From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:22 2008 Subject: No subject Message-ID: ... If no command is specified, you will be logged in on the remote host us- ing rlogin(1). ... i.e., if you don't give a command the behaviour is governed by the in.rlogind line in inetd.conf, not the in.rshd line. Whether you need a password or not also depends on the settings in /etc/pam.d/rlogin. A further comment: if you do allow rsh without password, you should have a line in.rshd : ALL in your /etc/hosts.deny file (actually I believe that file should contain the single line "ALL : ALL") and then specify the hosts that you want to connect to via rsh in /etc/hosts.allow. Martin ======================================================================== Martin Siegert Academic Computing Services phone: (604) 291-4691 Simon Fraser University fax: (604) 291-4242 Burnaby, British Columbia email: siegert@sfu.ca Canada V5A 1S6 ======================================================================== From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:23 2008 Subject: No subject Message-ID: --- Is KLAT2 really a "Beowulf?" Yes. Every hardware component is an unmodified commodity subsystem, available from multiple vendors. Although KLAT2 runs MPI applications unmodified, some of our system software is odd (but also free). For information on other "Beowulfs" see http://www.beowulf.org/, http://www.beowulf-underground.org/, and the Parallel Processing HOWTO. For an answer to "What's a Beowulf?" see the first entry in the Beowulf FAQ. --- > > Has anyone heard about KLAT2, it is a cheaper way for > supercomputing than > beowulf. It is hard to believe, does someone has any > information about it? > > -------------------------------------------------------------- > ------------ > Salim Mounir Alaoui salim@ee.fit.edu > Computer Science Dept. > salaoui@cs.fit.edu > Research Assistant. salim@ieee.org > Florida Institute of Technology > Melbourne, Florida > Voice: (407) 537-8025. > -------------------------------------------------------------- From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:25 2008 Subject: No subject Message-ID: One part of the cluster would act as a dictionary, one part would act as a grammar/comprehension manager, for the original as well as the translated text, one part would act as a "contextuator", a process remaining largely TBD one part would act as supervisor (having AI capabilities & learning procedures etc.) We have now defined a 4 part set of algorithms, all of which, I'm sure, can benefit from process migration. Recently, a friend of mine tested a translation software (a very good one by today's standards) between English & French. The sentence "Germanium is a good semi conductor" came out as "L'allemand est un mauvais chef d'orchestre"... as I said, its all in the context. Cheers, Al "jok707s@mail.smsu.edu" a écrit : > Let me throw some more fuel on the fire for the translation discussion: > > Suppose one had a large, complex web site that was also constantly changing. > Furthermore, suppose one wanted this entire site to be always available online > in a large number of different languages. Could the ongoing process of site > translation be parallelized by target language &/or page? My offhand guess > would be that splitting up the process for different languages could be > embarassingly parallel, but doing it by page might be more complex due to > links. > > Besides the translation process per se, what issues would be involved in the > relationship between the translating hardware/software and the web server that > would be keeping the whole thing online in a state as close to real-time as > possible? Is anyone using beowulf(s) for something like this already? > > Perhaps a country like India is already wrestling with something like this. > If I recall correctly, India has somewhere around fourteen official languages, > several hundred unofficial ones, and more dialects than anyone has ever tried > to count (somebody with the facts can correct my numbers)--and this is even > before you cross the border and try to reach the rest of the world. If > Internet access reaches the rural corners of India faster than the people out > there learn English, intense parallel translation could be very handy. > > Any interested parties who have not yet seen www.tranexp.com might want to > check out the InterTran software; this could give you a better idea of both > the possibilities and the problems involved here. > > Joel > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:26 2008 Subject: No subject Message-ID: "Source Code: A complete linux 2.4.0-test1 tree including the XFS filesystem is available for cvs checkout. The code is pre-Beta quality and although the basic functionality is operational, it may hang or crash your machine." I've used XFS only under IRIX. Has anybody tried it under Linux? Is it usable already? Martin Siegert Academic Computing Services phone: (604) 291-4691 Simon Fraser University fax: (604) 291-4242 Burnaby, British Columbia email: siegert@sfu.ca Canada V5A 1S6 From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:26 2008 Subject: No subject Message-ID: of the problems that will run a really long time one finds that there are obvious problem segmentations (exploring a parameter space in parallel, accumulating statistics in stochastic simulations in parallel, parallel processing independent images or datasets or initial conditions) that render them EP. One hardly ever writes a program to just run one (very long) time with just one set of input values. EP programs generally scale (in the parallel speedup sense) nearly perfectly on any hardware they'll fit on in the first place... rgb Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu From kus at free.net Sat May 17 01:00:01 2008 From: kus at free.net (Mikhail Kuzminsky) Date: Sat May 17 01:00:28 2008 Subject: Asante 1000Base-T NIC ? Message-ID: <200008031901.XAA15036@nocserv.free.net> Dear colleagues, does somebody know about Asante GigaNIX 1000 BaseT NIC ? The company write (www.asante.com) about Linux driver w/o any date about linux kernel version ... BTW: 1 )the only small GigE-over-copper switch I found (excluding Asante FriendlyNet 400) w/1000 BaseT was Intel NetStructure 470T (any other proposals are welcome !). I'll be very appreciate in any ideas/experience about interoperability of Avante NIC w/Intel switch ! 2) I've heard that Intel PRO/1000 will come soon w/1000 Base T interface. Does somebody know more ? Yours Mikhail Kuzminsky Zelinsky Institute of Organic Chemistry Moscow From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:29 2008 Subject: No subject Message-ID: "The experimental kernel from VA Linux, kernel 2.2.16-8.ext3.3, are at: http://ftp.valinux.com/pub/support/hjl/redhat/6.2/ It is based on 2.2.16-8 from RedHat rawhide with NFS V3, ext3 0.0.2f, new pcmcia-cs, ....., many other changes. There is no support whatsoever. If you have problems, which cannot be reproduced under stock 2.2.16 nor 2.2.16-8 from RedHat, post them to the NFS mailing" Jeff Davis http://www.jrdavis.net/ jdavis@hess.com If you pick up a starving dog and make him prosperous, he will not bite you. This is the principal difference between a dog and a man. -- Mark Twain ----- Original Message ----- From: "Jakob Østergaard" To: "Josip Loncaric" Cc: "Beowulf mailing list" Sent: Friday, August 18, 2000 3:32 PM Subject: Re: NFS 100:1 performance loss On Fri, 18 Aug 2000, Josip Loncaric wrote: > Has anyone seen this NFS performance problem? > > Our Beowulf has two servers w/RAID-0 arrays which deliver 45MB/s, > connected via Gigabit Ethernet. The machines are dual PIII/500 systems > with 512MB of RAM each. We are running Red Hat 6.2 updated to kernel > 2.2.16-3 and recently we updated to nfs-utils-0.1.9.1-1. We are running > 16 kernel nfsd threads on each machine. Here is how long it takes to > copy a 28,955,860 byte file from machine 1 to machine 2: > > rcp: 1.04 seconds (27.8 MB/s, where 1MB=10^6B) > ftp: 1.12 seconds (25.8 MB/s) > NFS 1KB: 12.35 seconds ( 2.3 MB/s, using rsize=wsize=1024) > NFS 8KB: 129.42 seconds ( 0.2 MB/s, using rsize=wsize=8192) > > Clearly, there is something very wrong with NFS, particularly with 8KB > rsize/wsize (which should have improved performance!). Our system > manager tells me that reduced Linux NFS performance with 8KB rsize/wsize > is a known problem, but even at 1KB our NFS is getting less than 10% of > the rcp or ftp performance. > > Any ideas? > Josip You do use the kernel nfs server right ? knfsd, not the old userspace nfs daemon... Our NFS server gets us some 5 MB/s over 100MBit, which is pretty good I think. Linux NFS is not ``excellent'' yet, but with the kernel NFS server it should be better than ``horrible'' at least :) Eventually, you could try the NFS patches (for both client and server) available somewhere at sourceforge (don't have the URL, but a quick search should get you there). They should resolve most of the annying errors you see with Linux NFS, and they should improve performance too I think. We don't use them yet however, so I don't have any experience to go with that one. -- ................................................................ : jakob@ostenfeld.dtu.dk : And I see the elder races, : :.........................: putrid forms of man : : Jakob Østergaard : See him rise and claim the earth, : : OZ9ABN : his downfall is at hand. : :.........................:............{Konkhra}...............: _______________________________________________ Beowulf mailing list Beowulf@beowulf.org http://www.beowulf.org/mailman/listinfo/beowulf From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:30 2008 Subject: No subject Message-ID: Seoul, Korea ----- Original Message ----- From: Horatio B. Bogbindero To: Sent: Wednesday, September 06, 2000 3:15 PM Subject: problems running SCALAPACK > > i have successfully made LAPACK work. thanks to help from the people on > this list. now. i have also successfully compiled SCALAPACK with no > problems. however, when i ran the xdlutime program for benchmarking there > are no results? > > here is the output: > -------------- > [wyy,/home/apps/libs/SCALAPACK/TESTING,02:15]$ mpirun N xdlutime > > Simple Timer for ScaLAPACK routine PDGESV > Number of processors used: 8 > > TIME N NB P Q LU Time Sol Time MFLOP/S Residual CHECK > ---- ----- --- --- --- --------- --------- -------- -------- ------- > > Comments for the output: > Column 1: the time is the wall clock time > Column 2 & 3: matrix size (N by N) and block size (NB by NB) > Column 4 & 5: grid size (P by Q) > Column 6: total time (in seconds) to run PDGESV > Column 7: mega flops per seconds > Column 8: if PDGESV works fine > > -------------- > > what could be wrong? pardon my ignorance. > > --------------------- > william.s.yu@ieee.org > > While we are sleeping, two-thirds of the world is plotting to do us in. > -- Dean Rusk > > > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf From newton at netific.com.netific.com Sat May 17 01:00:01 2008 From: newton at netific.com.netific.com (Wing Newton) Date: Sat May 17 01:00:31 2008 Subject: multi-ethernets LAN Message-ID: <200009111502.IAA30083@ws132.netific.com> A non-text attachment was scrubbed... Name: not available Type: text Size: 214 bytes Desc: not available Url : http://www.scyld.com/pipermail/beowulf/attachments/20000911/a9afdb97/attachment.ksh From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:31 2008 Subject: No subject Message-ID: a factor of 2-3 difference in performance in linear algebra calls, and is a lot cheaper than buying 2-3 times as many nodes:-)... I ask because I need ATLAS (and of course the actual LA packages) for all of the above to properly cover our base of installed systems. > > These packages will be included in our Beowulf 2 distribution we now > preparing beta testing. Please report any packaging problems with > these packages. (e.g. link errors caused by missing functions or > header files) The source tarballs for these packages were taken from > netlib.org. > > Rick Niles > Scyld Computing Corporation. > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf > -- Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu From kus at free.net Sat May 17 01:00:01 2008 From: kus at free.net (Mikhail Kuzminsky) Date: Sat May 17 01:00:33 2008 Subject: kickstart RH6.2 installation problems Message-ID: <200009251904.XAA01603@nocserv.free.net> Dear netters, I'm installing RH6.2 on the nodes of new Beowulf cluster. I want to do kickstart RH 6.2 installation on HDDs, having the necessary partitions (but they are w/o ext2fs created). But after start of installation I receive following traceback messages: File /usr/bin/anaconda, line 341 in ? extraModules=extraModules) File /usr/lib/anaconda/todo.py, line 332 in __init__ self.setClass(instClass) File /usr/lib/anaconda/todo.py, line 822, in setClass todo.addmount(dev,mntpoint,fstype,reformat) File /usr/lib/anaconda/todo.py, line 395, in addMount and install exited abnormally. It looks that the problems are w/partitions. I'll be very appreciate in ideas what is reason of errors. ks.cfg file contents is : lang en_US network --bootproto static --ip 192.168.0.10 --netmask 255.255.255.0 --gateway 192.168.0.4 ### Source File Location cdrom keyboard us ### Partitioning Information #zerombr yes zerombr no #clearpart --linux part / --size 141 --onpart hda2 ^^^^^^^^^^ (the result dosn't depend from the presence of size keywords) part swap --size 133 --onpart hda3 part /usr --size 3004 --onpart hda5 install ### Mouse Configuration mouse genericps/2 --emulthree ### Time Zone Configuration timezone --utc US/Eastern ### X Configuration xconfig --vsync 60 ### Root Password Designation rootpw paSSword ### Authorization Configuration auth --useshadow --enablemd5 ### Lilo Configuration lilo --linear --location mbr ### Package Designation %packages @ Base chkfontpath groff-perl ... ### Commands To Be Run Post-Installation %post echo "This is in the chroot" > /tmp/message Thanks for your help ! Mikhail Kuzminsky Zelinsky Institute of Organic Chemistry Moscow From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:34 2008 Subject: No subject Message-ID: Seoul,Korea P.S : I suggsted so called "Benchmark DB" like www.top500.org for the beowulf clusters few months ago. but I don't see many progresses from this beowulf site. Will you tell me who run the "Benchmark DB" now ? and if someone run the "Benchmark DB", Will it be linked in this www.beowulf.org ? or Will the www.beowulf.org (Scyld) make the "Benchmark DB" for the beowulfian in the world ? If Scyld have any intention to make a "Benchmark DB" in this beowulf, Will Scyld allow me make a korean Scyld Branch, and make a "Benchmark DB" there & link it to the beowulf list ? Thank you ----- Original Message ----- From: Kolbuszewski, Marcin To: Sent: Thursday, October 05, 2000 6:01 AM Subject: One-day workshop on building commodity clusters > Cluster-building workshop > > High Performance Computing program at the National Research > Council of Canada organizes a one day workshop devoted to > building computer clusters out of commodity components. > > The workshop is targeted at scientists and engineers who > use computing as an everyday research tools and are > considering clusters as an cost-effective option for their computing needs. > > Representatives of the leading providers of Linux and clustering > solutions as well as Canadian academic and government cluster > users will share their knowledge and experiences. > > The workshop's website is at > > http://www.iit.nrc.ca/HPC/workshop.html > > Marcin Kolbuszewski > > > ---------------------------------------------------------------------------- > -- > High Performance Computing Group Coordination Office > Institute for Information Technology C3.ca Association > National Research Council of Canada > > Rm 286, M-50, 1500 Montreal Road tel 613-998-7749 > Ottawa, Canada fax 613-998-5400 > K1A 0R6 e-mail > Marcin.Kolbuszewski@nrc.ca > > > PROGRAM > > > One-day workshop on building commodity clusters > 17 November 2000, Ottawa, Ontario > National Research Council of Canada > Sussex Laboratories > 100 Sussex Drive > > Organized by the internal High Performance Computing Program at the National > > Research Council of Canada. Contact ihpc@nrc.ca for more information > Schedule: > > 8:45 - 9:00am *Opening addresses > Dr. John Tse, Dr. Roger Impey, co-chairs of the iHPC program, NRC > > 9:00 - 9:45am Overview of Intel and Alpha based clusters > Stephen Fried, Microway Inc. > > 9:45-10:30 Overview of high speed interconnects > Keith Murphy and Tim Wilcox, Dolphin Interconnect AS > > 10:30-10:45 Coffee break > > 10:45-11:30 *Parallel computing on Linux clusters using Fortran, C and C++ > compilers and tools from The Portland Group, Inc. (PGI) > Douglas Miles, Portland Group International > > 11:30-12:15 *Question and answer session with three invited speakers > > 12:15-13:15 Lunch > > 13:15-13:45 *THOR - A Multi-purpose Commodity Component Supercomputer > Prof. James Pinfold, University of Alberta > > 13:45-14:15 *Chemical Supercomputing on the cheap: Cobalt cluster > Dr.Serguei Patchkovskii, University of Calgary > > 14:15-14:45 *Application of Cluster Technology to BioInformatics > Jacek Nowak, Plant Biotechnology Institute, NRC > > 14:45-15:00 Coffee break > > 15;00-15:30 Intel cluster with Wulfkits for CFD at ICPET > Ron Jerome, Institute for Chemical Process and Environmental Technology, NRC > > > 15:30-16:00 *Aerodynamic CFD Discovers Cluster Computing > Dr. Jerry Syms, Institute for Aerodynamic Research, NRC > > 16:00-16:30 *Using PBS > Dr. Gabriel MateescuResarch Computing Support Group, NRC > > 16:30-17:00 *Performance of Linux Clusters Compared to Some Traditional RISC > > Multiprocessor Systems > Dr. Douglas Ritchie, Resarch Computing Support Group, NRC > > 17:00 Closing comments > * indicates confirmed titles > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf From owiest at darwin.helios.nd.edu Sat May 17 01:00:01 2008 From: owiest at darwin.helios.nd.edu (Olaf Wiest) Date: Sat May 17 01:00:34 2008 Subject: 00.09.28 Scientist, LinuxClaster Admin, U Notre Dame, Indiana Message-ID: College of Science Science Computing Facilities University of Notre Dame Linux system administrator/scientist The College of Science at the University of Notre Dame has a position open for a Linux system administrator. See http://www.science.nd.edu/scf/open.html for details. The College (which includes Biological Sciences, Chemistry and Biochemistry, Mathematics, and Physics) has a mixed platform distributed computing environment (primarily Sun, SGI, Linux, Windows NT, and Macintosh). The appointments will be at faculty rank (assistant professional specialist) for those with suitable academic background. Applicants should have two or more years of experience in system administration using Linux. Duties will include installation and support of a production cluster of 50 or more Linux workstations. This system administrator will also be responsible for support of the desktop computers of scientists in the college who are running the Linux operating system. Preference will be given to candidates who have experience with distributed file systems (NFS, AFS), shell scripting (Bourne shell, PERL, etc.), and computer security. Interested applicants should submit a letter of application, resume, and names of three professional references to Prof. Kathie Newman, Associate Dean, College of Science, 229 Nieuwland Science Hall, University of Notre Dame, Notre Dame, IN 46556. Applications may be sent by email to newman.1@nd.edu. Review of applications will begin immediately and will continue until the positions are filled. EOE/AA MINORITIES AND WOMEN ARE ENCOURAGED TO APPLY. From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:34 2008 Subject: No subject Message-ID: difference between LE, HE-SL and HE? And from http://www.interpromicro.com/mb-superm.html, the price of DE6 is $595. Is it correct? Eric ----- Original Message ----- From: Douglas Eadline To: He, Ding Cc: Jared Hodge ; Sent: Wednesday, October 11, 2000 6:32 AM Subject: Re: ServerWorks chipset > On Tue, 10 Oct 2000, He, Ding wrote: > > > Does anybody happen to know which one is more suitable for clustering: > > SuperMicro 370DL3 and 370DE6 ? > > The 370DE6 uses the HE chipset which supports interleaved memory. > This could be a big win for some applications. > > However, the 370DE6 is not shipping yet and looks to > be quite expensive (~$1500). > > Doug > > > > Thanks. > > > > > > ----- Original Message ----- > > From: Jared Hodge > > To: > > Sent: Monday, September 25, 2000 10:06 AM > > Subject: ServerWorks chipset > > > > > > > Here's part of an E-mail that I received from Myricom on the subject: > > > > > > > > > We have tested Supermicro 370DLE 370DL3 and Supermicro P-III DME. > > > They all work fine. The 370DLE and DL3 have better performance > > > since they use the RCC chipset rather than the intel840. > > > > > > We have also tested almost every intel-840 board we could find > > > including Tyan, Supermicro, Iwill. > > > > > > If you are using PCI64 cards from Myricom, then the > > > RCC is really a good chipset to use since Myrinet boards > > > can take full advantage of the excellent PCI bandwidth. > > > > > > > > > We're looking at upgrading our beowulf using the 370DLE, 800EB Mhz > > > processors (one per node to start) and 512MB RAM. If anyone has > > > suggestions or warnings, please let me know. > > > > > > Jared > > > > > > > > > _______________________________________________ > > > Beowulf mailing list > > > Beowulf@beowulf.org > > > http://www.beowulf.org/mailman/listinfo/beowulf > > > > > > > > > _______________________________________________ > > Beowulf mailing list > > Beowulf@beowulf.org > > http://www.beowulf.org/mailman/listinfo/beowulf > > > > -- > ------------------------------------------------------------------- > Paralogic, Inc. | PEAK | Voice:+610.814.2800 > 130 Webster Street | PARALLEL | Fax:+610.814.5844 > Bethlehem, PA 18015 USA | PERFORMANCE | http://www.plogic.com > ------------------------------------------------------------------- > From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:35 2008 Subject: No subject Message-ID: These results were run before we had support for shared memory for local communication. So, on our quad processor nodes, if any of the mpi processes needed to communicate to another on the same board the communication would have to go through the pci bus, myrinet card, to the switch and back. We now have the shared memory support for local communication which cuts the zero length, half round-trip latency by better than a third. Doug On Wed, 18 Oct 2000, Andreas Boklund wrote: > > I replied to Daniel but i might as well translate my reply to him to > english and add a few things and send it to the list. > > I have build 2 clusters with the purpose of running fluent on them. > > The first one was > > FoxHole > 20 * Pentium III 500 128SDRAM > 1 * Cisco Catalyst 1900 > > > The second one is the one that im "fine tuning" right now. > > ????? > 1 * dual Pentium III 866 256RDRAM > 9 * dual Pentium III 800 256RDRAM > 1 * Cisco Catalyst 2900 > > > Our calculations has a LOT of chemistry in them so i dont think its fair > to draw any standard conclusions from them. > But maybe a few pointers: > > They dont require much bandwith. > Although each iteration i do with my new cluster sends approx 150-200MB of > information /iteration. One iteration takes just over a minute to > complete. This might sound like a lot but the switch is not an issue. > > On very few computers it has pretty good scalability but somewhere around > 10-12 processors the gain of adding another one is very small. (we need > the time so we added a few more anyway). > > This chart http://www.exjobb.udd.htu.se/2000/sv3/25/Image3.gif (is in > swedish) but it showes the time it takes for FOXHOLE with different number > of nodes to complete our problem, and it shows the time it takes for a > Dell with 4*550 and 1GBSDRAM. > > Fluent is pretty RAM hungry although for the nodes i "might" have been > able to use 128MB but im considering to go to 512 on the master. > > > > MVH > //Andreas > > > And if anyone know Swedish (i know that Daniel does) you can read our > Bachelor thesis on: > http://www.exjobb.udd.htu.se/2000/sv3/25/ > Unfortunately we had to cut away a lot of the technical stuff :( > > > > ********************************************************* > * Administator of Amy(studentserver) and Sfinx(Iris23) * > * * > * Voice: 070-7294401 * > * ICQ: 12030399 * > * Email: andreas@htu.se, boklund@linux.nu * > * * > * That is how you find me, How do -I- find you ? * > ********************************************************* > > > > > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf > -- Doug Johnson Systems Developer/Engineer The Ohio Supercomputer Center djohnson@osc.edu From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:37 2008 Subject: No subject Message-ID: it appears that the eepro100 loads as part of the second stage boot, but if that is the case, how does the system manage to retrieve the second stage boot from the Master node. Since I don't completely understand how all this works yet, I'm probably reading things wrong, but I would like to understand it! Anyway, any suggestions on how to get the eepro100 working on the slaves would be greatly appreciated. -- Stand Fast, tjg. Timothy Grant tjg@exceptionalminds.com Chief Technology Officer www.exceptionalminds.com Red Hat Certified Engineer (503) 246-3630 Avalon Technology Group, Inc. fax (503) 246-3124 >>>>>>The last power outage was: 5 days 1:19 hours ago<<<<< From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:38 2008 Subject: No subject Message-ID: 1) PVFS 2) OpenAFS 3) Coda 4) none of the above. turn a machine into an NFS server Is anyone willing to comment on the virtues/drawbacks of these solutions in a COW environment? (or to point me to references where this comparison is made?) > > Also, now that openafs has been released > > AFS has some very good features, but it also has some very evil ones. > > Regarding your primary idea, you might look into pvfs. -- Dr. Ivan Rossi - CIRB Biocomputing Unit e-mail: ivan@biocomp.unibo.it Web: http://www.biocomp.unibo.it/ivan From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:39 2008 Subject: No subject Message-ID: as a "scratch space" for parallel computation, and will give you a big increase in performance for those applications (shameless plug :). If you are concerned about redundancy (this is actually a good idea from the management perspective even if you don't care about redundancy), then I would recommend NFS for the actual /home directory. NFS survives PC failure no better than PVFS, but is at least easier to back up because it resides on one disk. It is also easier to make use of RAID for disk level protection if you use NFS because you only have to buy one extra disk to be able to use software raid mirroring. Mirroring every drive in a PVFS cluster could get expensive quickly. You don't necessarily need to purchase a new disk for home unless you need more than 18G for that purpose. It is perfectly acceptable to use the existing disk on your head node as an NFS exported home directory and then use your other 7 node disks as PVFS storage. This is how we typically setup clusters at our site and other people seem to have success with this configuration as well. hope this helps, -Phil Phil Carns Parallel Architecture Research Lab Clemson University From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:41 2008 Subject: No subject Message-ID: ----- Original Message ----- From: Yoon Jae Ho To: Sent: Saturday, December 02, 2000 3:24 PM Subject: Re: beoserv not seeing RARP > After I succeeded in setting up the beowulf2 using preview CD, I can run beoMPI with ease. > > but I found very strange situation. > > After I trun off (power off not booting ) the master node, The master node can't recognize the slave node. > > I mean the master node can't receive the RARP signal from the slave node. > > In the beosetup screen, I found the node "down" all day long. > > My PC is a 150Mhz Pentinum with 16 M RAM. > > So I install the master node from CD again, then It worked perfectly. > > Without turning off master node, I turn off slave node only several times. then the beosetup & beostatus works fine. > > but If I power off the master node & power on the master node, then the master node can't recognize the slave node again. > > the procedure of my power off was > > "halt" from the window's shell or "shotdown -h now" from the shell prompt. > > I think something is problem in the master node program -- maybe beoserv or my pc hardware for small RAM. > > Is there Anybody here same problem like me ? > > I need your help > > Thank you very much. > > ----------------------------------------------------------------------------------------------------------------------------------------- > Yoon Jae Ho > Economist > http://ie.korea.ac.kr/~superocm/ Korea Beowulf > > Imagination is more important than Knowledge A. Einstein. > ----------------------------------------------------------------------------------------------------------------------------------------- > > > ----- Original Message ----- > From: Yoon Jae Ho > To: Ken > Cc: > Sent: Thursday, November 30, 2000 10:21 AM > Subject: Re: beoserv not seeing RARP > > > > Me, same problem they don't even show up in the unknown column. > > > > but, i format the slave node from the advice of Daniel Persson myrridin@wlug.westbo.se > > Tue, 24 Oct 2000 > > > > http://www.beowulf.org/pipermail/beowulf/2000-October/010345.html > > > > I make a partition of slave node from the above instruction. > > > > Then, I succeeded the setting up the beowulf-2 using beosetup (down -> up ) > > > > > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:44 2008 Subject: No subject Message-ID: could drive a Cruise like missile without major difficulties. That was the direct reason for initial thought of export restrictions. But PS2 is built around a RISC processor (I _think_ IBMs Power something) so it is not that much more powerful that your dekstop... Marcin -----Original Message----- From: Dave Leimbach To: Ken Cc: epaulson@students.wisc.edu; beowulf@beowulf.org Sent: 12/22/00 8:10 AM Subject: Re: Serverworks LE versus VIA Chipset Actually there was an article an /. not too long ago about how PS2's have export restrictions due to the massive computing power stored within. I haven't seen the specs for a PS2 so I won't say its true or not. I just remember the old article saying that it was feared they might be used for weapons research which is exactly what /. claims Iraq is doing. Could be true... screweyer stuff has happened.. Dave On Thu, 21 Dec 2000, Ken wrote: > Actually, if you beleive NBC news, we should all be using Play Station > 2's. They actually had someone on saying that the PS2's are "1000 times > more powerfull than any desk top computer". I may have the quote a > little wrong, but that was pretty much what he said. > > I was a little skeptical when Slashdot pointed to a story about Iraq > building PS2 clusters to get around the sanction on computer equipment > but who am I to argue with NBC? ;-) > > -- > Ken Lowther > Youngstown, Ohio > http://www.atmsite.org > > _______________________________________________ > Beowulf mailing list > Beowulf@beowulf.org > http://www.beowulf.org/mailman/listinfo/beowulf > _______________________________________________ Beowulf mailing list Beowulf@beowulf.org http://www.beowulf.org/mailman/listinfo/beowulf From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:44 2008 Subject: No subject Message-ID: certainly be an exercise in extreme pain, culminating in bashing the damn things with a sledgehammer and throwing them into a ditch somewhere (or giving them to a children's hospital, which is a much better thing to do). However, it is entirely possible that Saddam is desperate enough (given the embargo that likely makes it moderately difficult for him to get even cheap Duron boxes) to be willing to throw his available computer hackers at the task. Perhaps he just wants them to train his pilots using flight simulators, since he's a bit short on planes...;-) The possibility of his building an "unmanned aerial vehicle" on top of the PS2 is not quite as crazy (and indeed is almost certainly feasible), but again I fail to see the motivation for using a PS2 instead of, say, an ordinary laptop computer (of which hundreds would fit into a single truck, once again). Hitler managed to build his "buzz bombs" with 1940's technology, so UAV's are clearly not particularly difficult to make. I'll bet that I could engineer one from scratch inside a month or two (right after I learned to fly, of course:-) that would have a 90% chance or better of working. These days, with satellites kindly providing us with electronically readable absolute locations (to within 10 meters or so) I find it excruciatingly difficult to believe that an ordinary laptop with a handful of servos couldn't be glued into a piper cub to build a UAV capable of taking off, flying to some predefined location along some predefined trajectory remaining (say) 15 meters above the ground (to clear ground trash but remain below most radar) and then crash itself into any given 100 meter circle within its fuel range. After all, the timescale of the steering and piloting corrections required to fly any non-jet plane is long, long, long compared to the speed of a laptop, and we're talking solving some simple negative feedback coarse grained differential equations here. As in "drifting to the left, move the stick to the right" while integrating the trajectory with realtime corrections from the GPS. Over seconds. Pretty much the same code used to write flight SIMULATORS, which have been around since the original IBM PC (which was ALSO almost certainly capable of flying a small plane). Maybe flying a jet through rugged mountains would be a bit trickier, but still -- a rack of playstation 2's? Why not dedicated moslem kamakazi's? They are (to be cynical) preprogrammed and probably more readily available, judging from the number of suicide attacks that occur in the middle east. Better at evasive maneuvers and dealing with accidents of the road, too. > could be true... screweyer stuff has happened.. This is certainly correct. And I'm certain that (given enough effort) PS2's COULD be used to build UAV's or even to do nuclear weapons research (we'll see if they are featured in the next major bid to Los Alamos;-). It would be a most peculiar statement of the distorting effects of the embargo if indeed it has driven Saddam to spend much money importing PS2's and expecting his computer folks to deal with all the manifold headaches of using them as general purpose computers (or as de facto programmable device controllers). OTOH, they >>do<< have a lot of onboard "smarts" for doing e.g. "enemy AI" and simple physics calculations, as any game system must (after all, what are they running but moderately sophisticated classical physics simulations?). I wonder if he's running the market on PDA's? or any of the myriad programmable device controllers available to the world today? Or if he's set up an ASIC shop in (say) Taiwan? In a world where devices as evil as singing christmas cards with "infinite" battery capacity exist (hammer time again:-) custom silicon is obviously WAY too cheap for our own good...;-) > > Actually, if you beleive NBC news, we should all be using Play Station > > 2's. They actually had someone on saying that the PS2's are "1000 times > > more powerfull than any desk top computer". I may have the quote a > > little wrong, but that was pretty much what he said. The /. article suggests that it is a 300 MHz, 128 bit CPU, "every bit as powerful as the processors found in most desktop and laptop computers" (no pun intended, I'm sure -- a bit >>more<< powerful would have me doubt my own understanding of binary logic:-). I'd suggest that this is a much fairer statement, although I'm also sure that the CPU is (as advertised) very good at doing graphics transformations. The system is still constrained by the current VLSI scale and silicon available to all systems designers, and still must talk to memory and peripherals over a bottlenecked bus. When a number like 1000x more powerful appears, I think back to the old days, when various "special purpose" CPU's touting incredibly fast (for their day) floating point speeds were knocking around the PC market. In many cases the claim was technically true from (as Obi Wan would say) a certain point of view -- if one loaded the CPU's registers in a particular way and was doing just the right mix of instructions and didn't have to go even to an L2 cache, one could get 10 or 100x the nominal throughput of a general purpose CPU plodding along on operations out of main memory. DSP's, for a while, were often touted for this sort of purpose, as was the (IIRC) i860, and when it comes right down to it, the PS2 is YADSP being offered up as a general purpose computer. The incredible pain of programming a DSP for a general purpose calculation without anything like an operating system or real programming language was conveniently ignored. The i860 faired somewhat better, but never had supercomputer vendors exactly quaking in their boots. The PS2 is in better shape than most of the DSP-based efforts -- it at least has been engineered from the beginning for >>something<< like general purpose use and does have a general purpose CPU integrated in the engine, but still... This long ramble, at the end, has a bit of a beowulf-relevant point. The beowulf concept (that differentiates it from the manifold virtual parallel machine efforts that preceded it) has from the beginning been one of COTS technology and mainstream, proper OS and programming support. It is quite difficult enough to engineer a beowulf with a high quality, open source operating system, over the counter devices (especially network devices) living on a standardized bus, excellent compilers, and venerable (and "mostly" debugged:-) communications libraries. Building "one-of-a-kind" cluster supercomputers out of components being used far from their intended purpose and missing most of the above is the sort of thing undertaken by companies like Thinking Machines -- it takes years and top-level expertise and a major investment in engineering and when you're finally done, you find that Moore's Law ate all of your efforts alive and you're lucky if you can recover development costs on sales before even the dimmest customer begins to realize that maintenance alone is costing more then the hardware that would replace the whole shebang. Or (in the case of the CM5) possibly even the costs of electricity and cooling. [Which also happens even with beowulfs -- ten 200 MHz pentium class nodes cost approximately $600/year to feed and are likely slower, for most purposes, than a single Duron node that costs $600.] I wonder if Iraq executes its computer scientists who fail? I wonder if all those PS2's are really Christmas gifts for Saddam's many relatives? I wonder when we'll see the first UAV terrorist attack? I wonder when people will finally wake up and realize that killing each other to achieve a better way of life is as oxymoronic as it gets? rgb -- Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:49 2008 Subject: No subject Message-ID: is change the /etc/beowulf/fstab file. Mine currently has the following entries: /dev/ram3 / ext2 fs_size=65536 0 0 none /proc proc defaults 0 0 none /dev/pts devpts gid=5,mode=620 0 0 $MASTER:/home /home nfs defaults 0 0 This fails, and I have trouble interpreting the error log attached at the end of this message. Now, who can help me? 1. What are the instructions for booting a diskless node with Scyld? 2. Is it possible to boot a diskless node without a Scyld floppy or CD-ROM? My nodes send out DHCP requests. Can I simply setup a dchp server to hand out /var/beowulf/. Will dhcpd conflict with beoserv over ports? Valentin --------------cut here------------- [root@scyld beowulf]# cat /var/log/beowulf/node.0 node_up: Setting system clock. mke2fs 1.18, 11-Nov-1999 for EXT2 FS 0.5b, 95/08/09 ext2fs_check_if_mountFilesystem label= OS type: Linux Block size=1024 (log=0) Fragment size=1024 (log=0) 128 inodes, 1024 blocks 51 blocks (4.98%) reserved for the super user First data block=1 1 block group 8192 blocks per group, 8192 fragments per group 128 inodes per group Writing inode tables: done Writing superblocks and filesystem accounting information: : No such file or directory while determining whether /dev/ram1 is mounted. done node_up: TODO set interface netmask. node_up: Configuring loopback interface. /dev/hda: No such device beoboot: /lib/modules/2.2.16-21.beo/modules.dep missing /usr/lib/beoboot/bin/node_modprobe: /lib/modules/2.2.16-21.beo/modules.dep: No such file or directory setup_fs: Checking /dev/ram3 (type=ext2)... e2fsck 1.18, 11-Nov-1999 for EXT2 FS 0.5b, 95/08/09 ext2fs_check_if_mount: No such file or directory while determining whether /dev/ram3 is mounted. Couldn't find ext2 superblock, trying backup blocks... The superblock could not be read or does not describe a correct ext2 filesystem. If the device is valid and it really contains an ext2 filesystem (and not swap or ufs or something else), then the superblock is corrupt, and you might try running e2fsck with an alternate superblock: e2fsck -b 8193 e2fsck: Bad magic number in super-block while trying to open /dev/ram3 setup_fs: FSCK failure. setup_fs: Creating ext2 on /dev/ram3... mke2fs 1.18, 11-Nov-1999 for EXT2 FS 0.5b, 95/08/09 ext2fs_check_if_mount: No such file or directory while determining whether /dev/ram3 is mounted. setup_fs: Mounting /dev/ram3 on /rootfs//... (type=ext2; options=defaults) setup_fs: Checking 192.168.2.1:/home (type=nfs)... setup_fs: Mounting 192.168.2.1:/home on /rootfs//home... (type=nfs; options=defaults) beoboot: /lib/modules/2.2.16-21.beo/modules.dep missing beoboot: /lib/modules/2.2.16-21.beo/modules.dep missing /usr/lib/beoboot/bin/node_modprobe: /lib/modules/2.2.16-21.beo/modules.dep: No such file or director\y /usr/lib/beoboot/bin/node_modprobe: /lib/modules/2.2.16-21.beo/modules.dep: No such file or directory node_modprobe: installing kernel module: nfs /tmp/nfs.o: unresolved symbol rpc_register_sysctl_Rbf9a77c0 /tmp/nfs.o: unresolved symbol rpc_wake_up_task_Rffa78ed9 /tmp/nfs.o: unresolved symbol rpc_do_call_R0fae8de2 /tmp/nfs.o: unresolved symbol rpc_proc_unregister_R5bd26000 /tmp/nfs.o: unresolved symbol rpc_allocate_R0cd1c989 /tmp/nfs.o: unresolved symbol rpcauth_lookupcred_R0366fdf8 /tmp/nfs.o: unresolved symbol rpc_clnt_sigunmask_R17abaa09 /tmp/nfs.o: unresolved symbol xdr_encode_string_Rabc0fe0c /tmp/nfs.o: unresolved symbol rpc_init_task_Rf4c99bc4 /tmp/nfs.o: unresolved symbol rpc_sleep_on_R41929c92 /tmp/nfs.o: unresolved symbol rpc_shutdown_client_Rb50bc549 /tmp/nfs.o: unresolved symbol rpc_create_client_R4589e663 /tmp/nfs.o: unresolved symbol rpciod_up_R375492a4 /tmp/nfs.o: unresolved symbol rpc_call_setup_R6f2441da /tmp/nfs.o: unresolved symbol rpc_proc_init_Rf56e5632 /tmp/nfs.o: unresolved symbol rpc_killall_tasks_R66ae6aea /tmp/nfs.o: unresolved symbol rpc_release_task_Re71e954e /tmp/nfs.o: unresolved symbol nlmclnt_proc_Rc02cb40f /tmp/nfs.o: unresolved symbol nfs_debug_Raf5bf6ef /tmp/nfs.o: unresolved symbol rpc_execute_R2f4e83ce /tmp/nfs.o: unresolved symbol rpc_clnt_sigmask_R3b8df6d4 /tmp/nfs.o: unresolved symbol xprt_create_proto_Rc88e4139 /tmp/nfs.o: unresolved symbol rpciod_down_Rbabf0f35 /tmp/nfs.o: unresolved symbol rpc_proc_register_R83e79004 /tmp/nfs.o: unresolved symbol xprt_destroy_Rea15ebb6 /tmp/nfs.o: unresolved symbol rpc_wake_up_next_R134f0e35 mount: fs type nfs not supported by kernel Failed to mount 192.168.2.1:/home on /home. From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:51 2008 Subject: No subject Message-ID: should be using as much of one language as possible. mod_perl would be my choice - you'll have more flexibility with mod_perl because Apache is more accessable overall to developers, IMHO. > > The database question. The database is large, but entirely static. > Would PostgreSQL be able to handle this comparably to Oracle? The > licenses for Oracle are not so very expensive anymore, but if we're > going towards a farm of boxes, this will start to cumulate. Otoh, > some Oracle stuff seems to be able to handle parallel databases > natively. Hmm. I think it might! I'm working with PostgreSQL, and I'm seeing comparable performance, but you'll need to do some benchmarking. If it doesn't compare now, it eventually will. Watch out for using joins in PostgreSQL - they can impact run times. > > Sorry for the bunch of clueless questions, but I'm not exactly a > computer person, nor is the company extremely competent in technical > questions (they're all a bunch of chemists, which just have been > working with computers for a long time). The budget for hardware seems > to be there, but I'm a new guy, and I can't afford buying a bunch > of expensive crap which will just sit there gathering dust. You're not clueless 'cause you're asking the right questions - this is the same type of thing directors and managers must contend with. You have a chance to sell beowulf to your higher-ups. Go for it! > > I would also welcome some pointers towards lists where questions > such as these are handled holistically. I'd settle for a bunch > of dedicated lists too, though. > > > > _______________________________________________ > Beowulf mailing list, Beowulf@beowulf.org > To change your subscription (digest mode or unsubscribe) > visit http://www.beowulf.org/mailman/listinfo/beowulf > ------_=_NextPart_001_01C092C6.98BBCDA0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable RE: phasing out Solaris/Oracle/Netscape with = Linux/PostgreSQL/Apache

Let me know if I can help beyond what I write = below.  Additional comments from the crowd welcome.  See = below . . .

> -----Original Message-----
> From: Eugene.Leitl@lrz.uni-muenchen.de
> [mailto:Eugene.Leitl@lrz= .uni-muenchen.de]
> Sent: Thursday, February 08, 2001 10:43 = PM
> To: linux-elitists@zgp.org; = pigdog-l@bearfountain.com;
> beowulf@beowulf.org
> Subject: phasing out Solaris/Oracle/Netscape = with
> Linux/PostgreSQL/Apache
>
>
>
> This is somewhat off-topic, so bear with = me.
>
> Um, I need some advice. The company I'm with = has a Solaris Sun with
> Netscape Enterprise Server. It is running a mix = of C, Perl, Oracle
> and Daylight. The latter beast currently = requires 1.5 GByte RAM,
> but is still creaking at all seams. We're going = to kick Daylight
> sooner or later, so the memory footprint will = drop, possibly a lot.
>
> I don't yet understand the architecture of the = chemical database
> we're running, so I don't know where the = bottlenecks are, but I need
> to build a Linux machine which could outperform = the Sun at a fraction
> of the price. I will set it up locally, and = hammer it with queries,
> doing measurements, stability tests, = etc.

This could get you.  Critical to understand some = basic things about the existing application.  Exporting data, etc = is your first concern - you need to know how to do that.  That = could very well set the pace of a conversion.

>
> I would rather like to use Athlons, especially = recent DDR
> Dual Athlons,
> but this doesn't seem to go too well with = required stability. This
> means I should use a dual Pentium III = motherboard, right? Does Linux
> handle these well? I heard these can take up to = 8 GByte RAM, I think
> I should start with 2 GBytes. I don't think the = application is CPU
> bound, but I don't know for sure yet. It = certainly seems to exercise
> disks strongly, so here is another = question:

Linux does handle them well, but I'm impartial to = FreeBSD.  Since you're using Sun OS already, you could continue = with Sun X86 version without making the big switch all at once - it's = free and a works well enough.  It would make a fine part of a = beowulf, and allow you to get used to Linux over a longer period of = time.  If you can make the transition in baby steps, it's = better.

>
> do I absolutely, positively need SCSI? I was = thinking about putting a
> second 100 EIDE host adapter in, and run disk = striping plus mirroring
> over 4 EIDE hard drives (the better models from = IBM). Or should I
> use a Dual-Pentium mumboard with onboard SCSI, = and buy several fast,
> hot & noisy scuzzys, soft-RAIDing them? = Perhaps even harware RAID?
> I don't think the disks need to be very large, = but they
> better be fast.
>

SCSI is GREAT, and you should set up redundant hot = swaps so if you crash, you insert a new disk, type "boot", = and you're back online with a node.  I think Sun stations = outperform the Intel boards on disk throughput, but you could = check.

> I would like to use Reiser FS, so this only = allows me RAID 0/1, right?
> Higher stability, lack of fscking delay in case = machine needs to be
> rebooted and no 2 GByte file size limit would = seem to be needed.
> Since this is ReiserFS, I should obviously go = with the latest stable
> kernel. I've been using Mandrake for my desktop = and small
> time testing,
> but for this application this is probably not = the way to go. Which
> distro should I choose? Debian?

Not sure on this one.  Not enough exposure with = all of those filesystems.  I know, use MFS, and load up machines = with 2GB of ram!  Just kidding.

>
> We seem to be moving towards an architecture = consisting of a bunch
> of Perl programs (it's not settled, but the few = routines we have are
> in Perl) communicating via sockets. (Right now = it's a mix of C and
> Perl, talking via named pipes). The queries (a = chemical structure
> drawn within a browser, using a Java applet or = a plug-in) are
> handled with a dispatcher (a cgi-bin Perl = thing). Sooner or
> later we will spread the query to a farm of = boxes, each containing
> a database, or segmenting a database across = several, cheap, redundant
> boxes. (But where not there yet). There's no = alternative to sockets,
> right?


You'll take a big performace hit running perl too = much - it's interpreted.  I know it's popular but with a big = project, you should get the training required to do it in a compiled = language, like C/C++.  If you insist on an interpreted language, = use Java - it's a 'cleaner' language to manage on a large scale.  = For socket work I would definitely stick with C - and create a library = specific to your needs that your perl programs could call.

>
> Right now a query can take up to minutes, so I = don't think mod_perl
> is needed. We don't get a lot of query hits, at = least not yet. Should
> I try using Apache mod_perl instead of the = Netscape Server
> nevertheless?

How much data does the longer queries access?  = If you're seeing less than sub-minute response times, it's a red = flag.  That's just a rule of  thumb I use.  I've seen = five second return times on several gigs of data with a Sun and = Oracle. 

From the standpoint of managing the code, regardless = of what you use, you should be using as much of one language as = possible.  mod_perl would be my choice - you'll have more = flexibility with mod_perl because Apache is more accessable overall to = developers, IMHO.

>
> The database question. The database is large, = but entirely static.
> Would PostgreSQL be able to handle this = comparably to Oracle? The
> licenses for Oracle are not so very expensive = anymore, but if we're
> going towards a farm of boxes, this will start = to cumulate. Otoh,
> some Oracle stuff seems to be able to handle = parallel databases
> natively. Hmm.

I think it might!  I'm working with PostgreSQL, = and I'm seeing comparable performance, but you'll need to do some = benchmarking.  If it doesn't compare now, it eventually = will.  Watch out for using joins in PostgreSQL - they can impact = run times.

>
> Sorry for the bunch of clueless questions, but = I'm not exactly a
> computer person, nor is the company extremely = competent in technical
> questions (they're all a bunch of chemists, = which just have been
> working with computers for a long time). The = budget for hardware seems
> to be there, but I'm a new guy, and I can't = afford buying a bunch
> of expensive crap which will just sit there = gathering dust.

You're not clueless 'cause you're asking the right = questions - this is the same type of thing directors and managers must = contend with.  You have a chance to sell beowulf to your = higher-ups.  Go for it!



>
> I would also welcome some pointers towards = lists where questions
> such as these are handled holistically. I'd = settle for a bunch
> of dedicated lists too, though.


>
> <off-topic/>
>
> = _______________________________________________
> Beowulf mailing list, = Beowulf@beowulf.org
> To change your subscription (digest mode or = unsubscribe)
> visit http://www.beowulf.org/mailman/listinfo/beowulf
>

------_=_NextPart_001_01C092C6.98BBCDA0-- From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:52 2008 Subject: No subject Message-ID: application. In the past, UNIX-based hardware environments had better floating-point performance, but that's been offset in the last few years by Moore's Law curves for large-volume products that have advanced faster than specialty products have, as well as the price and support cost differentials between these vendors' products. From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:52 2008 Subject: No subject Message-ID: environment, designed with priorities set by a much different market segment than traditional science and engineering. Windows NT® and now Windows 2000 were designed to meet the needs of those ISVs building products for businesses that are unable or unwilling to dedicate their best people to support their infrastructure (versus focusing on building solutions for their business mission), as well as the needs of a hardware community that required continuous integration of new devices and components." Thanks to Yoon Jae Ho for the link, I needed a good laugh today. Maybe I should send this to my friends as a joke, naa they probably wouldn't get it anyway. I have to say, although I can't really share your concern about the computing needs for Korea (being an American), I think we can all agree that the general state of ignorance towards the Beowulf project is something that needs to be remedied before Microsoft gets a chokehold on this market just like it has on the desktop market. Christoph Wasshuber wrote: > > Matt Links wrote: > > > > from http://www.microsoft.com/WINDOWS2000/hpc/indstand.asp > > > > "Until recently, most clustered computers ran UNIX or proprietary > > operating systems on proprietary hardware. Microsoft changed all > > that with Windows® 2000, the de facto industry-standard operating > > system. Now anyone with massive computing needs can create > > clusters using commercial off-the-shelf (COTS) PCs and a shrink-wrapped > > version of Windows 2000. " > > > > I don't know about the rest of you but this makes me laugh. > > It makes me very angry!! Because it is clearly a lie! Isn't there > a Better Business Bureau or somewhere to complain about? This is > clear misinformation. > > Chris.... > > _______________________________________________ > Beowulf mailing list, Beowulf@beowulf.org > To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf -- Jared Hodge Institute for Advanced Technology The University of Texas at Austin 3925 W. Braker Lane, Suite 400 Austin, Texas 78759 Phone: 512-232-4460 FAX: 512-471-9096 Email: Jared_Hodge@iat.utexas.edu _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:53 2008 Subject: No subject Message-ID: time, the nodes most of us worked with were SUN SPARCstations. We were using p4, Express, Strand88, and Linda back then --- but the "environment" that really caught people's imagination and pushed clustering forward was PVM (with its first public release in 1989 version 2.4 -- if my faulty memory severs me correctly). Throughout the early 90's, there were a series of clustering workshops at Florida State University (or was it the university of Florida --- I always get those two mixed up) where we got together and shared information about clustering. Those early clustering workshops were some of the most enjoyable workshops I've ever attended. Just about everyone who was active in clustering were at those meetings. It was an exciting time. The MPP world had nothing but scorn for us "young upstarts" and we really had to work hard to prove we were a viable option for supercomputing. If you look at those old workshop proceedings, you'll see that all the major components of a Beowulf cluster were available and in use in those days. Sure, there were major differences. For one things, we couldn't do anything to the OS kernel since we didn't have access to source code. But in terms of what the user of a cluser sees --- the parallel algorithms, parallel programming environments, batch-queue/Job-management, and cluster management systems -- we had it all way back in the early 90's. So yes, you can cluster with Sparc. It would be accurate to say -- from my point of view -- that Sparc is where clustering really took off. --Tim P.S. So now I'm ready for someone from DEC to tell me that clustering even predates SPARC. To a certain extent, that may be the case, but the old DEC clusters were not "shared nothing" clusters and were quite different beasts -- weren't they? Disclaimer: The opinions in this message are my own and do not reflect the views of my employer. -----Original Message----- From: Josip Loncaric [mailto:josip@icase.edu] Sent: Monday, February 19, 2001 7:24 AM To: Raul Romero Wells Cc: Scott.Brown@aventis.com; beowulf@beowulf.org Subject: Re: Beowulf on Sparc Raul Romero Wells wrote: > > Easy, as a Sun's dealer said: "you can instal Sun's linux on it..." > > Scott.Brown@aventis.com wrote: > > > This may seem to be a stupid question, pardon my ignorance. I'm looking for > > information on building Beowulf clusters with ultra sparc hardware. Any > > information would be greatly appreciated. Clusters do not require Linux -- in fact, a nice Solaris cluster was recently built at William & Mary: http://www.compsci.wm.edu/SciClone/ SciClone is a heterogeneous cluster, using 64 Ultra 5 (single CPU machines), 32+6 Ultra 60 (dual CPU machines), 4 Enterprise 420R servers (quad CPU each) and an Ultra 60 front end machine. This is a rich environment designed to support a wide variety of research activities. Sincerely, Josip -- Dr. Josip Loncaric, Senior Staff Scientist mailto:josip@icase.edu ICASE, Mail Stop 132C PGP key at http://www.icase.edu./~josip/ NASA Langley Research Center mailto:j.loncaric@larc.nasa.gov Hampton, VA 23681-2199, USA Tel. +1 757 864-2192 Fax +1 757 864-6134 _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:56 2008 Subject: No subject Message-ID: root node is a compute node as well, and so should be made homogeneous with the cluster. However, this is not stated explicitly that I noticed. Is this the case? Does anyone have any comments on the Racksaver RS-1200 compute node, in the 2-Athlon in 1U configuration (or even the 2-pentium in 1U config)? We like the node we have for testing, but wonder what life with 64 of them will be like. Thanks, Ray Jones MERL From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:57 2008 Subject: No subject Message-ID: computation time to communication time for even the smallest mesh of around 100 to 500:1, more for the larger meshes. If your application is CPU bound with a comp:comm time ratio like 1000:1 with a fairly small serial fraction, then it will get pretty much linear speedup out to eight nodes. If you increase it to 2000:1, then you'll get pretty much linear speedup out to eight nodes. If you increase it to 10^6:1 you'll get pretty much linear speedup out to eight nodes. All you're observing is that communications are irrelevant to your (essentially coarse grained) parallel application, at least for the switched 100BT that forms your basic network and granularities of your code. In which case yes, channel bonding is way overkill for your application, at least at the ranges you portray. Instead of buying multiple switches and trying to build fatter pipes, buy bigger switches and more nodes. Only when/where you start to see some sort of saturation and bending over of the scaling curve (like that which might be visible for 9 nodes in the 480x50 mesh) will fatter data pipes be useful to you. At that point, what increasing the network bandwidth WILL do for you is increase the number of nodes you can apply to the problem and still get (nearly) linear speedup. So on the 480x50 mesh, channel bonding and higher bandwidth keeps you near-linear a bit further out (from the look of this curve the serial fraction of your code is starting to compare to the parallel fraction, causing saturation -- maybe -- in the 8-9 node range, if this isn't an artifact of poor statistics). Even this isn't strongly affected by mere doubling (and doesn't much affect the slope of this curve < 1, which is what I'm using to guess that the serial fraction is becoming important) -- it is more of a log scale kind of thing. That is, increasing network speed by a factor of 10 might buy you a factor of two or three extension of the linear scaling region. To understand all this read the section on Amdahl's law, etc. (and look at the equations and figures) in the online beowulf book on brahma: http://www.phy.duke.edu/brahma b) The other possibility is that your communication pattern is dominated by tiny messages (still) interspersed with lots of computation. In this case your communications is probably latency dominated, not bandwidth dominated. Fattening the pipe won't help this; adding an extra channel will if anything INCREASE the latency (and hence slow the time per message). I only mention this because it is possible that you have lots of small messages whose aggregate time is still in the 1000:1 range for your larger meshes. In this case channel bonding would be a bad idea in general, not just not worth it for the scaling ranges you've achieved so far. However, if you rearrange your application to use bigger messages, it might once again be worthwhile. Hope this helps. Your scaling is really rather good out to 8-10 nodes -- eyeballing it it looks impossibly good for the largest mesh, but I rather hope this is because you are plotting the results for single run times and better statistics would make the line behave. Although there >>are<< nonlinear breakpoints -- breaking the bigger mesh up means more of it runs in cache, which can actually yield better CPU rates (which then decrease the comp:comm ratio and really make a priori estimation of the speedup curve a bitch, actually:-). rgb -- Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:57 2008 Subject: No subject Message-ID: myself. I can hack on Beowulf clustering -- including new kernel modules -- all day long and the build process usually lasts 5s of seconds. Regards, Dan Ridge Scyld Computing Corporation From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:00:59 2008 Subject: No subject Message-ID: bye, Thommy -----Original Message----- Received: from mail.mi8.com ([63.68.243.13]) by mi8smtp02.mi8.com with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2650.21) id GHCTW5JW; Sun, 4 Mar 2001 02:44:58 -0500 Received: from mout03.kundenserver.de (unverified) by mail.mi8.com (Content Technologies SMTPRS 4.1.2) with SMTP id for ; Sun, 4 Mar 2001 02:45:01 -0500 Received: from [195.20.224.238] (helo=mx05.kundenserver.de) by mout03.kundenserver.de with esmtp (Exim 2.12 #2) id 14ZTBy-0003Mq-00 for ThomasBoehme@boehme.mi8.com; Sun, 4 Mar 2001 08:44:34 +0100 Received: from [216.254.93.179] (helo=blueraja.scyld.com) by mx05.kundenserver.de with esmtp (Exim 2.12 #3) id 14ZTAJ-0006aO-00 for mail@thomas-boehme.de; Sun, 4 Mar 2001 08:42:51 +0100 Received: from blueraja.scyld.com (localhost [127.0.0.1]) by blueraja.scyld.com (8.9.3/8.9.3) with ESMTP id TAA19501; Sat, 3 Mar 2001 19:32:53 -0500 Received: from daytona.gci.com (mail.gci.com [205.140.80.57] (may be forged)) by blueraja.scyld.com (8.9.3/8.9.3) with ESMTP id VAA23625 for ; Wed, 28 Feb 2001 21:51:41 -0500 Received: from PickupDirectory by daytona.gci.com with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2650.21) id FPFVJZL7; Wed, 28 Feb 2001 17:48:03 -0900 Received: from VGER.KERNEL.ORG (199.183.24.194[199.183.24.194 port:2021]) by windsor.gci.com Mail essentials (server 2.422) with SMTP id: <331549@windsor.gci.com> for ; Wed, 28 Feb 2001 2:08:52 PM -0900 smtpmailfrom Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 28 Feb 2001 18:06:05 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 28 Feb 2001 18:05:57 -0500 Received: from adsl-64-167-146-94.dsl.snfc21.pacbell.net ([64.167.146.94]:43784 "EHLO eleanor.wdhq.scyld.com") by vger.kernel.org with ESMTP id ; Wed, 28 Feb 2001 18:05:46 -0500 Received: from localhost (newt@localhost) by eleanor.wdhq.scyld.com (8.9.3/8.9.3) with ESMTP id SAA05593; Wed, 28 Feb 2001 18:06:37 -0500 X-Authentication-Warning: eleanor.wdhq.scyld.com: newt owned process doing -bs Date: Wed, 28 Feb 2001 18:06:37 -0500 (EST) From: Daniel Ridge X-Sender: newt@eleanor.wdhq.scyld.com To: beowulf@beowulf.org cc: Linux Kernel Mailing List Subject: Re: Will Mosix go into the standard kernel? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org Sender: beowulf-admin@beowulf.org Errors-To: beowulf-admin@beowulf.org X-Mailman-Version: 1.1 Precedence: bulk List-Id: Discussion of topics related to Beowulf clusters X-BeenThere: beowulf@beowulf.org Fellow Beowulfers, I have yet to hear a compelling argument about why any of them should go into the standard kernel -- let alone a particular one or a duck of a compromise. The Scyld system is based on BProc -- which requires only a 1K patch to the kernel. This patch adds 339 net lines to the kernel, and changes 38 existing lines. The Scyld 2-kernel-monte kernel inplace reboot facility is a 600-line module which doesn't require any patches whatsoever. Compare this total volume to the thousands of lines of patches that RedHat or VA add to their kernel RPMS before shipping. I just don't see the value in fighting about what clustering should 'mean' or picking winners when it's just not a real problem. Scyld is shipping a for-real commercial product based on BProc and 2-kernel-Monte and our better-than-stock implementation of LFS and and we're not losing any sleep over this issue. I think we should instead focus our collective will on removing things from the kernel. For years, projects like ALSA, pcmcia-cs, and VMware have done an outstanding job sans 'inclusion' and we should more frequently have the courage to do the same. RedHat and other linux vendors have demonstrated ably that they know how to build and package systems that draw together these components in an essentially reasonable way. Regards, Dan Ridge Scyld Computing Corporation On Tue, 27 Feb 2001, Rik van Riel wrote: On Tue, 27 Feb 2001, David L. Nicol wrote: > > I've thought that it would be good to break up the different > > clustering frills -- node identification, process migration, > > process hosting, distributed memory, yadda yadda blah, into > > separate bite-sized portions. > > It would also be good to share parts of the infrastructure > between the different clustering architectures ... > > > Is there a good list to discuss this on? Is this the list? > > Which pieces of clustering-scheme patches would be good to have? > I know each of the cluster projects have mailing lists, but > I've never heard of a list where the different projects come > together to eg. find out which parts of the infrastructure > they could share, or ... > > Since I agree with you that we need such a place, I've just > created a mailing list: > > linux-cluster@nl.linux.org > > To subscribe to the list, send an email with the text > "subscribe linux-cluster" to: > > majordomo@nl.linux.org > > > I hope that we'll be able to split out some infrastructure > stuff from the different cluster projects and we'll be able > to put cluster support into the kernel in such a way that > we won't have to make the choice which of the N+1 cluster > projects should make it into the kernel... > > regards, > > Rik > -- > Linux MM bugzilla: http://linux-mm.org/bugzilla.shtml > > Virtual memory is like a game you can't win; > However, without VM there's truly nothing to lose... > > http://www.surriel.com/ > http://www.conectiva.com/ http://distro.conectiva.com/ > > > > > _______________________________________________ > Beowulf mailing list, Beowulf@beowulf.org > To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf > - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf From rogers at seuss.chem.upenn.edu Sat May 17 01:00:01 2008 From: rogers at seuss.chem.upenn.edu (Christopher L. Rogers) Date: Sat May 17 01:01:03 2008 Subject: Memory Issues: /proc/kcore vs. free Message-ID: <200103221352.IAA291162@seuss.chem.upenn.edu> folks: i am trying to determine whether or not i have some bad memory or not on some of my beowulf nodes. One thing i noticed right away is that ls -l of /proc/kcore maxs out at 940 MB for physical ram greater than 1 GB. I'm running 2048 MB, which the bios detects. free reports 2074 MB but as i said earlier /proc/kcore reports 940 MB. (i do have mem=2048M appended in my lilo.conf) So is this cause for alarm? Which is the true amount the system sees? I need to routinely run jobs on these nodes that are about 1 to 1.25 GB., so it makes a difference. The system in question is a dual PIII Supermicro 370DL3 with their latest bios. The bios reports the 2 GB of ram as PC133, and the vendor claims it's true registered ECC. The kernel is a stock Redhat 7.0 "enterprise" kernel (2.2.16-22enterprise) Thanks for all your help. -Chris Rogers crogers@sas.upenn.edu From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:05 2008 Subject: No subject Message-ID: extract) "2.2 What is a Beowulf ?=20 Famed was this Beowulf: far flew the boast of him, son of Scyld, in the = Scandian lands. So becomes it a youth to quit him well with his father's = friends, by fee and gift, that to aid him, aged, in after days, come = warriors willing, should war draw nigh, liegemen loyal: by lauded deeds = shall an earl have honor in every clan. Beowulf is the earliest = surviving epic poem written in English. It is a story about a hero of = great strength and courage who defeted a monster called Grendel. See = History to find out more about the Beowulf hero.=20 There are probably as many Beowulf definitions as there are people who = build or use Beowulf Supercomputer facilities. Some claim that one can = call their system Beowulf only if it is built in the same way as the = NASA's original machine. Others go to the other extreme and call Beowulf = any system of workstations running parallel code. My definition of = Beowulf fits somewhere between the two views described above, and is = based on many postings to the Beowulf mailing list: " What this may not clearly identify clearly is the 'motivation' for = calling these machines by this name. The 'lore' I've hard is that the = 'Grendel' monster that one wanted to {destroy - or avoid at least} was = the then-typical 'supercomputer' - with it's (at least) monster-like = appetite for ($, maintenance costs, downtime, programming investment). = On the other hand, the vendors of the day tended to identify their = 'best' machines by names more like 'Thor' I think. I wouldn't be = surprised if the person who suggested the moniker visits here, though, = so maybe we'll get the 'true' story. ------=_NextPart_000_004E_01C0BAAD.C7EA3F20 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
From Jacek Radajewski & Douglas Eadline's excellent 'Beowulf = HowTo' see=20 the introduction:  http://www= .linuxdoc.org/HOWTO/Beowulf-HOWTO.html#toc2 =20 (I hope that I don't need to ask for copyright approval for this = extract)
 

Famed was this Beowulf: far flew the boast of him, son of Scyld, = in the=20 Scandian lands. So becomes it a youth to quit him well with his father's = friends, by fee and gift, that to aid him, aged, in after days, come = warriors=20 willing, should war draw nigh, liegemen loyal: by lauded deeds shall an = earl=20 have honor in every clan. Beowulf is the earliest surviving epic = poem=20 written in English. It is a story about a hero of great strength and = courage who=20 defeted a monster called Grendel. See Histo= ry to=20 find out more about the Beowulf hero.=20

There are probably as many Beowulf definitions as there are people = who build=20 or use Beowulf Supercomputer facilities. Some claim that one can call = their=20 system Beowulf only if it is built in the same way as the NASA's = original=20 machine. Others go to the other extreme and call Beowulf any system of=20 workstations running parallel code. My definition of Beowulf fits = somewhere=20 between the two views described above, and is based on many postings to = the=20 Beowulf mailing list: "

What this may not clearly identify clearly is the 'motivation' for = calling=20 these machines by this name.  The 'lore' I've hard is that the = 'Grendel'=20 monster that one wanted to {destroy - or avoid at least} was the = then-typical=20 'supercomputer' - with it's (at least) monster-like appetite for ($, = maintenance=20 costs, downtime, programming investment).  On the other hand, the = vendors=20 of the day tended to identify their 'best' machines by names more like = 'Thor' I=20 think.  I wouldn't be surprised if the person who suggested the = moniker=20 visits here, though, so maybe we'll get the 'true' story.

 

------=_NextPart_000_004E_01C0BAAD.C7EA3F20-- From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:07 2008 Subject: No subject Message-ID: > "Emacs is a hideous monstrosity, but a functional one. > On the other hand, vi is a masterpiece of elegance. Sort > of like a Swiss Army knife versus a rapier." Also note: http://www.splange.freeserve.co.uk/img/viman.gif From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:07 2008 Subject: No subject Message-ID: > Felix S. Gallo wrote this account > of a paintball match at which was decided the eternal > question of which Unix text editor is objectively superior. > > > vi trounced emacs, 3-1, in a four-game paintball match. > > Apparently the emacs users were confused by the simple > > nature of their paintball equipment, or were stricken by > > carpal tunnel syndrome... Rumors that emacs really lost > > because their guns took 5 minutes to load, weighed 500 lbs, > > included an undecipherable and outmoded built-in scripting > > language, and had 19 different modifier keys next to the > > trigger, were deemed baseless. Commented one grinning vi > > user, ":1,$s/emacs/lunchmeat/g, baby!" From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:07 2008 Subject: No subject Message-ID: > I have to confess that I'm a "vi guy." When I was new to > UNIX, I actually started out using emacs (not GNU emacs, but > Gosling's Emacs, in the early days of the emacs wars) and > loved it. But then one day, I discovered that someone had > blown away the custom emacs profile that I had gotten > comfortable with. It took about a week for me to get it > restored, and in the meantime, I switched to vi, and got > hooked. > > What I especially liked about vi was precisely that I didn't > have to have all these nifty customizations to make it > really usable. What really distinguished vi from emacs in my > mind was that it was just there on any system I sat down at- > -tremendously portable. I didn't have to copy my emacs > profile around to sit down and do useful work on someone > else's machine. From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:07 2008 Subject: No subject Message-ID: > I think psychoanalyze-pinhead is the important lesson > of GNU Emacs. > > ---Bennett Todd In case you don't know what he's talking about here, see this post to the Humanist Discussion Group, from the archive page at: http://lists.village.virginia.edu/lists_archive/Humanist/v12/0506.html > Date: Thu, 18 Mar 1999 07:27:50 +0000 > From: C M Sperberg-McQueen > Subject: Re: 12.0499 ... Eliza? > > > On Tue, 16 Mar 1999 22:13:34 +0000 (BST), in Humanist 12.0499, > Jim Marchand asked: > > > > What has happened to Eliza? If you remember, it was a > > great program written by Josef Weizenbaum way back when > > we had no memory. As a kind of Rogerian psychologist, > > it used your questions to reflect answers back to you. > > > Strictly speaking, Eliza was a general script-driven > pattern-matching and reply-generating program; the Rogerian > analyst was provided by the Doctor script. Since it was the > funniest and made the most plausible use of Eliza's radical > lack of real-world knowledge -- and also because it was the > only script included in the original publication, perhaps -- > the Doctor script is the only thing Eliza was ever known > for. > > > > I used it for a number of purposes back when I was > > teaching and wrote programs such as Freud, The Bavarian > > Inkeeper, The Wimpy Psychologist, which were fun, if > > not always good. ... > > > > Anyway, one hears that a psychologist out in California > > is still using the program, but I haven't seen it > > mentioned in years. Maybe I'm just reading the wrong > > things. Does anybody know anything about the present > > state of Eliza? > > Weizenbaum's Eliza was never, as far as I know, a supported > product, so it's present state is presumably the same as > ever: a clever program described in a journal article (the > article I know was in Communications of the ACM, but there > may have been others). > > > Undergraduates are sometimes given the task of > reimplementing Eliza (in general, or with a hard-coded > script), and there are lots of versions around. In that > sense, Eliza will surely never die while computers are used > with text interfaces. > > > Perhaps the most ubiquitous version of Eliza now available > is the 'doctor' program shipped as part of GNU emacs (M-x- > doctor, for those of you who have not tried it). This may be > used alone, but (emacs being what it is) can also be hooked > up to take its input not from the user, but from a database > of quotations from Zippy the Pinhead (M-x-psychoanalyze- > pinhead, a short sample of which is appended). If anyone has > been wondering whether to learn vi or emacs, the existence > of psychoanalyze-pinhead in the emacs distribution should > definitely tip the balance. One way or the other. > > > -C. M. Sperberg-McQueen Senior Research Programmer, > University of Illinois at Chicago > > > ----- A few lines of output from psychoanalyze-pinhead ----- > I am the psychotherapist. For your protection, I have been > censored according to the Communications Decency Act. > Please, describe your problems. Each time you are finished > talking, type RET twice. > > > I'm into SOFTWARE! > > > Why do you say you are into software? > > > I'm GLAD I remembered to XEROX all my UNDERSHIRTS!! > > > What makes you believe that? > > > Everybody is going somewhere!! It's probably a garage sale > or a disaster Movie!! > > > Maybe your plans have something to do with this. > > > Now I'm having INSIPID THOUGHTS about the beautiful, round > wives of HOLLYWOOD MOVIE MOGULS encased in PLEXIGLAS CARS > and being approached by SMALL BOYS selling FRUIT.. > > > Is it because of your life that you say and being approached > by small boys selling fruit? > > > My EARS are GONE!! > > > Are you sure that your ears are gone? > > > While my BRAINPAN is being refused service in BURGER KING, > Jesuit priests are DATING CAREER DIPLOMATS!! > > > Is it because of the people you hang around with that you > are going through all this? > > > .... etc., etc., etc. From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:08 2008 Subject: No subject Message-ID: the 100 MHz FSB. The increase that you would see from 800 MHz to 866 or 933 MHz would be marginal, but so is the cost. Get the what you can with the money you have (after other expenses. > * Can any ECC SDRAM fit in 370DLE mobo?. I do remember that someone > mentioned that high quality memory chips should be used. For 370DLE, what > can be the best choice of memory chips?. Any brand? You should see the required (and recommended) memory listing that Supermicro has. http://www.supermicro.com/TECHSUPPORT/FAQs/Memory_vendors.htm > Network > Since the cluster size is small, we will afford gigabit networking. > 3Com Superstack 3 Switch 4900 (12-port) (on copper wire) (it is one of the > cheapest gigabit switch which is around $5000; is there any other > alternative around this price?) > 3Com Gigabit NICs (is there another brand which is more cost-effective?.) I'm not an expert on any of these, all I can say is don't skimp on the network. Many clusters are built around the network (like ours) since many applications are very bandwidth/latency sensitive. Try talking to network tech support and seeing what motherboards, etc. they recommend. I doubt they'll have any problems with the 370DLE, but it's best to check. -- Jared Hodge Institute for Advanced Technology The University of Texas at Austin 3925 W. Braker Lane, Suite 400 Austin, Texas 78759 Phone: 512-232-4460 FAX: 512-471-9096 Email: Jared_Hodge@iat.utexas.edu From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:18 2008 Subject: No subject Message-ID: difference between "shared memory" programming and "message passing" programming. Most clusters don't offer shared memory services. At most they might provide one-sided messaging like the Cray SHMEM library, which is message passing that can be written relatively compactly compared to MPI, the usual library people use for message passing. -- greg From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:20 2008 Subject: No subject Message-ID: system. There are only two motherboards that I know of that offer video, dual ethernet and ATA100 they are Tyan's 2515, and GigaByte's new GA-6VXDR7. The Tyan had some problems(what else is new) but a new firmware release should of fixed them. The GigaByte GA-6VXDR7 has the VIA Apollo chipset and from what we've seen about 6~8% slower...but about $90.00 cheaper that Tyan's 2515. Cheers, Steve Gaudet ..... <(©¿©)> ---------------------- Angstrom Microsystems 200 Linclon St., Suite 401 Boston, MA 02111-2418 pH:617-695-0137 ext 27 home office:603-472-5115 cell:603-498-1600 e-mail:sgaudet@angstrommicro.com http://www.angstrommicro.com From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:21 2008 Subject: No subject Message-ID: guess they have their own way of doing it. David From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:25 2008 Subject: No subject Message-ID: Much of the energy consumed by data centers is needed to cool the buildings to keep the equipment from overheating. Backers of the Netricity project say the North Slope's cold temperatures and dry air make it a perfect environment for a server farm and its isolation provides security. Heat from the servers, together with waste heat from the power plant, could be used to warm the building. Any cost of heating would be minimal compared to cooling in the south, Dodson said. http://www.latimes.com/wires/wbusiness/20010618/tCB00V5140.html Steve Wickert Protein Pathways From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:28 2008 Subject: No subject Message-ID: to generate eficient code than the hardware performance. I made some test with STREAM on a DS20 system with gcc and Compaq C, results were quite differents. I think it's dificult to conclude something on memory bandwidth with a gcc compiled STREAM. A+ -- Guignon Thomas, guignon@artabel.net Tel: 01 69 18 95 95 Fax: 01 69 18 95 96 ARTABEL S.A. 69, rue de Paris 91400 Orsay From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:29 2008 Subject: No subject Message-ID: business environment (like for transactions processing or stress analysis), this cost is very much a reality. It's important for our directors/administrators/venture capitalists to understand how we are going to deal with the "incremental costs" of our clusters. Keeping track of the time to recompile our code (as effortless as it may be), and comparing that with, say upgrading nodes with Windows NT is something your boss will appreciate. The time it takes for Intel to stop producing the I32 architectures is your timeline to work with when establishing a porting cost per unit of time. Peter Dell touched on this in a recent paper . . . Richard Schilling Webmaster / Web Integration Programmer Affiliated Health Services Mount Vernon, WA USA http://www.affiliatedhealth.org phone: 360.856.7129 From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:30 2008 Subject: No subject Message-ID: The TCP/IP connections could not be established for some reason. I purposely started an FTP server on the slave so that I can test forming TCP/IP connections afterward. ftp'ing to the slave just hung, although I could sniff 2-way packets from 10.0.0.2-615 to 10.0.0.1-2223 (bpmaster's port). I'm guessing that the connection was spoiled since the slave put up this message: bproc: connect: connect failed, errno=111 bpslave: short read - lost connection to master rebooting in 30 seconds "ifconfig -a" on the master and slave showed that they were properly bonded. I also tricked the slave into giving me a shell so that I could type stuff after it lost connection with the master. I did this by "bpsh 0 csh -f /tmp/shell" /tmp/shell looks like: tcsh < /dev/console > /dev/console From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:30 2008 Subject: No subject Message-ID: showed that it was properly bonded. Of course, it rebooted me in 30 seconds :-( At this point, I'll buy a whole new switch if it makes my life easier! Any ideas? -Mike -- Michael J. Weller, M.Sc. office: (972) 235-7881 x.242 weller@zyvex.com cell: (214) 616-6340 Zyvex Corp., 1321 N Plano facsimile: (972) 235-7882 Richardson, TX 75081 icq: 6180540 From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:35 2008 Subject: No subject Message-ID: 1 NIC transfer with FTP: (PUT and GET): 150 Opening BINARY mode data connection for kern.tar.gz. 8186400 bytes sent in 0.765 secs (1e+04 Kbytes/sec) 150 Opening BINARY mode data connection for kern.tar.gz (8186400 bytes). 8186400 bytes received in 0.779 secs (1e+04 Kbytes/sec) 2 NICs: [root@n0 /tmp]# insmod bonding Using /lib/modules/2.4.7mw2/kernel/drivers/net/bonding.o [root@n0 /tmp]# ifconfig bond0 inet 10.0.0.2 netmask 255.255.255.0 [root@n0 /tmp]# ifenslave bond0 eth0 master has no hw address assigned; getting one from slave! The interface eth0 is up, shutting it down it to enslave it. [root@n0 /tmp]# ifenslave bond0 eth1 FTP transfer: 150 Opening BINARY mode data connection for kern.tar.gz. 8186400 bytes sent in 1.44 secs (5.6e+03 Kbytes/sec) 150 Opening BINARY mode data connection for kern.tar.gz (8186400 bytes). 8186400 bytes received in 1.28 secs (6.2e+03 Kbytes/sec) So, it's not Scyld's 2.2 kernel causing this. The switches are capable of performing 10000 Kbytes/sec as proven by the single-NIC transfer rates, so one would think that it's completely OS-dependent. To further verify that the BONDING is setup properly, I monitored the RX and TX packets of the bond0, eth0 and eth1 interfaces on both ends, to make sure that BOTH NICs were operational: One bug I noticed was that the RX packets for bond0 is 0. You can add up TX for eth0 and eth1 to give you the TX for bond0. I don't know if this is relevant since it's functional: [root@n0 /root]# ifconfig -a bond0 Link encap:Ethernet HWaddr 00:01:02:ED:C4:C8 inet addr:10.0.0.2 Bcast:10.255.255.255 Mask:255.255.255.0 UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:42282 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:36870337 (35.1 Mb) eth0 Link encap:Ethernet HWaddr 00:01:02:ED:C4:C8 inet addr:10.0.0.2 Bcast:10.255.255.255 Mask:255.255.255.0 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1 RX packets:31673 errors:0 dropped:0 overruns:488 frame:0 TX packets:39068 errors:0 dropped:0 overruns:0 carrier:6 collisions:0 txqueuelen:100 RX bytes:24773173 (23.6 Mb) TX bytes:40110808 (38.2 Mb) Interrupt:11 Base address:0xe000 eth1 Link encap:Ethernet HWaddr 00:01:02:ED:C4:C8 inet addr:10.0.0.2 Bcast:10.255.255.255 Mask:255.255.255.0 UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1 RX packets:17075 errors:0 dropped:0 overruns:458 frame:0 TX packets:17475 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100 RX bytes:13483933 (12.8 Mb) TX bytes:14076285 (13.4 Mb) Interrupt:9 Base address:0xe400 Using my "measure" script which takes differences, here's a before and after: root@n0 /root]# ./measure bond0 rx 0 tx 0 eth0 rx 0 tx 0 eth1 rx 0 tx 0 [root@n0 /root]# ./measure bond0 rx 0 tx 6532 eth0 rx 3269 tx 3266 eth1 rx 3264 tx 3266 The rx's and tx's matched on the other end as well. the RX for bond0 is always 0 (bug?) Any idea on what I should try next? Thanks! -- Michael J. Weller, M.Sc. office: (972) 235-7881 x.242 weller@zyvex.com cell: (214) 616-6340 Zyvex Corp., 1321 N Plano facsimile: (972) 235-7882 Richardson, TX 75081 icq: 6180540 From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:42 2008 Subject: No subject Message-ID: consumption of each node. We have an Athlon Cluster(1.33 GHZ) and those things are very power hungry. We underestimated how much they draw and had to install an extra circuit in addition to the ones we had planned for... Perhaps there is yet another efficiency measure - MFlops/kwatt or something like that. So for the I-cluster that would have been 75 Gflops/(210 * 50 W/1000) = 7.14GFlops/kW Roger, out of curiosity could you do a similar calculation? Thanks, Scott Shealy ----- Original Message ----- From: "Roger L. Smith" To: "RICHARD,BRUNO (HP-France,ex1)" Cc: "'Scott Shealy'" ; Sent: Thursday, September 27, 2001 2:38 PM Subject: RE: Paper showing Linpack scalability of mainstream clusters > > Bruno, > > Hmmm, I thought that some of the others were purely ethernet based, but > after doing some quick research, I guess I'll stand corrected. > > If you intend "mainstream" to mean only single-processor desktop-type > machines, then I'll completely concede your point. > > As of the last Top 500 list, our cluster was listed as 158th, and is > entirely ethernet based, using a single 100Mb/s interconnect to each node, > and GigE interconnects between switches. However, our nodes are 1U with > dual processors, so I guess maybe it doesn't fit the definition of > mainstream that you stated. It doesn't predate the I-cluster, but it does > at least tie with it. :-) > > -Roger > > On Thu, 27 Sep 2001, RICHARD,BRUNO (HP-France,ex1) wrote: > > > Hi Roger, > > > > What we mean by "mainstream" is that these could be your grandma's machine, > > unmodified except for the Software. > > Actually I-Cluster *is* the first cluster of this type to enter the TOP500. > > Some PC-based clusters are already registered there of course, but they > > cannot be called "mainstream": Most require specific (non-mainstream at > > all!) connectivity such as Myrinet, SCI, Quadrix... Some are based on PCs > > equipped with several LAN boards (not mainstream either). > > If you restrict to off-the-shelf monoprocessor (excluding non-mainstream > > Alpha, MIPS and such) interconnected through standard Ether100, no cluster > > ever entered the TOP500 list. And I-Cluster is still the only one to be > > there. Let me know if you think otherwise. > > > > Regards, -bruno > > _____________________________________________ > > Bruno RICHARD - Research Program Manager > > HP Laboratories > > 38053 Grenoble Cedex 9 - FRANCE > > Phone: +33 (4) 76 14 15 38 > > bruno_richard@hp.com > > > > > > -----Original Message----- > > From: Roger L. Smith [mailto:roger@ERC.MsState.Edu] > > Sent: Thursday, September 27, 2001 15:33 > > To: RICHARD,BRUNO (HP-France,ex1) > > Cc: 'Scott Shealy'; beowulf@beowulf.org > > Subject: RE: Paper showing Linpack scalability of mainstream clusters > > > > > > On Tue, 18 Sep 2001, RICHARD,BRUNO (HP-France,ex1) wrote: > > > > > Sorry Scott, I sent you a wrong reference. The actual link is > > > http://www.hpl.hp.com/techreports/2001/HPL-2001-206.html. Enjoy, > > > -bruno > > > > > > I'd be REALLY interested in hearing how you justify the following statement > > in the paper: > > > > "Being the first ones to enter the TOP500 using only mainstream hardware > > (standard PCs, standard Ethernet connectivity)...". > > > > _\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_ > > | Roger L. Smith Phone: 662-325-3625 | > > | Systems Administrator FAX: 662-325-7692 | > > | roger@ERC.MsState.Edu http://WWW.ERC.MsState.Edu/~roger | > > | Mississippi State University | > > |_______________________Engineering Research > > |Center_______________________| > > > > > _\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_ > | Roger L. Smith Phone: 662-325-3625 | > | Systems Administrator FAX: 662-325-7692 | > | roger@ERC.MsState.Edu http://WWW.ERC.MsState.Edu/~roger | > | Mississippi State University | > |_______________________Engineering Research Center_______________________| From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:43 2008 Subject: No subject Message-ID: - is it a LAM bug? - is it a 3c59x driver bug? - is it a 2.4 kernel bug? Besides this problem I have encountered by now several RedHat 7.1 machines on campus (UP or SMP) that had network problems which could be solved by including the "noapic" option in lilo.conf. Are there chances that the APIC problems in the 2.4 kernels are resolved soon (there seem to be changes to the APIC code in 2.4.10, but I still have problems)? Is there a performance hit related to the "noapic" option? Anyway, with the release of mpich-1.2.2 this problem isn't as pressing anymore as it was a few weeks ago. The performance of MPI jobs under mpich-1.2.2 is much improved, particularly for smaller message sizes. Big thankyou to the mpich developpers! Martin ======================================================================== Martin Siegert Academic Computing Services phone: (604) 291-4691 Simon Fraser University fax: (604) 291-4242 Burnaby, British Columbia email: siegert@sfu.ca Canada V5A 1S6 ======================================================================== From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:44 2008 Subject: No subject Message-ID: consumption of each node. We have an Athlon Cluster(1.33 GHZ) and those things are very power hungry. We underestimated how much they draw and had to install an extra circuit in addition to the ones we had planned for... Perhaps there is yet another efficiency measure - MFlops/kwatt or something like that. So for the I-cluster that would have been 75 Gflops/(210 * 50 W/1000) = 7.14GFlops/kW Roger, out of curiosity could you do a similar calculation? Thanks, Scott Shealy ----- Original Message ----- From: "Roger L. Smith" To: "RICHARD,BRUNO (HP-France,ex1)" Cc: "'Scott Shealy'" ; Sent: Thursday, September 27, 2001 2:38 PM Subject: RE: Paper showing Linpack scalability of mainstream clusters > > Bruno, > > Hmmm, I thought that some of the others were purely ethernet based, > but after doing some quick research, I guess I'll stand corrected. > > If you intend "mainstream" to mean only single-processor desktop-type > machines, then I'll completely concede your point. > > As of the last Top 500 list, our cluster was listed as 158th, and is > entirely ethernet based, using a single 100Mb/s interconnect to each > node, and GigE interconnects between switches. However, our nodes are > 1U with dual processors, so I guess maybe it doesn't fit the > definition of mainstream that you stated. It doesn't predate the > I-cluster, but it does at least tie with it. :-) > > -Roger > > On Thu, 27 Sep 2001, RICHARD,BRUNO (HP-France,ex1) wrote: > > > Hi Roger, > > > > What we mean by "mainstream" is that these could be your grandma's machine, > > unmodified except for the Software. > > Actually I-Cluster *is* the first cluster of this type to enter the TOP500. > > Some PC-based clusters are already registered there of course, but > > they cannot be called "mainstream": Most require specific > > (non-mainstream at > > all!) connectivity such as Myrinet, SCI, Quadrix... Some are based on PCs > > equipped with several LAN boards (not mainstream either). If you > > restrict to off-the-shelf monoprocessor (excluding non-mainstream > > Alpha, MIPS and such) interconnected through standard Ether100, no cluster > > ever entered the TOP500 list. And I-Cluster is still the only one to > > be there. Let me know if you think otherwise. > > > > Regards, -bruno _____________________________________________ > > Bruno RICHARD - Research Program Manager > > HP Laboratories > > 38053 Grenoble Cedex 9 - FRANCE > > Phone: +33 (4) 76 14 15 38 > > bruno_richard@hp.com > > > > > > -----Original Message----- > > From: Roger L. Smith [mailto:roger@ERC.MsState.Edu] > > Sent: Thursday, September 27, 2001 15:33 > > To: RICHARD,BRUNO (HP-France,ex1) > > Cc: 'Scott Shealy'; beowulf@beowulf.org > > Subject: RE: Paper showing Linpack scalability of mainstream clusters > > > > > > On Tue, 18 Sep 2001, RICHARD,BRUNO (HP-France,ex1) wrote: > > > > > Sorry Scott, I sent you a wrong reference. The actual link is > > > http://www.hpl.hp.com/techreports/2001/HPL-2001-206.html. Enjoy, > > > -bruno > > > > > > I'd be REALLY interested in hearing how you justify the following statement > > in the paper: > > > > "Being the first ones to enter the TOP500 using only mainstream hardware > > (standard PCs, standard Ethernet connectivity)...". > > > > _\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_ > > | Roger L. Smith Phone: 662-325-3625 | > > | Systems Administrator FAX: 662-325-7692 | > > | roger@ERC.MsState.Edu http://WWW.ERC.MsState.Edu/~roger | > > | Mississippi State University | > > |_______________________Engineering Research > > |Center_______________________| > > > > > _\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_ > | Roger L. Smith Phone: 662-325-3625 | > | Systems Administrator FAX: 662-325-7692 | > | roger@ERC.MsState.Edu http://WWW.ERC.MsState.Edu/~roger | > | Mississippi State University | > |_______________________Engineering Research Center_______________________| From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:44 2008 Subject: No subject Message-ID: there is also a yp lookup that happens as long as you have files and nis in nsswitch.conf, even if the user is root and happens to be found in files. Has anyone else seen this? Steve Timm ------------------------------------------------------------------ Steven C. Timm (630) 840-8525 timm@fnal.gov http://home.fnal.gov/~timm/ Fermilab Computing Division/Operating Systems Support Scientific Computing Support Group--Computing Farms Operations On Thu, 4 Oct 2001, Tim Carlson wrote: > On Thu, 4 Oct 2001, Donald Becker wrote: > > > > If you were running > > > 1000 small jobs in a couple of minutes I could imagine having problems > > > authenticating against any non-local mechanism. > > > > Hmmm, a reasonable goal is running a small cluster-wide job every > > second. I suspect the NIS delays alone take longer than one second with > > just a few nodes. > > So I ran the following test on one of our small clusters. > > 6 client NIS nodes with one NIS master (front end node) and no NIS slave > servers. Dual 800Mhz Pentium IIIs connected on a fast ethernet switch. > Forgive my sloppy C shell programming :) > > The "script" which is basically 100 rsh calls and some NIS work on > looking up the ownership of a file. > > I am doing an ls on /tmp which contains only 3 or 4 files, but I own two > of them so NIS is consulted for file ownership. I took NFS delays out by > going to /tmp. > > > #!/bin/tcsh > set i=0 > while ($i < 100) > rsh $1 ls -l /tmp > /dev/null > set i=`expr $i + 1` > end > > [tim@frontend-0 tim]$ time ./script compute-0-0 > real 0m12.704s > user 0m0.520s > sys 0m0.440s > > So if the job takes zero time and connecting to a machine takes zero time > then the NIS overhead is about 1/8 of a second. I ran this a half a dozen > times and the run varied between 10 and 13 seconds. > > Now I point this script at 6 nodes at the same time (or at least as fast > as I can type a return in 6 xterms) and the mean time per run is about 31 > seconds. That puts my potential NIS delay at a maximum of 1/3 of a > second. But I have also launched 600 jobs in 31 seconds. > > Two examples from the larger test: > > [tim@frontend-0 tim]$ date; time ./script compute-0-0 > Thu Oct 4 21:06:53 PDT 2001 > > real 0m30.905s > user 0m0.600s > sys 0m0.540s > > [tim@frontend-0 tim]$ date; time ./script compute-0-2 > Thu Oct 4 21:06:52 PDT 2001 > > real 0m30.075s > user 0m0.530s > sys 0m0.710s > > > Before and after "ps -ax | grep ypserv" on the master node. > 639 ? S 73:08 ypserv > 639 ? S 73:10 ypserv > So I used 2 seconds of CPU time with ypserv > > > My first version of the script was a "touch /tmp/testfile" and produced > similar results. My /etc/nsswitch.conf files go "files nis" and the only > entry in /etc/passwd on the compute nodes is root > > I am willing to be enlightened as to how my test is flawed. I'll run > different tests if asked. Is my test too trivial? > > Tim > > Tim Carlson > Voice: (509) 376-0300 > Email: Tim.Carlson@pnl.gov > EMSL UNIX System Support > > > _______________________________________________ > Beowulf mailing list, Beowulf@beowulf.org > To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf > From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:45 2008 Subject: No subject Message-ID: driver is an external driver provided by Qlogic and incorporated into the "redhat" version of the Linux kernel. The version I have is current w/ the latest off their website. It's hard to find any information regarding linux and these devices, since most people don't have this equipment. Does anyone here have any experience or suggestions. Thanks in advance! Regards, Derek R. -- Junior Linux Geek 713-817-1197 (cell) 713-781-4000 x2267 (office) "Linux users, fanatical. No way... HEY! Get that MCSE up on the altar, Tux must be appeased!" --=-x1H/azGXuumGFHORUIzY Content-Type: text/html; charset=utf-8 All,
Does anyone have any experience or know any gotchas when connecting a node to a fibre switch (that drives a tape SAN)?  Here's the hardware and situation :

gateway node is an IBM x340 (dual Intel PIII, 2Gb RAM, etc.  fairly generic high-end server-class machine) w/ a QLogic 2200 series Fibre Channel adapter, attached to an IBM SAN Fibre Channel Switch 2109 Model S16 (actually a number of them interconnected, ea. 16 port).  There are a number of IBM Magstar 3590E1 model tape drives (high-end geophysical use) attached to the SAN switch.  Some of these tape drives are connected directly to the switch (fibre from tape drive to switch), others are UltraWide Differential SCSI going to a switch that converts that to fibre, then to the main fibre switch.

We are running Redhat Linux 7.1, kernel 2.4.3 (redhat updates version, w/ the extra patches/tweaks and drivers they include), all patches and updates that applicable installed.

So far, I have connected an individual tape drive directly to the node, everything came up fine, could read/write, run all mt commands, etc. just fine (performance was actually a good deal better than w/ the SCSI connections I've previously used).  Then I plugged the node directly into the fibre switch and booted it.  The kernel boots, sees devices, sees the Qlogic adapter, sees the 29 devices and their addresses attached to it, and then just sits there.  The messages are :

( at boot time - w/out the link to the fibre switch plugged in - from /var/log/messages )
...
qla2x00: detect() found an HBA
qla2x00: VID=1077 DID=2200 SSVID=1077 SSDID=2
(scsi): Found a QLA2200  @ bus 1, device 0x5, irq 20, iobase 0x4d00
scsi(2): Configure NVRAM parameters...
scsi(2): Verifying loaded RISC code...
scsi(2): Verifying chip...
scsi(2): Waiting for LIP to complete...
scsi-qla0-adapter-node=200000e08b034ec1;
scsi-qla0-adapter-port=210000e08b034ec1;
scsi2 : QLogic QLA2200 PCI to Fibre Channel Host Adapter: bus 1 device 5 irq 20
        Firmware version:  2.01.27, Driver version 4.28
...

( same - from /var/log/dmesg )

identical to above

( when link is plugged in -  from /var/log/messages )
...
scsi(2): LIP reset occurred
scsi(2): LOOP UP detected
scsi(2): Waiting for LIP to complete...
scsi2: Topology - (F_Port), Host Loop address  0xffff
qla2100_update_host: Host table Full.scsi2: Topology - (F_Port), Host Loop address  0xffff
scsi2: Topology - (F_Port), Host Loop address  0xffff
...

( when link is plugged in - dmesg output )

same as above

( from /proc/scsi/qla2x00/2 )

QLogic PCI to Fibre Channel Host Adapter for ISP2100/ISP2200/ISP2300:
        Firmware version:  2.01.27, Driver version 4.28
Request Queue = 0x1fb38000, Response Queue = 0x1fb30000
Request Queue count= 256, Response Queue count= 256
Number of pending commands = 0x0
Number of queued commands = 0x0
Number of free request entries = 251
Number of mailbox timeouts = 0
Number of ISP aborts = 0
Number of loop resyncs = 0
Number of retries for empty slots = 0
Number of reqs in retry_q = 0
Number of reqs in done_q = 0

SCSI Device Information:
scsi-qla0-adapter-node=200000e08b034ec1;
scsi-qla0-adapter-port=210000e08b034ec1;
scsi-qla0-target-0=5005076300004507;
scsi-qla0-target-1=5005076300003ef6;
scsi-qla0-target-2=5005076300003497;
scsi-qla0-target-3=5005076300003c45;
scsi-qla0-target-4=5005076300004029;
scsi-qla0-target-5=5005076300002408;
scsi-qla0-target-6=5005076300003290;
scsi-qla0-target-7=5005076300002864;
scsi-qla0-target-8=50050763000035be;
scsi-qla0-target-9=1000006045161784;
scsi-qla0-target-10=100000604516234d;
scsi-qla0-target-11=5005076300003acd;
scsi-qla0-target-12=500507630000273a;
scsi-qla0-target-13=5005076300003338;
scsi-qla0-target-14=5005076300003295;
scsi-qla0-target-15=10000060451617d5;
scsi-qla0-target-16=10000060451620fe;
scsi-qla0-target-17=500507630000406c;
scsi-qla0-target-18=50050763000035d2;
scsi-qla0-target-19=50050763000041ac;
scsi-qla0-target-20=50050763000035d3;
scsi-qla0-target-21=1000006045161f40;
scsi-qla0-target-22=1000006045161784;
scsi-qla0-target-23=500507630000404f;
scsi-qla0-target-24=500507630000331b;
scsi-qla0-target-25=5005076300004038;
scsi-qla0-target-26=5005076300003812;
scsi-qla0-target-27=500502e020000159;
scsi-qla0-target-28=500502e0200000c3;
scsi-qla0-target-29=500502e02000015f;
scsi-qla0-target-30=500502e020000171;
scsi-qla0-target-31=50050763000030e3;

SCSI LUN Information:
(Id:Lun)

From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:45 2008 Subject: No subject Message-ID: Regards,
Derek R.
-- 
Junior Linux Geek
713-817-1197 (cell)
713-781-4000 x2267 (office)
"Linux users, fanatical.  No way...
HEY! Get that MCSE up on the altar,
Tux must be appeased!"
--=-x1H/azGXuumGFHORUIzY-- From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:49 2008 Subject: No subject Message-ID: > Fujitsu Builds First InfiniBand Linux Cluster > > Section: 02. Today's News > > Fujitsu Ltd's research laboratories have built the first > Linux-based supercomputer cluster using the open standard > InfiniBand interconnect. The cluster, which was built from > 16 two-way Primergy Intel-based servers, was created using > InfiniBand fabric that provided a maximum data transfer rate > of 2.5Gbps. Both Fujitsu and Intel Corp are members of the > InfiniBand Trade Association, and IBM is the other big > cheerleader for InfiniBand technology. > > Unlike many Intel-based Linux clusters, which rely on the > open source Beowulf clustering software, the Fujitsu > InfiniBand cluster uses a clustering program called SCore, > the development of which was commissioned by the Japanese > Ministry of Economy, Trade and Industry and performed by > Real World Computing Partnership (RWCP), a consortium based > in Tsukuba, Japan. The SCore clustering program does not, > like Beowulf, use the TCP/IP protocol stack for interserver > communication, but rather relies on a much leaner and meaner > SCore protocol that can communication over Ethernet, Fast > Ethernet, Gigabit Ethernet or Myrinet supercomputing links. > > Using the Myrinet interconnect, an SCore cluster with 64 > dual-processor NEC Express 5800 servers with 800MHz > Pentium-III processors and 512Mb of main memory were able to > deliver 146.9Mbps of bandwidth. Gigabit Ethernet links > delivered half of that bandwidth on the same cluster, and > triple Fast Ethernet links delivered half again that > bandwidth under the best of circumstances. > > That the InfiniBand interconnect fabric could deliver a > sustained bandwidth of 2.5Gbps on a similar SCore cluster > shows the reason why IBM, Fujitsu, Intel, Sun Microsystems > Inc, Hewlett-Packard Co, Compaq Computer Corp and other > heavy promoters of InfiniBand are looking for this > technology to transform supercomputer clusters, and any > server-to-peripheral as well as server-to-server > connections. > > The Fujitsu SCore cluster was built using pre-release > versions of InfiniBand hardware and software from Intel. > Fujitsu demonstrated the SCore Infiniband cluster running a > molecular dynamics calculation application called Amber that > was developed at the University of California in San > Francisco. The Amber application was built using Intel's > 32-bit Fortran compilers for the Pentium III processors. > > The RWCP lab has been prototyping supercomputer clusters > based on Unix workstations and Unix and Linux servers since > 1995. The lab has demonstrated SCore clustering on Sun > Sparc, Compaq Alpha, and Intel Pentium platforms. The > biggest SCore cluster it has built to date is a 512 > processor cluster comprised of 256 dual- processor NEC > Express 5800 servers with 933MHz Pentium III processors and > 512Mb of memory. This cluster, called the SCore cluster III, > ran the TurboLinux 6.1 version of Linux and the SCore 4.1 > version of the SCore clustering software. This cluster > delivered a peak performance of 955 gigaflops using a mix of > Myrinet and Fast Ethernet interconnect. From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:50 2008 Subject: No subject Message-ID: probably important. All the best db From kus at free.net Sat May 17 01:00:01 2008 From: kus at free.net (Mikhail Kuzminsky) Date: Sat May 17 01:01:50 2008 Subject: Athlon MP vs Athlon XP Message-ID: <200111052011.XAA04207@nocserv.free.net> Dear colleagues, I think about buying of Tyan S2460 motherboards for Beowulf. According the data I have, Athlon XP (Palomino core) microprocessors can work successfully w/this mobos. But there is also Athlon MP microprocessors w/same Palomino core w/same OPGA package w/same voltages and w/same frequencies beginning from 1333 (1500+). They costs, as I understand, higher than corresponding MP models. Sorry, what is the difference between MP and XP chips ? Both, if my source was correct, supports cache coherence. Yours Mikhail Kuzminsky Zelinsky Institute of Organic Chemistry Moscow From kus at free.net Sat May 17 01:00:01 2008 From: kus at free.net (Mikhail Kuzminsky) Date: Sat May 17 01:01:50 2008 Subject: Athlon MP vs Athlon XP In-Reply-To: <3C4B2812.D133809D@lnxi.com> from "Patrice Duffort" at Jan 20, 2 01:26:58 pm Message-ID: <200111061659.TAA10767@nocserv.free.net> According to Patrice Duffort > > Dear Mikhail, > > The XPs and MPs have the same core but the XP is essentially a crippled MP. Lesser overall performance. > I'm sorry, do you have some test results of performance or some data about microarchitecture difference ? It's not too obviously how to prepare chip w/decreased MP performance, but working correctly in SMP environment. For example, I should suppress something like split transactions handling etc. It looks too expensive to prepare special chip with a bit different microarchitecture. Yours Mikhail Kuzminsky Zelinsky Institute of Organic Chemistry Moscow From bogus@does.not.exist.com Sat May 17 01:00:01 2008 From: bogus@does.not.exist.com () Date: Sat May 17 01:01:51 2008 Subject: No subject Message-ID: posted here, the cache protocol allows sharing data between the CPUs via the northbridge directly. With direct access to eachother's caches MOESI is required. However, it would seem to me that to employ this high speed inter-cpu communications system, you'd need some specific code. Or am I thinking of this wrong - say for Gromacs, which uses shared memory on SMP boards (well, lam mpi does) for communication, if the shared memory area is in one CPU's cache, the other CPU can then access it directly from that cache? This would speed things up a fair bit I would expect. Does anyone have benchmarks on MP processors for Gromacs, MPQC and G98? /kc > > 3DNow! Professional is actually Intel's SSE, which Intel's compiler can > generate vectorized code to take advantage of. > > Rayson > > > --- Velocet wrote: > > What packages support MOESI and 3DNow!Professional? When will they? > > > > /kc > > -- > > Ken Chase, math@velocet.ca * Velocet Communications Inc. * > > To