlTTolJHUDwDipK3/FfOa2BwkMhFpS1HWsZko8aUpJ1mQ5NHKmtlMpzqDeMxZ 1vKWuGQmOEuyH16Ok5fj/tHWFAu2zn7684Lt9OQ7HzjKeZrkT3mU5kHMeU5W/vOh6wzoIQfq TYMChaGyEtQfIcrRdEoUjhTVnzwtuiB97tOMHU1pMT9Kyy8usqAkzQlGQaXRbKrtOioFJEs/ 6dJ4wjSmJTVpCvl5RbVB4KhITapS67LUTOUUiDvlpsYI+k2g1mSmnqrpEK+j1K569atJZSpY u+rUpwYwqskMZUWtqhOsckqrLQTUWOcKVrHSla6a4ihaaznVtbIVJ25FlJgYw0LGHNWud00s BACg2MaS9aaE1ek2+QpP2Y30r6UUqtgGC1C7ItaxY/0saB0r2ruqzYJ77allf4rZmASWTnVx /qhhkVra0TbVtrhlLG4TG1kBplaZy2Rta1/y2jo5yX2i1e1uv6rc5Sq2ts4NawISsFHfTla1 nhtuUI8VW2M5rrTQ3W14o0tW8s4VANOlLmqvC9yXVlW7Linu9prV3Nual7b1vS9z86vfxaI3 vdU9K3vVKlLhwhcw59RoeMdrWwbr18HO/W961RvLAXfTrwd2LU3tNNu68ne/kO0vc0XsVQlP OMADtHBfRfreDDvyrRz+cFjVloEa2/jGOM5xBjRlXggvty4TPvF6kUnZ9E3TxTN5bXcdF1rG 6PjJUI6ylHH8qsbK+MFBDjKKX6jiynatxUgO56Bi7GEATPnMaE5z/pR5PGMSkzXLWh6yO7Fr vTBr+EtLrq2T1cznPvuZypDtsJVNDGcKW5fIdC6hne/4pDzn9zp/jrSkJ83nmxY6y1v+LYHd u2jiLsnRTTUzpUdN6lJLmtCXznSXMdzpitQzPOBZ6HzDpFxIm/rWuM41lFFdaBSjNa0XDi6Y W82QDr06mtlpVtQOWxddO/vZzuY1nH29TWCvmNPEpsg962lPg8R42XuGtrjHPWppY7rCqy7w sLOtED3i0Zf5dHS4yU3vevfZ3HE+NKLbu9p1sxs+C+22NBVs74IbXM34FrK+58zv7P57Itye JLzpEqmDW/ziT044gNG970371N8PLyey/rft7Ypj/OQY1/h0qZ1uYYdcIq8mzx6TvSuU2/zg Kjd0iluO7ZeTRF43D3q9c87yjl+75z5fjMmFzvRnE53jDA+2y5M+EqA3/eq5fvrCBdpwh1M9 JFbHuthJrXJVW9jLxAP51yG+9LG7PdJll3PUj/7xtYO97W/PO8IvrfCtgzTRl7W7RcKu98Kf Oe5QnyjgDSx4V+Pd8JDXMeL93sUis7rxsnl85Ddf48kL2Oh0lx0dMe8RwnOe857f+dxD73XS a8T0p4986rkMerTn0vWD13zsCz97TUsd6bi/COx3z3u+bzzxf+962oP/et0T/+295znwma9t 5z9/7NGvPVWp/o+R4V/f7dlfve0Dz/1iW//7Vw8/15Wv6PKzveboL77xda769Xs8nmp3f8nh H/+8q1/x7Md4+tdu59d/Qfd/yXd/lpV/SRdz7hZxnlGABmhzCFh52HV7ridz0ERONEdrEwh9 80d/tCd+l5eBAtdLyJZPjhMuHwh+IWh2JKhu+idOJyhNBdOC2PeCcgeACrh87keD8MYuN4iD WFeBLdV1GGiCHEhy5SSBRIhzOoh8FoiE5Cd4QJiCFMd/Tyh0RshTVCiAVpgdMtchKjiEW8iF UUh5R9iDSTiAC+F9Z3hxXShVXtaGbpgQcBiHUDh/MGh/v9dvdwhzTqiH5DaH1laH/lUYiB0Y JoR4c4YYUjlkHjOniASohY0oh2n4eXNnewsAgZSIh4N4iU6XifXHg6yXM574ifARiqKoa4/I fqoIEYciaq24h8bXh6ZYh6kYi/tnibU4dKQ4gn5YWWQIdtnxeiSVh784iny4g1N4f8VojFjo asnIistIdsEYVZaHS+22R++GgrAWa/j0TOJoSuH4jRaljNeYddmofdzYbhvYSw0yTiR3bOQo j0vYbbsoRuq4jrc2h9uYM8UWj0xIjuERcwPHhPKIHfsIRf3oj9jYjjGIigMpcQRJTvV4jPeY jxZ5jOlojRD5Z1qniX7YV9HobReJSs80j+MoSY90hR1p/lAPGZKTNpKlmIDddJIoiZGzEU0G aY4eOY/3qJI9+ZG+SJO4ZpPCaIpT1ZC8+IYgiZR7R4p79U5O+ZSVaIZSyY5UOVmdGI5YKYbT AYEzuZX3FoL0t1focZWUeGwcWIbzZpamppRqiTxhuZMjt3+DJZdc2Yxq+EkIdZdiOZTelmd8 mZRoWXQCJYmCCY9BuGSMeJhzKZHu1ImN6ZjTWJaSOWWPyJaq+JKZGZWbKXmUKVCXCR9jSJai OZpUVpogdZrdt5qsaWNdqJOw6RCaOZuk6ZdqqJOwxm3UMYmeCY8it1BjuZHDuTi5qZutyZsk SUt4qJHuVoPJCXBiGHAxeU/4/tRIy8mctOmaS0Sc12kdKfiTKwkeD/iW2vmW+eiWutSd3rlj 4NmQxgaO1OmSp0SYoFmcHtmf2lmdXwOf3vl/bNmTQAmW+YFQQHiQeJmZ6MiedjSL8eln6meb AXeg27mBwJmXG8qfmemhAOo0EjqhlUaZFqqe+ImCF9qeHBqP06iPIApOI0qiaZZ9JzpwGBmT xSmUHOmSRdmf5fSjODpPAqqbNnqbH1Gksxl9IYqkWRmZNIpmZVcXvtWkTgqKRxmlGfeCkoig V9p8WaqlOVZ2BmKlXyprYSqmN0ama3mmpSebS8qldummHaGkrDmlgUmnYKqVarqlfMiYegpx 0Yhc/rTYp815i5YZqDAXmgRSIJnSp2Rqpl86cREYKosSKBM6pYoKcZm5KGaCqZupcTe6qXjp gIzhqZ76qEiZcKNKqvwZcQ94Hah6qaD6i6zqqpwKobsYKLO6lihSq3GIb6n4myKHoJK6nZOY UOiJopiVrIL6J586Taiqqh9obsMqnfpxn7Lhn68Koy+Kebwqib2aIsBKfNLmm4WZrUHIk+d5 jkIKpLDqojxafpkyrtEKKKg3bftYnxiKovqpny4Kr6XakpQafMAJKPZ6IHT0lX/if5h2lQYK jsIZqwoZnAyJnQ4qpCw5gPSJsL0qreSKr0yHasOJoeupod2qoxkarx7a/pLfGpbh+qv2Sq22 uHINYY856on0eJHZiazWqaLlWYNXGq4gm7COKrKFeGLSgY7niZ3zSoPlOJiw+q7LqqzadazO 5LFG+6tIi5g2i6uLgbXOFKTQurW+2qVdC3crJ7ZgW6xF0bFl+7F4Uq6HR11s27YNWhUxK7Mz m7ZjSl14W3VgEbNFa7YM27Cdd7eB27Ib4aw9Ua+Gq7COu7iNm6QvyxMHG7eGq7iUC6SWSxRw K6sz27lGwbkypbXkSrpvGxZ7a7qqK7iL8bZ5+ro+4br5sbq0e1Feuq22S0+5SxYn+7sOCaHC u0OzyxGTW7wJwrkFq7zLCxLN67zYk7zSCzrR/lu9ynm51KhQ2Asd16ttUNu9sUG9nPq94osy sXq+tBO+6iuiedm+Afq+8LszDji/9nu/pHKseXq8xeaux4maFCuW+CtT3dehIYqzPMutLmu+ A4y8/0ue9Jia3Ouj9qmRVvugQBu0WPifPtvAP5eS1gGwPfuTC7mx+1nC2sGs7GmxVLsnJ9mq GrAmEQHD2boXf8GgI5yQebSxQNu0I3eQOKyeyOm5O7uENLwiMDwYMNekhuEXJ5yRK6zBBdnB VKzC8krEAtehaTIjdeEAqbEjCGAaoiEkQCIQm9EZdQEZkkEZGnAB9nEAlREYEwAAXhwkAJAZ qDkfrdEXLLmfjdGi/tsBmn0cxSyqoxxMqSopv71bFoFxGRMwx168HKwxF8NRHI8hIamBGwCA G4cRxp5xGHUByafBGzWSrh7ixEDJkxnsjRFMwhoIrxKMnulbtUa8rlarJzGSy3bRHUbCH/Gh ARjwAKOBnt8BKHl8u+iLsS+nyzIiH/RhH/iRrc0cHPbBGsHpAAcQzS/MzHYTKF/XyADwyHQM GhLAGpmBIZ+MydMcyADQG8EZJJ2xIaUcpM3hwVDBxeMMGmBsGj2SxmWsATeyznJMx+9hIXVM I0NinBpQJPb8FLmREVZyErmRJQ3tFAwQzhohwyRx0ZWhAG1S0SAd0hE0c2SZmpm7w4ssnNJH QZQDm6HaqtJqIcgDq7MaDNNpwb7tKZzya9NogdNCm7kMzNNd4dMTR9MuLdRjQdQJnMVIHdP1 6ZOx7KNB3dRUXdVWfdVYndVavdVc3dVe/dVgHdZiPdZkXdZmfdZondZqvdZs3dZu/dZwHddy Pdd0Xdd2fdd4ndd6vdd83dd+/deAHdiCPdiEXdiGfdiIndiKvdiM3diObVUBAQA7 --2oS5YaxWCcQjTEyO-- From owner-freebsd-smp Thu May 1 11:27:56 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA08099 for smp-outgoing; Thu, 1 May 1997 11:27:56 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id LAA08093 for ; Thu, 1 May 1997 11:27:54 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA04751; Thu, 1 May 1997 11:25:52 -0700 From: Terry Lambert Message-Id: <199705011825.LAA04751@phaeton.artisoft.com> Subject: Re: Where to start SMP? To: smp@csn.net (Steve Passe) Date: Thu, 1 May 1997 11:25:51 -0700 (MST) Cc: bruce@zuhause.mn.org, freebsd-smp@FreeBSD.org In-Reply-To: <199705011749.LAA07241@Ilsa.StevesCafe.com> from "Steve Passe" at May 1, 97 11:49:15 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > Where's the recommended location for starting SMP? I'm doing it in > > /usr/local/etc/rc.d, but I was wondering if I was missing something. > > I think it should be added to /etc/rc and /etc/rc.conf now that the > > SMP code is in -current. > > -- > there is no 'recommended' place. experiment, let us know what works well... It would help posters like this, now that the merge is done, to have an "SMP" file in /etc/i386/conf" for an SMP version of "GENERIC". Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Thu May 1 11:56:57 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA09655 for smp-outgoing; Thu, 1 May 1997 11:56:57 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA09648 for ; Thu, 1 May 1997 11:56:54 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id MAA07653; Thu, 1 May 1997 12:55:50 -0600 (MDT) Message-Id: <199705011855.MAA07653@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Terry Lambert cc: bruce@zuhause.mn.org, freebsd-smp@FreeBSD.org Subject: Re: Where to start SMP? In-reply-to: Your message of "Thu, 01 May 1997 11:25:51 PDT." <199705011825.LAA04751@phaeton.artisoft.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 12:55:50 -0600 Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Hi, > > > Where's the recommended location for starting SMP? I'm doing it in > > > /usr/local/etc/rc.d, but I was wondering if I was missing something. > > > I think it should be added to /etc/rc and /etc/rc.conf now that the > > > SMP code is in -current. > > > > -- > > there is no 'recommended' place. experiment, let us know what works well... > > It would help posters like this, now that the merge is done, to have > an "SMP" file in /etc/i386/conf" for an SMP version of "GENERIC". do you mean srs/sys/i386/conf? there is an SMP section in LINT there. if you mean /etc/i386/? forgive me, I haven't had time to figure out this new scheme for doing things. I tried make -n install in src/etc to see just what gets put into /etc, but it appears to be a NO-OP... -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 12:22:10 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA11054 for smp-outgoing; Thu, 1 May 1997 12:22:10 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id MAA11045 for ; Thu, 1 May 1997 12:22:07 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id NAA07779; Thu, 1 May 1997 13:21:43 -0600 (MDT) Message-Id: <199705011921.NAA07779@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Andreas Klemm cc: smp@freebsd.org Subject: Re: Made nice HTML page about FreeBSD SMP performance In-reply-to: Your message of "Thu, 01 May 1997 19:44:12 +0200." <19970501194412.06645@klemm.gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 13:21:43 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, I've started a page for benches, and placed this bench page by Andreas Klemm as the first entry, thanx Andreas! http://www.freebsd.org/~fsmp/SMP/benches.html -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 13:03:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA12849 for smp-outgoing; Thu, 1 May 1997 13:03:55 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA12844 for ; Thu, 1 May 1997 13:03:52 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id NAA04939; Thu, 1 May 1997 13:01:58 -0700 From: Terry Lambert Message-Id: <199705012001.NAA04939@phaeton.artisoft.com> Subject: Re: Where to start SMP? To: smp@csn.net (Steve Passe) Date: Thu, 1 May 1997 13:01:57 -0700 (MST) Cc: terry@lambert.org, bruce@zuhause.mn.org, freebsd-smp@FreeBSD.org In-Reply-To: <199705011855.MAA07653@Ilsa.StevesCafe.com> from "Steve Passe" at May 1, 97 12:55:50 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-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > It would help posters like this, now that the merge is done, to have > > an "SMP" file in /etc/i386/conf" for an SMP version of "GENERIC". > > do you mean srs/sys/i386/conf? there is an SMP section in LINT there. > if you mean /etc/i386/? forgive me, I haven't had time to figure out > this new scheme for doing things. I tried make -n install in src/etc > to see just what gets put into /etc, but it appears to be a NO-OP... Sorry... /sys/i386/conf/SMP (or /usr/src/sys/i386/conf/SMP). Basically, a copy of "GENERIC" with the default SMP options set, so they can "config SMP ; cd ../../compile/SMP ; make depend all". Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. From owner-freebsd-smp Thu May 1 13:13:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA13545 for smp-outgoing; Thu, 1 May 1997 13:13:26 -0700 (PDT) Received: from agora.rdrop.com (root@agora.rdrop.com [199.2.210.241]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA13540 for ; Thu, 1 May 1997 13:13:24 -0700 (PDT) Received: from news1.gtn.com by agora.rdrop.com with smtp (Smail3.1.29.1 #17) id m0wN2DG-0009HJC; Thu, 1 May 97 13:12 PDT Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id WAA02422; Thu, 1 May 1997 22:00:17 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id VAA00470; Thu, 1 May 1997 21:45:20 +0200 (CEST) Message-ID: <19970501214520.51610@klemm.gtn.com> Date: Thu, 1 May 1997 21:45:20 +0200 From: Andreas Klemm To: Bruce Albrecht Cc: freebsd-smp@freebsd.org Subject: Re: Where to start SMP? References: <199705011356.IAA01945@zuhause.mn.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 In-Reply-To: <199705011356.IAA01945@zuhause.mn.org>; from Bruce Albrecht on Thu, May 01, 1997 at 08:56:17AM -0500 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Thu, May 01, 1997 at 08:56:17AM -0500, Bruce Albrecht wrote: > Where's the recommended location for starting SMP? I'm doing it in > /usr/local/etc/rc.d, but I was wondering if I was missing something. > I think it should be added to /etc/rc and /etc/rc.conf now that the > SMP code is in -current. There are problems, to start the 2nd CPU automatically ... See the comments in the commitlogs or in LINT, don't remember the exact location ... I have created an alias in root's .cshrc smp sysctl -w kern.smp_active=2 This starts nicely the 2nd (or more) CPU. -- powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html From owner-freebsd-smp Thu May 1 13:22:16 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA13892 for smp-outgoing; Thu, 1 May 1997 13:22:16 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA13883 for ; Thu, 1 May 1997 13:22:12 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id WAA05934; Thu, 1 May 1997 22:15:33 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id WAA01639; Thu, 1 May 1997 22:03:10 +0200 (CEST) Message-ID: <19970501220310.30978@klemm.gtn.com> Date: Thu, 1 May 1997 22:03:10 +0200 From: Andreas Klemm To: Steve Passe Cc: smp@freebsd.org Subject: Re: Made nice HTML page about FreeBSD SMP performance References: <19970501194412.06645@klemm.gtn.com> <199705011921.NAA07779@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 In-Reply-To: <199705011921.NAA07779@Ilsa.StevesCafe.com>; from Steve Passe on Thu, May 01, 1997 at 01:21:43PM -0600 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Thu, May 01, 1997 at 01:21:43PM -0600, Steve Passe wrote: > Hi, > > I've started a page for benches, and placed this bench page by > Andreas Klemm as the first entry, thanx Andreas! > > http://www.freebsd.org/~fsmp/SMP/benches.html Wow, fine ! Now I only have to get access to freefall, there still seem to be network problems... -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Thu May 1 13:29:38 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA14508 for smp-outgoing; Thu, 1 May 1997 13:29:38 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA14502; Thu, 1 May 1997 13:29:33 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id OAA08130; Thu, 1 May 1997 14:29:26 -0600 (MDT) Message-Id: <199705012029.OAA08130@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: "Eric J. Chet" cc: "Jonathan M. Bresler" , msmith@atrad.adelaide.edu.au, macgyver@db-net.com, Don@PartsNow.com, hackers@freebsd.org, smp@freebsd.org Subject: Re: 7860 In-reply-to: Your message of "Thu, 01 May 1997 15:59:31 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 14:29:25 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > > what's the skinny on this board. i am looking to get into > > the smp fray and need a new mobo. > > jmb > > > > Hello > Steve turned me on to this board, It's what we use for > FreeBSD-SMP. It's a rock solid 430HX board with an onboard 7880 and > 512k of pipeline burst cache. Performance is very good. The only down > side to this board is it has a ATX form factor but requires a AT > power supply. I believe we are talking about the GA586DX here. I am told that gigabyte is now shipping a new version that uses the ATX PS connector. If buying this board be sure to verify that your vendor will be supplying you with the new version. I haven't seen the original mail (from jmb) yet, but as to the issue of a good SMP mb, I recommend going with the P6 at this point in time. Gigabyte 686DX Dual Pentium Pro ATX $ 365 Intel Pentium Pro 166 512K $ 490 Intel Pentium Pro 166 512K $ 490 ---- $1345 Gigabyte GA-586DX-512 Dual Pentium $315 Intel Pentium 166 $215 Intel Pentium 166 $215 ---- $745 $1345 - $ 745 $ 600 so you have about $600 difference, but for a system that will probably cost you $3000+ this isn't that much more. And this will give you a box thats about TWICE as FAST!!! And you could save a few $hundred by going with the 256k cache versions: Pro 150 256K $199, actually CHEAPER than the P5x166 (but avoid 150mHz, means a bus speed of 60mHz vs 66mHz for 166mHz). This makes the difference almost nil, (but I really recommend the 512k cache). You *probably* could overclock 150s to 166, I have been overclocking my P6-166x512s to 200mHz for almost a month now without any sign of trouble. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 13:47:28 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA15587 for smp-outgoing; Thu, 1 May 1997 13:47:28 -0700 (PDT) Received: from news.quick.net (donegan@news.quick.net [207.212.170.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA15581 for ; Thu, 1 May 1997 13:47:18 -0700 (PDT) Received: (from donegan@localhost) by news.quick.net (8.8.5/8.6.9) id NAA08454; Thu, 1 May 1997 13:46:46 -0700 (PDT) Date: Thu, 1 May 1997 13:46:46 -0700 (PDT) From: "Steven P. Donegan" To: Andreas Klemm cc: Steve Passe , smp@freebsd.org Subject: Re: Made nice HTML page about FreeBSD SMP performance In-Reply-To: <19970501220310.30978@klemm.gtn.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Thu, 1 May 1997, Andreas Klemm wrote: > On Thu, May 01, 1997 at 01:21:43PM -0600, Steve Passe wrote: > > Hi, > > > > I've started a page for benches, and placed this bench page by > > Andreas Klemm as the first entry, thanx Andreas! > > > > http://www.freebsd.org/~fsmp/SMP/benches.html > > Wow, fine ! Now I only have to get access to freefall, there still > seem to be network problems... > > -- > Andreas Klemm | klemm.gtn.com - powered by > Symmetric MultiProcessor FreeBSD > http://www.freebsd.org/~fsmp/SMP/SMP.html > http://www.freebsd.org/~fsmp/SMP/benches.html > After seeing this I did a simple test on my machine (Tyan Tomcat III-D with WD IDE drive, 128M of RAM, and the CPU's are a pair of 75's overclocked to 90). I only did the make -j 1 and make -j 8, but the results were pretty nifty - clock time went from 13:28 to 7:25 nice :-) Steven P. Donegan donegan@quick.net From owner-freebsd-smp Thu May 1 13:51:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA15798 for smp-outgoing; Thu, 1 May 1997 13:51:33 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA15791 for ; Thu, 1 May 1997 13:51:29 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id OAA08276; Thu, 1 May 1997 14:50:57 -0600 (MDT) Message-Id: <199705012050.OAA08276@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: "Eric L. Hernes" cc: Andreas Klemm , smp@freebsd.org Subject: Re: Made nice HTML page about FreeBSD SMP performance In-reply-to: Your message of "Thu, 01 May 1997 15:35:40 CDT." <199705012035.PAA07951@jake.lodgenet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 14:50:57 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > >Wow, fine ! Now I only have to get access to freefall, there still > >seem to be network problems... > > umm, www.freebsd.org != freefall.freebsd.org anymore... > I don't think freefall is running a httpd anymore this is correct, hub is serving httpd now, but www.freebsd resolves to hub. what does traceroute www.freebsd.org show the failure point to be? it should finally get you to: hub.FreeBSD.ORG (204.216.27.18) -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 13:56:48 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA16251 for smp-outgoing; Thu, 1 May 1997 13:56:48 -0700 (PDT) Received: from bacall.lodgenet.com (bacall.lodgenet.com [205.138.147.242]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA16245 for ; Thu, 1 May 1997 13:56:46 -0700 (PDT) Received: (from mail@localhost) by bacall.lodgenet.com (8.6.12/8.6.12) id PAA21162; Thu, 1 May 1997 15:56:30 -0500 Received: from garbo.lodgenet.com(204.124.123.250) by bacall via smap (V1.3) id sma021155; Thu May 1 15:56:12 1997 Received: from jake.lodgenet.com (jake.lodgenet.com [10.0.11.30]) by garbo.lodgenet.com (8.6.12/8.6.9) with ESMTP id PAA15369; Thu, 1 May 1997 15:56:12 -0500 Received: from jake.lodgenet.com (localhost.lodgenet.com [127.0.0.1]) by jake.lodgenet.com (8.8.5/8.6.12) with ESMTP id PAA08127; Thu, 1 May 1997 15:56:09 -0500 (CDT) Message-Id: <199705012056.PAA08127@jake.lodgenet.com> X-Mailer: exmh version 2.0gamma 1/27/96 To: Steve Passe cc: "Eric L. Hernes" , Andreas Klemm , smp@freebsd.org Subject: Re: Made nice HTML page about FreeBSD SMP performance In-reply-to: Your message of "Thu, 01 May 1997 14:50:57 MDT." <199705012050.OAA08276@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 15:56:08 -0500 From: "Eric L. Hernes" Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Steve Passe writes: >> umm, www.freebsd.org != freefall.freebsd.org anymore... >> I don't think freefall is running a httpd anymore > >this is correct, hub is serving httpd now, but www.freebsd resolves to hub. >what does traceroute www.freebsd.org show the failure point to be? I was just confused for a while when some of my old links to freefall.freebsd.org didn't work... I figured it out when I had no problem getting to `www.freebsd.org'. Then Andreas mentioned he couldn't get to `freefall', so I wondered if he had the same problem. >it should finally get you to: > >hub.FreeBSD.ORG (204.216.27.18) > >-- >Steve Passe | powered by >smp@csn.net | Symmetric MultiProcessor FreeBSD > > eric. -- erich@rrnet.com http://rrnet.com/~erich erich@freebsd.org http://www.freebsd.org/~erich erich@lodgenet.com From owner-freebsd-smp Thu May 1 14:00:06 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA16504 for smp-outgoing; Thu, 1 May 1997 14:00:06 -0700 (PDT) Received: from bacall.lodgenet.com (bacall.lodgenet.com [205.138.147.242]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id OAA16492 for ; Thu, 1 May 1997 14:00:04 -0700 (PDT) Received: (from mail@localhost) by bacall.lodgenet.com (8.6.12/8.6.12) id PAA14161; Thu, 1 May 1997 15:36:07 -0500 Received: from garbo.lodgenet.com(204.124.123.250) by bacall via smap (V1.3) id sma014156; Thu May 1 15:35:43 1997 Received: from jake.lodgenet.com (jake.lodgenet.com [10.0.11.30]) by garbo.lodgenet.com (8.6.12/8.6.9) with ESMTP id PAA14968; Thu, 1 May 1997 15:35:44 -0500 Received: from jake.lodgenet.com (localhost.lodgenet.com [127.0.0.1]) by jake.lodgenet.com (8.8.5/8.6.12) with ESMTP id PAA07951; Thu, 1 May 1997 15:35:40 -0500 (CDT) Message-Id: <199705012035.PAA07951@jake.lodgenet.com> X-Mailer: exmh version 2.0gamma 1/27/96 To: Andreas Klemm cc: Steve Passe , smp@freebsd.org Subject: Re: Made nice HTML page about FreeBSD SMP performance In-reply-to: Your message of "Thu, 01 May 1997 22:03:10 +0200." <19970501220310.30978@klemm.gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 15:35:40 -0500 From: "Eric L. Hernes" Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Andreas Klemm writes: >On Thu, May 01, 1997 at 01:21:43PM -0600, Steve Passe wrote: >> Hi, >> >> I've started a page for benches, and placed this bench page by >> Andreas Klemm as the first entry, thanx Andreas! >> >> http://www.freebsd.org/~fsmp/SMP/benches.html > >Wow, fine ! Now I only have to get access to freefall, there still >seem to be network problems... umm, www.freebsd.org != freefall.freebsd.org anymore... I don't think freefall is running a httpd anymore > >-- >Andreas Klemm | klemm.gtn.com - powered by > Symmetric MultiProcessor FreeBSD > http://www.freebsd.org/~fsmp/SMP/SMP.html > http://www.freebsd.org/~fsmp/SMP/benches.html > eric. -- erich@rrnet.com http://rrnet.com/~erich erich@freebsd.org http://www.freebsd.org/~erich erich@lodgenet.com From owner-freebsd-smp Thu May 1 14:06:26 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA16877 for smp-outgoing; Thu, 1 May 1997 14:06:26 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA16870 for ; Thu, 1 May 1997 14:06:23 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id PAA08374; Thu, 1 May 1997 15:05:52 -0600 (MDT) Message-Id: <199705012105.PAA08374@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Andreas Klemm cc: Bruce Albrecht , freebsd-smp@freebsd.org Subject: Re: Where to start SMP? In-reply-to: Your message of "Thu, 01 May 1997 21:45:20 +0200." <19970501214520.51610@klemm.gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 15:05:51 -0600 Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, > On Thu, May 01, 1997 at 08:56:17AM -0500, Bruce Albrecht wrote: > > Where's the recommended location for starting SMP? I'm doing it in > > /usr/local/etc/rc.d, but I was wondering if I was missing something. > > I think it should be added to /etc/rc and /etc/rc.conf now that the > > SMP code is in -current. > > There are problems, to start the 2nd CPU automatically ... > See the comments in the commitlogs or in LINT, don't > remember the exact location ... > > I have created an alias in root's .cshrc > > smp sysctl -w kern.smp_active=2 > > This starts nicely the 2nd (or more) CPU. I do the same thing since I don't necessarily want the 2nd CPU to start automatically when I'm debugging the kernel. But the "automatic startup" problem is specific to having the kernel attempt to do it early on in the boot process. unless something has changed, it should be OK to have something in /usr/local/etc/rc.d start the 2nd CPU. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 14:13:50 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA17444 for smp-outgoing; Thu, 1 May 1997 14:13:50 -0700 (PDT) Received: from cais.cais.com (root@cais.com [199.0.216.4]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA17438 for ; Thu, 1 May 1997 14:13:48 -0700 (PDT) Received: from earth.mat.net (root@earth.mat.net [205.252.122.1]) by cais.cais.com (8.8.5/8.7.3) with SMTP id RAA17079; Thu, 1 May 1997 17:12:26 -0400 (EDT) Received: from Journey2.mat.net (journey2.mat.net [205.252.122.116]) by earth.mat.net (8.6.12/8.6.12) with SMTP id RAA18569; Thu, 1 May 1997 17:12:23 -0400 Date: Thu, 1 May 1997 17:11:25 -0400 (EDT) From: Chuck Robey To: Andreas Klemm cc: Walter Haslbeck , smp@FreeBSD.org, everybodyunix@wup.de, jk@ct.heise.de Subject: Re: Dual PPro Mainboard f. SCO SMP In-Reply-To: <199705011339.PAA02810@klemm.gtn.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk On Thu, 1 May 1997, Andreas Klemm wrote: > [Posted and mailed] > > In article <6VndaKB5SfB@surfer.ods.de>, > WH@ODS.de (Walter Haslbeck) writes: > [snip] > > BTW: Wie siehts eigentlich z.Z. an der FreeUnix Front aus in Sachen SMP? > > Gibts irgendwo im Web Benchmarkergebnisse, die SCO SMP mit Linux/SMP > > bzw. FreeBSD/SMP vergleichen? > > I'm writing this in english, so I can write this also to the > FreeBSD SMP mailinglist. I suppose the people there are interested > in numbers and figures as well ;-) > > I'll put together some nice gif files later, that are suitable to > be put onto a Web Server. > > I made something like an application level benchmark by compiling > my FreeBSD custom kernel with different kernels and job parameters > for make (-j). You can see this kind of tests already in PC magazines > like C't, where they bench the compile time for a Linux kernel. > > For reference: > Tyan Titan Pro ATX, 64 MB RAM, 2 x 200 MHz Pro [some deletions of data] Andreas, I think you really ought to double check this. When I did my own checks on my SMP setup, I found that just running make -j 8 (without going smp) gave a serious speedup. There WAS additional speedup when I turned on smp, but it's NOT 2 for 1. To do the benchmark correctly, you have to run the same program in both places, and turning on the -j 8 feature for smp only is fooling you. > > 2) Using the SMP kernel with 2 CPU's and make -j 4 and -j 8 is > really fast > > The performance boost you get now with FreeBSD SMP > _for_this_kind_of_application is factor 1.9 !!! > > To sum up: you nearly double the performance of your system !!! > > I think important here is, that every PPro CPU has it's own 2nd > level cache. > > Andreas /// > > -- > powered by > Symmetric MultiProcessor FreeBSD > http://www.freebsd.org/~fsmp/SMP/SMP.html > > ----------------------------+----------------------------------------------- Chuck Robey | Interests include any kind of voice or data chuckr@eng.umd.edu | communications topic, C programming, and Unix. 9120 Edmonston Ct #302 | Greenbelt, MD 20770 | I run Journey2 and picnic, both FreeBSD (301) 220-2114 | version 3.0 current -- and great FUN! ----------------------------+----------------------------------------------- From owner-freebsd-smp Thu May 1 14:47:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA19289 for smp-outgoing; Thu, 1 May 1997 14:47:03 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA19284 for ; Thu, 1 May 1997 14:46:56 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id PAA08539; Thu, 1 May 1997 15:45:42 -0600 (MDT) Message-Id: <199705012145.PAA08539@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Chuck Robey cc: Andreas Klemm , Walter Haslbeck , smp@FreeBSD.ORG, everybodyunix@wup.de, jk@ct.heise.de Subject: Re: Dual PPro Mainboard f. SCO SMP In-reply-to: Your message of "Thu, 01 May 1997 17:11:25 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 01 May 1997 15:45:42 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, >Andreas, I think you really ought to double check this. When I did my >own checks on my SMP setup, I found that just running make -j 8 (without >going smp) gave a serious speedup. There WAS additional speedup when I >turned on smp, but it's NOT 2 for 1. To do the benchmark correctly, you >have to run the same program in both places, and turning on the -j 8 >feature for smp only is fooling you. being a bi-noid I always have trouble with statistical type math, but I would set the difference shown by these figures as 1.79 using "-j4" for UP and "-j8" for SMP is fair, as these produce the best results for each case. the time for UP at -j4: 213.17 the time for SMP at -j8: 119.08 213.17 / 119.08 == 1.789696 not quite 2-1, but not shabby. Chuck, you might want to re-try your test compiles, we have gained some speed with recent changes. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Thu May 1 16:43:11 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA25917 for smp-outgoing; Thu, 1 May 1997 16:43:11 -0700 (PDT) Received: from mail.ai.com (Mail.ai.com [204.144.182.5]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id QAA25878; Thu, 1 May 1997 16:42:50 -0700 (PDT) From: rich@vegasone.com Message-Id: <199705012342.QAA25878@hub.freebsd.org> Received: from Dial-up323 ([208.193.202.83]) by mail.ai.com (Post.Office MTA v3.0 release 0121 ID# 0-32332U1000L100S10000) with SMTP id ACF193; Thu, 1 May 1997 17:17:51 -0600 Date: Thu, 01 May 1997 16:15:01 PDT Subject: Add LIVE Girls to your website! Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk To: undisclosed-recipients:; Want to increase your profits by adding LIVE strippers to your adult website? Check out what Virtual Net Productions has to offer: We will provide a Live Stripshow with chat for your website starting at only $500 per month and NEVER more than $1600.00 a month!! (read on for details) The $500 rate is based on 744 hours (The amount of hours in a month for 1 person). - Additional hours beyond 744 will be rated at 60 cents per hour. You may or may not use additional hours depending on the amount of traffic you receive on your website. I can tell you that we have had our stream on over 100 different websites for the past year and only the top 10 sites exceeded 700 hours for any given month, so for the majority of adult websites our stream will never cost more than $500 per month. But what if you do use more? No problem cause we are capping the stream at $1600! You will never pay over $1600 no matter how many hours you use. And we are throwing in chat at no additional cost! So just to recap, get our stream on your website for $500 - no setup fees, you are paid in advance for the month for only $500. Then if you use more than 744 hours you pay .60 cents per hour up until you reach $1600, then you pay no more, and chat is included! Our stream has been used reliably on sites like Cybererotica, Sexmuseum and Newd.com just to name a few. Don't waste your money with untried services that can't match our quality or reliability, not to mention our prices! But don't take my word for it, see our girls in action for yourself. Just visit http://www.mojave.net/stream USER ID: visitor PASSWORD: live213 If you need any additional information ask for Rich or Glenn.. Rich Olson Virtual Net Productions OFFICE: 1-800-314-4285 or 1-702-360-4939 E-MAIL: rich@vegasone.com From owner-freebsd-smp Thu May 1 17:17:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id RAA28244 for smp-outgoing; Thu, 1 May 1997 17:17:23 -0700 (PDT) Received: from horst.bfd.com (horst.bfd.com [204.160.242.10]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id RAA28233; Thu, 1 May 1997 17:17:19 -0700 (PDT) Received: from harlie.bfd.com (bastion.bfd.com [204.160.242.14]) by horst.bfd.com (8.8.5/8.7.3) with SMTP id RAA13665; Thu, 1 May 1997 17:17:12 -0700 (PDT) Date: Thu, 1 May 1997 17:17:12 -0700 (PDT) From: "Eric J. Schwertfeger" To: Steve Passe cc: "Eric J. Chet" , "Jonathan M. Bresler" , msmith@atrad.adelaide.edu.au, macgyver@db-net.com, Don@PartsNow.com, hackers@FreeBSD.ORG, smp@FreeBSD.ORG Subject: Re: 7860 In-Reply-To: <199705012029.OAA08130@Ilsa.StevesCafe.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Thu, 1 May 1997, Steve Passe wrote: > And you could save a few $hundred by going > with the 256k cache versions: Pro 150 256K $199, actually CHEAPER than > the P5x166 (but avoid 150mHz, means a bus speed of 60mHz vs 66mHz for > 166mHz). This makes the difference almost nil, (but I really recommend > the 512k cache). You *probably* could overclock 150s to 166, I have been > overclocking my P6-166x512s to 200mHz for almost a month now without any > sign of trouble. I've had my hands on three P6-150's, and all will make world at 180. I still run them at 166 for the bus speed. From owner-freebsd-smp Fri May 2 00:53:51 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id AAA19209 for smp-outgoing; Fri, 2 May 1997 00:53:51 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id AAA19204 for ; Fri, 2 May 1997 00:53:45 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id JAA07098; Fri, 2 May 1997 09:45:14 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id JAA10114; Fri, 2 May 1997 09:33:26 +0200 (CEST) Message-ID: <19970502093326.00399@klemm.gtn.com> Date: Fri, 2 May 1997 09:33:26 +0200 From: Andreas Klemm To: Steve Passe Cc: Chuck Robey , Walter Haslbeck , smp@FreeBSD.ORG, everybodyunix@wup.de, jk@ct.heise.de Subject: Re: Dual PPro Mainboard f. SCO SMP References: <199705012145.PAA08539@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="/9DWx/yDrRhgMJTb" X-Mailer: Mutt 0.69 In-Reply-To: <199705012145.PAA08539@Ilsa.StevesCafe.com>; from Steve Passe on Thu, May 01, 1997 at 03:45:42PM -0600 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk --/9DWx/yDrRhgMJTb Content-Type: text/plain; charset=us-ascii On Thu, May 01, 1997 at 03:45:42PM -0600, Steve Passe wrote: > Hi, > > >Andreas, I think you really ought to double check this. When I did my > >own checks on my SMP setup, I found that just running make -j 8 (without > >going smp) gave a serious speedup. There WAS additional speedup when I > >turned on smp, but it's NOT 2 for 1. To do the benchmark correctly, you > >have to run the same program in both places, and turning on the -j 8 > >feature for smp only is fooling you. I did the benchmarks in single user mode and automated with a script. Everytime in the _same_ directory /bisdn/compile/BISDNSMP. I used for Uniprocessor and SMP kernel the same kernel config file, I use -pipe -O for compilation. Before starting all benchmarks I did a make depend all and then removed only *.o. >From now on every benchmark ran in the same directory, with a script in single user mode and every run started with a 'rm *.o'. I think more accurate isn't possible ;-) > being a bi-noid I always have trouble with statistical type math, > but I would set the difference shown by these figures as 1.79 > > using "-j4" for UP and "-j8" for SMP is fair, as these produce > the best results for each case. > > the time for UP at -j4: 213.17 > the time for SMP at -j8: 119.08 Well yes, these are the values I was referring, too. > 213.17 / 119.08 == 1.789696 > > not quite 2-1, but not shabby. Well, am I wrong ? I saw it the other way around. Single processor mode gave us the value 213.17. This is the 100% value. SMP gives us a better values. How good ? 119.08 / 213.17 = 0.55 If it would be exactly 0.5, then it would be half the time, so a 100% speedup, so factor 2. 0.55 is exactly 90% speed gain, factor 1.9. > Chuck, you might want to re-try your test compiles, we have > gained some speed with recent changes. I can give you my script and kernel config file, so that you are nearly doing the same ... -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html --/9DWx/yDrRhgMJTb Content-Type: text/plain; charset=us-ascii Content-Disposition: attachment; filename=kbench #! /bin/sh prefix=/bisdn/compile cd ${prefix}/BISDNSMP for i in 1 2 4 8 16 do rm -f *.o sync time make -j $i all >> ${prefix}/kbench.out 2>> ${prefix}/kbench.err sync done --/9DWx/yDrRhgMJTb Content-Type: text/plain; charset=us-ascii Content-Disposition: attachment; filename=BISDNSMP # # BISDN kernel # machine "i386" cpu "I686_CPU" ident BISDNSMP maxusers 64 # SMP Stuff options SMP # Symmetric MultiProcessor Kernel options APIC_IO # Symmetric (APIC) I/O options SMP_INVLTLB # TLB shootdowns (ONLY with APIC_IO) # Debugging #options DDB #options KTRACE #kernel tracing options SHOW_BUSYBUFS # busy buffers on shutdown ? # Networking options INET #InterNETworking options IPFIREWALL #firewall options IPFIREWALL_VERBOSE #print information about dropped packets options "IPFIREWALL_VERBOSE_LIMIT=100" #limit verbosity # filesystems options FFS #Berkeley Fast Filesystem options NFS #Network File System options MFS #Memory File System options PROCFS #Process filesystem options MSDOSFS #MS DOS File System options "CD9660" #ISO 9660 filesystem options NSWAPDEV=3 #Allow this many swap-devices. # misc options options "COMPAT_43" #Compatible with BSD 4.3 [KEEP THIS!] options UCONSOLE #Allow users to grab the console options SYSVSHM,SYSVSEM,SYSVMSG #shared memory (X11) options "MD5" options COMPAT_LINUX # Linux Binary compatibility config kernel root on sd1 # ISA and PCI BUS support controller isa0 controller pci0 # Floppy Disk Controller controller fdc0 at isa? port "IO_FD1" bio irq 6 drq 2 vector fdintr disk fd0 at fdc0 drive 0 # AHA 2940 PCI Controller controller ahc0 # SCSI Devices controller scbus0 device sd0 # Harddisk 0 - DOS/FreeBSD SMP device sd1 # Harddisk 1 - FreeBSD Boot device sd2 # Harddisk 2 - FreeBSD local device st0 # TDC 4222 device cd0 # TOSHIBA XM-5701TA 3136 options AHC_TAGENABLE # tagged command queueing options AHC_ALLOW_MEMIO options AHC_SCBPAGING_ENABLE options SCSI_REPORT_GEOMETRY # SCO compatible system console device sc0 at isa? port "IO_KBD" tty irq 1 vector scintr options MAXCONS=4 # number of virtual consoles # floating point unit device npx0 at isa? port "IO_NPX" flags 0x1 irq 13 vector npxintr # serial devices on mainboard device sio0 at isa? port "IO_COM1" tty irq 4 vector siointr device sio1 at isa? port "IO_COM2" tty irq 3 vector siointr # parallel device on mainboard device lpt0 at isa? port? tty irq 7 vector lptintr # PS/2 mouse on mainboard device psm0 at isa? port "IO_KBD" conflicts tty irq 12 vector psmintr options "PSM_ACCEL=1" # PS/2 mouse acceleration # Joystick device joy0 at isa? port "IO_GAME" # Network 3COM PCI # device vx0 device ed0 at isa? port 0x280 net irq 10 iomem 0xd0000 vector edintr # Soundblaster 16 # SoundBlaster DSP driver - for SB, SB Pro, SB16, PAS(emulating SB) # SoundBlaster 16 DSP driver - for SB16 - requires sb0 device # SoundBlaster 16 MIDI - for SB16 - requires sb0 device # Yamaha OPL-2/OPL-3 FM - for SB, SB Pro, SB16, PAS controller snd0 device sb0 at isa? port 0x220 irq 5 conflicts drq 1 vector sbintr device sbxvi0 at isa? port? irq? drq 5 conflicts device sbmidi0 at isa? port 0x330 irq? conflicts device opl0 at isa? port 0x388 conflicts # Pseudo devices pseudo-device loop pseudo-device ether pseudo-device log #Kernel syslog interface (/dev/klog) pseudo-device vn 1 #Vnode driver (turns a file into a dev.) pseudo-device tun 1 #user mode ppp pseudo-device bpfilter 2 #Berkeley packet filter pseudo-device pty 16 pseudo-device gzip # Exec gzipped a.out's # BISDN options IPI_VJ # Van Jacobsen header compression support #options "IPI_DIPA=3" # send ip accounting packets every 3 seconds options TELES_HAS_MEMCPYB # bisdn 0.97 # Teles S0/16.3 ###################################################### IRQ 9 ## controller tel0 at isa? port 0xd80 net irq 9 vector telintr pseudo-device disdn pseudo-device isdn pseudo-device ipi 1 pseudo-device ispy 1 #pseudo-device itel 1 --/9DWx/yDrRhgMJTb-- From owner-freebsd-smp Fri May 2 01:04:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id BAA19476 for smp-outgoing; Fri, 2 May 1997 01:04:25 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id BAA19471 for ; Fri, 2 May 1997 01:04:22 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id CAA10747; Fri, 2 May 1997 02:04:06 -0600 (MDT) Message-Id: <199705020804.CAA10747@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Andreas Klemm cc: Chuck Robey , Walter Haslbeck , smp@FreeBSD.ORG, everybodyunix@wup.de, jk@ct.heise.de Subject: Re: Dual PPro Mainboard f. SCO SMP In-reply-to: Your message of "Fri, 02 May 1997 09:33:26 +0200." <19970502093326.00399@klemm.gtn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 02 May 1997 02:04:05 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > being a bi-noid I always have trouble with statistical type math, > ... > Well, am I wrong ? I saw it the other way around. Single processor > mode gave us the value 213.17. This is the 100% value. > > SMP gives us a better values. How good ? > > 119.08 / 213.17 = 0.55 > > If it would be exactly 0.5, then it would be half the time, so > a 100% speedup, so factor 2. 0.55 is exactly 90% speed gain, > factor 1.9. as I said, this type math ain't my thing... whether its 1.79 or 1.9 I'm pretty happy its as good as it currently is. Knowing how much room we have for improvement from these base numbers is very encouraging. We are quite inefficient at kernel locking, once we get our act together there it should really fly! And the current version seems to be VERY stable so I'm doubly happy!! -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Fri May 2 03:21:02 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id DAA23969 for smp-outgoing; Fri, 2 May 1997 03:21:02 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [194.77.0.15]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id DAA23964 for ; Fri, 2 May 1997 03:21:00 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id MAA24034 for smp@FreeBSD.ORG; Fri, 2 May 1997 12:15:22 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id LAA07001; Fri, 2 May 1997 11:49:56 +0200 (CEST) Message-ID: <19970502114956.56936@klemm.gtn.com> Date: Fri, 2 May 1997 11:49:56 +0200 From: Andreas Klemm To: smp@FreeBSD.ORG Subject: Re: Add LIVE Girls to your website! References: <199705012342.QAA25878@hub.freebsd.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 In-Reply-To: <199705012342.QAA25878@hub.freebsd.org>; from rich@vegasone.com on Thu, May 01, 1997 at 04:15:01PM -0800 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Thu, May 01, 1997 at 04:15:01PM -0800, rich@vegasone.com wrote: > Rich Olson > Virtual Net Productions > OFFICE: 1-800-314-4285 or 1-702-360-4939 > E-MAIL: rich@vegasone.com Is somebody near him already _toasting_ him for this ? Does somebody have a good connection to his ISP ? Or should I start digging around with whois, to get his ISP ? Just curious ... -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Fri May 2 03:39:09 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id DAA24473 for smp-outgoing; Fri, 2 May 1997 03:39:09 -0700 (PDT) Received: from news1.gtn.com (news1.gtn.com [194.77.0.15]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id DAA24468 for ; Fri, 2 May 1997 03:39:02 -0700 (PDT) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) with UUCP id MAA25799; Fri, 2 May 1997 12:30:40 +0200 (MET DST) Received: (from andreas@localhost) by klemm.gtn.com (8.8.5/8.8.2) id MAA08732; Fri, 2 May 1997 12:16:40 +0200 (CEST) Message-ID: <19970502121639.37715@klemm.gtn.com> Date: Fri, 2 May 1997 12:16:39 +0200 From: Andreas Klemm To: Steve Passe Cc: Chuck Robey , Walter Haslbeck , smp@FreeBSD.ORG, everybodyunix@wup.de, jk@ct.heise.de Subject: Re: Dual PPro Mainboard f. SCO SMP References: <19970502093326.00399@klemm.gtn.com> <199705020804.CAA10747@Ilsa.StevesCafe.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.69 In-Reply-To: <199705020804.CAA10747@Ilsa.StevesCafe.com>; from Steve Passe on Fri, May 02, 1997 at 02:04:05AM -0600 X-Disclaimer: A free society is one where it is safe to be unpopular X-Operating-System: FreeBSD 3.0-CURRENT Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk On Fri, May 02, 1997 at 02:04:05AM -0600, Steve Passe wrote: > > > being a bi-noid I always have trouble with statistical type math, > > ... > > Well, am I wrong ? I saw it the other way around. Single processor > > mode gave us the value 213.17. This is the 100% value. > > > > SMP gives us a better values. How good ? > > > > 119.08 / 213.17 = 0.55 > > > > If it would be exactly 0.5, then it would be half the time, so > > a 100% speedup, so factor 2. 0.55 is exactly 90% speed gain, > > factor 1.9. > > as I said, this type math ain't my thing... whether its 1.79 or 1.9 > I'm pretty happy its as good as it currently is. Knowing how much room we > have for improvement from these base numbers is very encouraging. > We are quite inefficient at kernel locking, once we get our act together > there it should really fly! And the current version seems to be VERY stable > so I'm doubly happy!! Yes indeed, fast and stable !!! BTW: I'd like to create some more figures for further comparisons - dual PPro System using Pentiums with 512k cache - quad PPro System - dual Pentium 200 - dual Pentium 200 MMX It would be interesting to see - how much performance gain you get using 512k 2nd level cache CPU's vs. 256k cache - how much performance (loss) you get, when 2 CPU's have to share the same 2nd Level cache (Pentium/MMX case) - how much system overhead you get using a 4 CPU system - what a MMX CPU brings for such things like compilation jobs But ... we'll certainly never get really compareable results... -- Andreas Klemm | klemm.gtn.com - powered by Symmetric MultiProcessor FreeBSD http://www.freebsd.org/~fsmp/SMP/SMP.html http://www.freebsd.org/~fsmp/SMP/benches.html From owner-freebsd-smp Fri May 2 07:15:46 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA02256 for smp-outgoing; Fri, 2 May 1997 07:15:46 -0700 (PDT) Received: from aviion.ts.kiev.ua (aviion.ts.kiev.ua [193.124.229.12]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id HAA02251 for ; Fri, 2 May 1997 07:15:42 -0700 (PDT) Received: from nbki.ipri.kiev.ua by aviion.ts.kiev.ua with ESMTP id PAA24900; (8.6.11/zah/2.1) Fri, 2 May 1997 15:01:53 GMT Received: from cki.ipri.kiev.ua by nbki.ipri.kiev.ua with ESMTP id QAA10011; (8.6.9/zah/1.1) Fri, 2 May 1997 16:00:05 +0100 Received: from 194.44.146.14 (mac.ipri.kiev.ua [194.44.146.14]) by cki.ipri.kiev.ua (8.7.6/8.7.3) with SMTP id QAA00331; Fri, 2 May 1997 16:11:01 +0300 (EET DST) Message-ID: <3369DA00.5A6B@cki.ipri.kiev.ua> Date: Fri, 02 May 1997 15:11:27 +0300 From: Ruslan Shevchenko Reply-To: rssh@cki.ipri.kiev.ua Organization: IPRI X-Mailer: Mozilla 3.01Gold (Macintosh; I; 68K) MIME-Version: 1.0 To: Chris Csanady CC: Ben Black , Chuck Robey , FreeBSD-SMP@FreeBSD.org Subject: Re: SMP References: <199704280416.XAA12986@nyx.pr.mcs.net> Content-Type: text/plain; charset=koi8-r Content-Transfer-Encoding: 7bit Sender: owner-smp@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk Chris Csanady wrote: > > >i sit corrected. i expect they will eventually migrate to a fully > >threaded kernel. > > As will we I hope. I was hoping to work on pushing the locks down > into the syscalls earlier, but I ran into some trouble. I really > knew very little about assembly, and our locks really are not up > to it yet. :( > > Besides, the general concencus was that we didn't want to deal with > it now.. > > --Chris Csanady > > >On Sun, 27 Apr 1997, Chris Csanady wrote: > > > >> > >> >freebsd-smp is not the best example of how to do SMP. it uses the > >> >simplest method: one giant kernel lock. i don't know that it is > >> >particularly representative of advanced SMP operating systems (though > >> >linux also uses a giant kernel lock). > >> > >> Actually, linux has moved to a slightly finer grain system. Now they > >> have seperate locks for the run queues, scheduler, and some other > >> things.. > >> > >> --Chris Csanady > >> In principle, the *correct* way: split kernel on subsystems, write IDL interfases (of course, on paper, not in srsc) , and analize. From owner-freebsd-smp Fri May 2 07:25:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA02828 for smp-outgoing; Fri, 2 May 1997 07:25:08 -0700 (PDT) Received: (from jmb@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA02783; Fri, 2 May 1997 07:24:30 -0700 (PDT) From: "Jonathan M. Bresler" Message-Id: <199705021424.HAA02783@hub.freebsd.org> Subject: Re: Dual PPro Mainboard f. SCO SMP To: andreas@klemm.gtn.com (Andreas Klemm) Date: Fri, 2 May 1997 07:24:30 -0700 (PDT) Cc: smp@csn.net, chuckr@mat.net, WH@ODS.de, smp@FreeBSD.ORG, everybodyunix@wup.de, jk@ct.heise.de In-Reply-To: <19970502121639.37715@klemm.gtn.com> from "Andreas Klemm" at May 2, 97 12:16:39 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Andreas Klemm wrote: > > - how much performance (loss) you get, when 2 CPU's have > to share the same 2nd Level cache (Pentium/MMX case) this one is of particular interest to me, anyone set up to test this? given numbers for a Pentium dual proc system and a PetniumPro dual proc i can normalize the output ;) jmb From owner-freebsd-smp Fri May 2 07:45:17 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA03902 for smp-outgoing; Fri, 2 May 1997 07:45:17 -0700 (PDT) Received: from uhf.wdc.net (uhf.wdc.net [198.147.74.44]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA03892 for ; Fri, 2 May 1997 07:45:12 -0700 (PDT) Received: from localhost (bad@localhost) by uhf.wdc.net (8.8.5/8.6.12) with SMTP id KAA02748 for ; Fri, 2 May 1997 10:45:29 -0400 (EDT) Date: Fri, 2 May 1997 10:45:28 -0400 (EDT) From: Bernie Doehner To: smp@freebsd.org Subject: Re: Add LIVE Girls to your website! In-Reply-To: <19970502114956.56936@klemm.gtn.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi Gang: > > Rich Olson > > Virtual Net Productions > > OFFICE: 1-800-314-4285 or 1-702-360-4939 > > E-MAIL: rich@vegasone.com I tried calling this number last night to SPAM his voicemail in retaliation, but the 800 number is unreachable. Please note: This heathen is also the administrative contact for the vegasome.com, so politely complaining to him does no good. > Is somebody near him already _toasting_ him for this ? I encourage all of you to send your death threats to this heathen and send copies to the points of contact for netxs.net (vegasone's ISP), in particular rduran@netxs.net, dave@netxs.net. > Does somebody have a good connection to his ISP ? Even if you don't. I encourage to all of you to send your death threats to as many people related to him. > Or should I start digging around with whois, to get his ISP ? Try traceroute vegasone.com and you'll see that netxs.com is their immediate upstream and agis is the next one up. But I wouldn't pester them. Working in the ISP business we have had to deal with this junk time and time again and most of us install either procmail or sendmail SPAM site filters. I already wrote a letter to postmaster@freebsd.org to see if that might be possible here too. If some of you agree with me, please also write him in support of SPAM filtering. Best Regards, Bernie From owner-freebsd-smp Fri May 2 07:51:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA04222 for smp-outgoing; Fri, 2 May 1997 07:51:25 -0700 (PDT) Received: from uhf.wdc.net (uhf.wdc.net [198.147.74.44]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA04217 for ; Fri, 2 May 1997 07:51:20 -0700 (PDT) Received: from localhost (bad@localhost) by uhf.wdc.net (8.8.5/8.6.12) with SMTP id KAA02767 for ; Fri, 2 May 1997 10:52:25 -0400 (EDT) Date: Fri, 2 May 1997 10:52:24 -0400 (EDT) From: Bernie Doehner To: smp@freebsd.org Subject: Re: Add LIVE Girls to your website! (fwd) 2 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk FYI: So that you don't swamp Jonathan's mailbox. Bernie ---------- Forwarded message ---------- Date: Fri, 2 May 1997 06:58:47 -0700 (PDT) From: "Jonathan M. Bresler" To: Bernie Doehner Subject: Re: Add LIVE Girls to your website! (fwd) Bernie Doehner wrote: > > Dear Postmaster: > > Would you please consider adding a SPAM filter to FreeBSD.org's sendmail? > Now doubt you have received counteless reports of this poor excuse for a > human being spaming the smp mailing list. I hope you'll seriously consider > adding SPAM filtering rules to your sendmail.cf. Bernie, spam is a pain in the ass for all of us. i have examined the various methods of trying to stop spam thru sendmail.cf, but none of them are particularly successful it seems to me. we refuse mail that has originated from any of a large number of sites...unfortunately the list keeps growing. we have just added ai.com (functioned as a relay) and vegasone.com to the list. if you know of a particular method of stopping spam, please let's talk about it. we may start accepting mail from subscribers only, this will help some, but the obvious work around will be used by others. jmb -- Jonathan M. Bresler FreeBSD Core Team, Postmaster jmb@FreeBSD.ORG FreeBSD--4.4BSD Unix for PC clones, source included. http://www.freebsd.org/ PGP 2.6.2 Fingerprint: 31 57 41 56 06 C1 40 13 C5 1C E3 E5 DC 62 0E FB From owner-freebsd-smp Fri May 2 09:06:55 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id JAA07996 for smp-outgoing; Fri, 2 May 1997 09:06:55 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id JAA07979; Fri, 2 May 1997 09:06:16 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id KAA14190; Fri, 2 May 1997 10:05:33 -0600 (MDT) Message-Id: <199705021605.KAA14190@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: "Jonathan M. Bresler" cc: andreas@klemm.gtn.com (Andreas Klemm), chuckr@mat.net, WH@ODS.de, smp@FreeBSD.ORG, everybodyunix@wup.de, jk@ct.heise.de Subject: Re: Dual PPro Mainboard f. SCO SMP In-reply-to: Your message of "Fri, 02 May 1997 07:24:30 PDT." <199705021424.HAA02783@hub.freebsd.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 02 May 1997 10:05:33 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > > - how much performance (loss) you get, when 2 CPU's have > > to share the same 2nd Level cache (Pentium/MMX case) > > this one is of particular interest to me, anyone set up to test this? > given numbers for a Pentium dual proc system and a PetniumPro dual proc > i can normalize the output ;) both with COPTFLAGS = -O dual P5-133mHz with shared 512kb external cache. wide 7200rpm SCSI disk. # make clean && make depend && time make -j8 318.39s real 458.19s user 126.32s system dual P6-200mHz each with 512kb internal cache (P6-166 overclocked to 200mHz): older narrow 5400 SCSI disk. /usr/src mounted async,noatime. # make clean && make depend && time make -j8 104.08s real 129.70s user 53.88s system --- on a similar note, I played with the share/mk/* files trying to get "-j8" into them for a world build. I kept hitting places where the Makefile/*.mk files didn't define dependancies closely enough to avoid build errors (links made b4 files copied, etc.). But by timing the build of lower level sub-dirs I saw 25%+ speedup where it worked. This was just by adding "-j8" to bsd.subdir.mk. If someone who had some experience with these guys were to work a little at it I think it could be made to fly. -j4 speeds up even UP kernel builds, so it would be useful for both UP and SMP worlds. It would be nice if it could be done so that it was an /etc/make.conf setting. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Fri May 2 13:05:06 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA19931 for smp-outgoing; Fri, 2 May 1997 13:05:06 -0700 (PDT) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA19925 for ; Fri, 2 May 1997 13:05:02 -0700 (PDT) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.5/8.8.5) with ESMTP id OAA14997 for ; Fri, 2 May 1997 14:05:00 -0600 (MDT) Message-Id: <199705022005.OAA14997@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: smp@FreeBSD.ORG Subject: APIC whitepaper Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Fri, 02 May 1997 14:04:59 -0600 Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, I have created the first draft of a white paper on the APIC sub-system: http://www.freebsd.org/~fsmp/SMP/papers/apicsubsystem.txt please review and return comments. I am more interested in issues of content than style at this point. In particular I would like to know what people feel is missing and/or unclear in the current version. -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD From owner-freebsd-smp Fri May 2 23:14:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id XAA15009 for smp-outgoing; Fri, 2 May 1997 23:14:33 -0700 (PDT) Received: from mail0.iij.ad.jp (mail0.iij.ad.jp [202.232.2.113]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id XAA15002 for ; Fri, 2 May 1997 23:14:31 -0700 (PDT) Received: from uucp2.iij.ad.jp (uucp2.iij.ad.jp [202.232.2.202]) by mail0.iij.ad.jp (8.8.5+2.7Wbeta5/3.5Wpl4-MAIL) with SMTP id PAA16733; Sat, 3 May 1997 15:14:28 +0900 (JST) Received: (from uucp@localhost) by uucp2.iij.ad.jp (8.6.12+2.4W/3.3W9-UUCP) with UUCP id PAA13945; Sat, 3 May 1997 15:14:28 +0900 Received: from tyd1.tydfam.iijnet.or.jp (tyd1.tydfam.iijnet.or.jp [192.168.1.2]) by tydfam.iijnet.or.jp (8.8.5/3.4W2-uucp) with ESMTP id MAA04792; Sat, 3 May 1997 12:00:27 +0900 (JST) Received: from localhost.tydfam.iijnet.or.jp (localhost.tydfam.iijnet.or.jp [127.0.0.1]) by tyd1.tydfam.iijnet.or.jp (8.8.5/3.4Wnomx) with SMTP id MAA03736; Sat, 3 May 1997 12:00:27 +0900 (JST) Message-Id: <199705030300.MAA03736@tyd1.tydfam.iijnet.or.jp> X-Authentication-Warning: tyd1.tydfam.iijnet.or.jp: localhost.tydfam.iijnet.or.jp [127.0.0.1] didn't use HELO protocol To: smp@csn.net Cc: smp@FreeBSD.ORG Subject: Re: Dual PPro Mainboard f. SCO SMP Reply-To: ken@tydfam.iijnet.or.jp In-Reply-To: Your message of "Fri, 02 May 1997 02:04:05 -0600" References: <199705020804.CAA10747@Ilsa.StevesCafe.com> X-Mailer: Mew version 1.55 on Emacs 19.34.2, Mule 2.3 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Date: Sat, 03 May 1997 12:00:26 +0900 From: Takeshi Yamada Sender: owner-smp@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk I am very interesting in how fast the 512K cache P6 is comparing with 256K cache one. It will help me to mkae a good decision for future investment. Can anyone kindly practice such comparison work, and upload the result?