Slashdot Mirror

> And how precisely are they going to save money ? And save money relative to what ? The old system ? (I kind of doubt it) The same hardware system with a proprietay OS ? Maintenance costs ?

See the links modded up to (5, informative) elsewhere. The general idea is that these days you buy a Linux CoW or Beowulf cluster instead of upgrading the ageing Cray. And for some reason it's still newsworthy, though people have been doing it for the better part of a decade now.

Not that I mind the good PR for Linux, but it is a curious phenomenon that this kind of detail of a big corporation's IT affairs is considered newsworthy.

I mean, really, what was the last 'innovation' that occured in the *nix /world?

Jeez... are you serious? Come on, Unix is one of the more important platforms for research, if not the most important. It is flexible, it is reliable, most of the scientific community is familiar with it. And these days it is also free!

Just talking about Linux I could point you to Berlin, some guys with rather interesting ideas for building user interfaces. Or the Beowulf Project, for massive distributed computing. Or RTLinux (and KURT), for full featured real-time operating systems. How about ReiserFS, that takes database-like balanced trees to the filesystem level. Or SELinux, a research prototype of a high-security operating system.

And the list goes on and on (forgive me for not looking up links, go Google for these ones): SPIN (a dynamically extensible operating system written in Modula-3, runs on Linux), all the research stuff at Mosix (including distributed shared memory, grid management, network RAM and more), the Hello Project (an operating system in Standard ML atop Linux), all the emulation stuff which hardly needs to be introduced, and all the kernel work for supporting different processor architectures.

Also note BDS's Kame Project, an advanced implementation of IPV6 and IPSec; the evolutionary scheduler for Linux; the networking kernel stuff, including the QoS work; OpenBIOS; the User-mode Linux kernel. Look up also the "C10K problem" for an interesting paper on server performance, (and while you're on that, khttpd and TUX kernel webservers).

Unix gave you the Internet, for root's sake. How much more "innovative" does it needs to get?

Just like the Iraqis and the Chinese, I do all my nuclear weapons testing on my Playstation 2 Supercomputer!

fp!

There should be a slashdot "whack-a-mole" (whack-a-troll?) game where nerds pop up instead of moles and say things like Imagine a Beowulf cluster of these, and Yeah, but can you run linux on it.

There already is. It's here. But you get at least three hammers: the "-1 Troll", "-1 Flamebait" and "-1 Overrated" hammers. You have to sign up and then watch for a few weeks, however, before you can use the hammers, or at least that's how I understand it.

Just out of curiousity, and on an only remotely related note, how many people here have ever actually seen a beowulf cluster? I think they're like nerd unicorns.

I went to the Beowulf site once but didn't inhale^H^H^H^H^H^Hdownload.

Maybe you're right. Maybe The Beowulf Cluster is geekdom's response to all the broken flying car promises of Popular Mechanics.

On second thought, no one here's seen one because it doesn't improve Quake III fps and doesn't run on Windows. :) "I promote Linux and Ogg Vorbis but use Windows and WMP for my MP3's. But I'm leet, really." <whack!>

Whenever an underpowered but inexpensive computing device is released, many Anonymous Cowards suggest that a fellow could turn a cluster of such devices into a supercomputer. Most of the time, such posters refer to the Beowulf clustering project. Exhibit A, the most common form of the joke:

Subject: Can you imagine...
Comment
...a Beowulf cluster of these?

The joke has been old for several years.

So brondsem asked: "would a beowulf cluster [of self-organizing circuits] create the Internet?" I'm still not sure what (s)he meant by that, as Al Gore created the commercialized Internet.

OSX makes a good desktop system for people who want a no-hassles UNIX system that runs out of the box. It makes it easy for people with modest computer experience to install and maintain their machine. And it runs a bunch of commercial desktop apps.

Once you are talking number crunching with a compute cluster, you are almost certainly better off with Linux-based systems. Linux has extensive cluster administration tools. Cluster installation and maintenance is much easier than OSX installation and maintenance. You can easily run any GUI-based Linux programs remotely using X11, which still beats Apple's remote desktop software in both functionality and performance. You get automatic process migration across a cluster with OpenMOSIX. There is much more numerical and scientific software available for Linux than for OSX, much of it open source; while porting to OSX usually isn't hard, it does require some effort.

Also, you do much better in terms of hardware. While XServe pricing is OK, Pentium and AMD-based servers are still cheaper and offer better performance, and they are offered by many vendors in many different configurations. And, if you like, people already use 64bit Itanium-based machines, or you can still get Alpha-based 64bit compute servers.

Maybe they use channel bonding? ;)

Can I have a Beowulf cluster of clones?

;)

I just wouldn't feel like I am part of the /. community if I didn't mention this at least once:

Can you imagine a beowulf cluster of these?

You can find out what it's all about here: http://www.beowulf.org/intro.html

1368 wasted CPU's

Are you kidding? With the possible exception of Cray and maybe the Hitachi(can't find info on that one... they seem to be out of the supercomputer business), nobody builds single-unit supercomputers any more. The scalability with clustering and shared memory over high speed networking overcomes the contention problems with massively parallel processors, though the Numa-Q may let us put more cpus in a single box. Check out the AlphaServer SC, the RS/6000 SP, both supercomputers in the top 10 (with the SP dominating... awesome switch for the shared mem), and the Beowulf, or (no shared memory here), the Mosix, clustering. If I were building a linux supercomputer, I think i'd rather go with a pile of dual-cpu units anyway, cutting down on resource contention.

With distributed processing like these systems, adding another unit adds about another unit worth of processing power, whereas adding cpus to a SMP configuration gives diminishing returns. As long as you build the infrastructure along with the growth (add switch capacity), the sky's the limit.

> A beowulf cluster is not limited to Linux, it could run on top of any OS. I believe NASA did the original design work to be OS agnostic.

Strictly speaking, Beowulf is a Linux-specific phenomenon, though there's no reason "Beowulf-like" clusters could not be built on other OSes. (However, you may need the OS's source code if you want a true cluster rather than a cluster emulator that runs in application space.)

Riiiight... a Google search for fluorinert "mustard gas" turns up a total of 10 hits, the ones that mention Fluorinert turning into mustard gas are from some thread on Beowulf clusters. And that claim was only made by one person... so I'm supposed to believe it?

Well, I don't... Mustard gas contains sulfur and chlorine... in particular, it's C4H8Cl2S. On the other hand, Fluorinert is a fluorocarbon--it contains carbon, fluorine, and maybe hydrogen. There's no way you're going to get mustard gas from burning it (in air at least). In fact, Fluorinert is non-flammable. Good luck setting it on fire in the first place!

That said, it does decompose into hydrofluoric acid and other stuff if you heat it too much, so I guess that would be a bad thing :) But it's not mustard gas.

Here's the MSDS for FC-71: http://www.sisweb.com/referenc/articles/fc71.htm and FC-5311: http://www.sisweb.com/referenc/articles/fc5311.htm

Thanks for the FUD, Bob Glamm!

I suggest the Feds put the database on Oracle running on a Beowulf cluster of Linux machines, running the newest kernel.

Then Slashdot would LOVE it.

;)

Err, it was written by physicists...

Will the 10 GHz CPU outperform the current 1GHz CPU? Why?

Today's limitations seem to be the hard disk with access in the millisecond range (~MHz). A few orders of magnitude up is main memory speed. Even Rambus runs at less than 1GHz clock speed and doesn't get much gain for that! Please look up the following: what length is a light-nanosecond. Memory systems must be less than this in physical dimensions, or the increase in clock speed is useless. But modern computer problems require really large capacity memories.

The conclusion is: a 10 GHz CPU will be useless without a comparable speed for memory and an almost as fast speed for hard disk, all currently impossible. Nobody will buy a 10GHz CPU with a 1GHz memory speed and even slower hard disk!

P.S. Here is a hint to the problem. The really big super-computer boys have concluded that the solution is a move to massively parallel machines, nothing else else will do. This means really big $$$. Home computers just don't count in this world. Intel is just trying to fake you out!

Imagine a Beowulf cluster of these!

Seriously though, it is easy to take a good parallel code and murder it by going for a Beowulf option. I assume that your CFD code is a heavilly parallel beast. From what I have seen from the links available here one must be very careful about the relative importance of the node communication bandwidth and the node latency on the code in question.

In my opinion this decision boils down to the following:

How interdependent is the parallel code? If the code is trivially parallel, as many of mine are, then a cluster of the cheapest ix86 machines running Linux is an excellent bet.

Are you certain that the code will work on the platforms in question? Perhaps invest a couple of kiloquid (thousand pounds) or so in a few ix86 machines for a feasibility study, if it doesn't then you have cut your losses and have a few machines to play Quake on ;-). For the ultra cheap option, if your Prof is unhappy about gambing any dosh (money) on buying new stuff, stick Linux on the NT machines and try those.

How far through your PhD are you? Quite seriously, if you're anything like me you may find that you end up spending most of your time tweaking computers rather than doing physics. Building your own supercomputer could easilly suck your entire PhD down the time toilet.

I expect you have done this already but ask the people at Fluent Inc. what experience they have on parallelizing their codes.

Finally, I would say while the support you may get from commercial *NIX vendors is generally good it can also be very expensive. While this is not necessarilly a problem in a commercial datacentre environment, in academia it can be crippling. If you and your co-workers are willing to expend time and effort into trying to get the code working on a Beowulf style cluster it could well be a good long term investment.

Good luck!

P.S. When I checked this post, www.beowulf.org was down so you may want to check the google cache.

Look here.

The xbox acts as a cooker itself! no need to buy a microwave to warm up that pizza, just put it on the xbox!
Is the x-box big enough for this? I doubt it since this was meant as a family machine. That means you need BIG pizza's. The xbox is only atx size

Expensive heating is now cut down thanks to the myraid of heat exhauts on the Xbox
A Beowulf cluster of ... would solve this problem much faster.

Use the controller as an inexpensive door stop: big enough for even the heaviest of doors.
Or better, use your Dreamcast, it is is real cheap now.

coming soon: a hack to reverse the exhauts so the xbox doubles as a vacum!
Silly, why do you want to heat your dust?

just my 2 cents

Clustering is such a fascinating area on its own and the article is so shallow that I'm curious why they've published it at all. They could mention the potential benefit of cluster computing as well as examples of some working clusters like Beowulf or Mosix or even the famous fact that there is a cluster among top 500 supercomputers.

First you could got here
And then here

How hard can it be, it's not like you're launching rockets...Wait a second, 2 days to buy, quick access to cash, damn he probably is launching rockets!

If you don't find any answers to your quest then you could always buy 10 dual-processor machines, configure one and then copy its HD image to the other 9 ( I have never tried this ).

I wonder whether /. uses any existing cluster like Beowulf or Mosix?

Beowulf Clusters are not that hard to build. the only differnce between building an 18 node cluster and a 4 node cluster is the number of computers. If you have the hardware, some free time, and the patience, then setting one up is not that hard. The hard part is actually doing something with one. Programming in MPI and PVM is not an easy task even with a degree in CS (unless you have no life). Check out Beowulf.org for more info.

Cluster: A computer consisting of many smaller computers. A cluster acts like ONE computer.

Network: A group of computers connected together for data communication, not necessarily acting like one machine.

Go to Beowulf.org for more information.

Excuse me, but haven't they considered Beowulf clusters? I think they are better in both scalability and price. Even some clusters managed to rank among 100 fastest computers.

You obviously haven't looked very hard for information :)

SETI@home works in client-server fashion: your desktop computer asks the main server for a chunk of data, then chews on the data and talks to the server again. This is massively parallel computation, but it isn't peer-to-peer, it's client/server.

When you put data on this fiber ring, within a very short time all the computers on the ring have seen the data. So if you want a bunch of computers to cooperate on a job, this would be a great way for them to update each other on what they are doing. If you did it right, you would have massively parallel distributed processing: all the computers in Canada tied into a single InterComputer. And just as Napster can spread popular songs around where a single FTP server would be hammered, an InterComputer potentially could handle truly large computations that any single computer (or even Beowulf cluster) couldn't.

Multicast data packets aren't new; that's why they said it takes only a few changes to try out their ideas. Multicast packets are currently designed to die fairly quickly so they can't clog a network up too much; these guys want the packets to go all the way around the ring.

P.S. That joke about the backhoe chopping the fiber was only a little bit funny, and then only the first time. When a backhoe hits a cable today, half of Canada does not lose Internet service! It isn't a trivial ring; it has some redundancy redundancy.

steveha

Yeah, tried that once with a video card, with the same result. Don't understand why people think they need to continually add components to individual PC's when additional nodes to the cluster are so easy to plug in. Of course, you might lose 16 machines for 5 seconds on occasion when you need to daisy-chain another hub. >:)


Linux rocks!!! www.dedserius.com

Check this out:

How to pronounce Linux
How to pronounce Linux
A poll from Linux.com

Funny that this guy is so confused that he is mistakenly informing people to pronounce it the WRONG way. :)

Duh. You just shit in your own nest.

It is LIN-ix. When I lived in the dorms, we had one guy that always called it the incorrect way, LYE-nix. When we finally had the heart to tell him, he killed himself.

The moral of the story is, pronounce Linux as LIN-ix, or you may shoot yourself one day.

It's a way to build a supercomputer by networking lots of standard computers together. For example, if you had 10 Pentium3s you could link them and get 10x the power of a Pentium3 (on certain special tasks). See beowulf.org for details. Beowulfs are used when a supercomputer is needed on a low budget, and can rival conventional supercomputers at some tasks.

Saddam has obviously been using a pseudonym while trolling for information! He's probably been using us for years, asking about the suitability of various non-computer computers as components of a beowulf cluster!

The bastard!

Assuming the Beowulf thing is supposed to refer to the parallel computing project, and not to some guy, the link should point to: http://www.beowulf.org/ and not http://www.beowulf.com/.

But fortunately the some-guy has linked to the project too. Thank you, some-guy.

Having discussed extensively design possibilities with the big walker operator and worked a bit on a Walking robot repairs with him while with SRL (www.srl.org), I can attest that the Honda robot is definately an impressive achievement. Stable walking is difficult enough on tripedal and quadropedal robots, and the hacking done at SRL only succeeded in 2 and 3 legged robots, with not much hill-climbing ability, and only moderate speeds. The feedback circuitry and balancing techniques needed for a biped like the Honda robot is, sadly, not yet within reach of even the sophisticated hackers without a ton of money.

The Honda robot qualifies as an "impressive start"... its 2.0km/h speed of the Honda robot is not impressive for a lightweight tri or quad walker, but it is for a bipedal robot of human size.
The fact that it can climb stairs is especially cool, given the extensive rebalancing done every moment in a step, and the feedback sensors needed to read these motions properly.

Certainly there are lifting robots which can hoist many tons, so the 5.0kg/hand weight limit seems skimpy - but not when considering that this machine can allegedly walk and perform complex arm manipulations while holding this weight. Sadly, its continuous runtime before recharge is only 30 minutes, but I suspect later versions will take advantage of increasing innovation in charge/weight ratios in batteries, and perhaps solar panels for space use (an obvious application of these robots would be EVAs for the ISS or other craft).

Cool, in terms of integration with other systems, is the use of wireless ethernet as the comm standard, rather than some proprietary system. This means this robot could be controlled by base-station systems of arbitrary complexity - including a Beowulf cluster running a complex AI system like Webmind. This means that while technology is not quite there yet to put any advanced computational intelligence inside a biped robot, it can be controlled by advanced systems running at fixed-position stations through LAN technology - a good compromise in terms of merging the state-of-the-art in Robotics with AI to try to build towards a better convergence.

Regarding the robustness, it appears that the 25 minutes of runtime is the primary limitation in terms of continuous operation - there is no data I was able to find on failure rates or the fault tolerance of the sensors or computational systems on-board.

As for hacks vs. new general purpose algorithsm... They obviously do not reveal tremendous amounts of details, but suffice it to say that the engineering done to build 3 successive models of bipedal robots that can walk and climb (stairs, hills) represents fundamental work in robot dynamics engineering which, while parameter tweaked for this robot's operations, is certainly applicable (with some tweaks or modifications, as with all engineering techniques) to other bipedal robot applications.

The wireless lan comm technology, improved user interface (over the previous version), and sensor systems are all also certainly reusable in similar robots (indeed, likely also in multi-legged robots).

However, as it is a commercial product AND I do not read Japanese, I was not able to find any papers on specific algorithms to give a more detailed analysis...

Here are some useful resources I did find:

The official site in English

An article about the robot's walking functions

Images of the robot at UIUC

Biped Robots in General

Robodex Robotics Conference

Whut?

[snip]...a real aversion to intelligent discussion of issues.

Now hold on just one second! Personally, I have found many a good conversation going on here at /.
Why, just the other day I had sat down at my Beowulf cluster with a nice toasty pants full of hot grits to discuss relativity with Natalie Portman.
Granted, it can be difficult to hold a conversation with a waif he is not only emaciated and vapid, but petrified, but I tell you, we were solving the mysteries of the UNIVERSE, I say!

/me sits back to watch all my karma fly away...

I think the last version of Extreme Linux was (searches for his Extreme Linux CD) is based on RedHat Linux 5.0 - it's a little out of date now - code has moved on considerably.

For you I would like to recommend some reading:

Building Linux Clusters by David HM Spector published by O'Reilly, (hmmm site seems to be down, come back later, or check Google cached version)

This book comes with a CD together with clustering software. It also comes with step-by-step instructions. I believe, however, that there are some errata, which means that some hacking will need to be done to get your cluster online.
It also goes through some aspects of choosing hardware etc...

A more in-depth resource, without step-by-step instructions, but with in-depth discussions on granularity of Beowulf systems and whether they are actually good for the tasks you have in hand is:
How to Build a Beowulf, A guide to the implementation and application of PC Clusters by the MIT Press

Also check the The Beowulf Project Site and the The Beowulf Underground Site

Have fun!
---

*Cough* Who is this?

AFAIK, it _is_ named after the mythical dude that killed Grendel. Even if it isn't, it's still a rather cool name.

Hey, even my .sig is ontopic!

> Is anybody actually using Beowulf in a non-academic, non-experimental system?

Visit www.beowulf.org.

Yes, there is a strong correlation between "Beowulf" and "academic", but it's not 1.0. And there's an even stronger correlation with "experimental", since massive number crunching is so important to modelling.

But if you skim the list at the link, you'll see that a lot of the applications have commercial potential, even if (mostly) currently used in research: applications in chemisty, meteorology, graphics, economics, satellite imagery, materials design, etc.

Also, a googling turns up this older article claiming that Beowulfs were in use (or at least under study) at Boeing, Bristol Myers, and Proctor & Gamble. I wouldn't be surprised to find that a lot of companies were beowulfing, since most companies don't make a big PR noise about their IT infrastructure, and might even prefer to keep it secret as if it were a competitive advantage.

Finally, I vaguely remember an older /. article talking about a company that uses a 1000-node Beowulf running genetic algorithms to produce patentable software products.

--

And about your second paragraph:

1. Linux might just wind up feeding some of the world's hungry. I could see where extremely low-cost computing could certainly help Third-World countries. 100% of the web servers in .td, .ne, .lr., gq, .cf, and .dj run Linux (source: Internet Operating System Counter). Even superficially speaking, that has to create some jobs.

   You can't tell me that countries with lots of manpower and very little money wouldn't benefit from what amounts to essentially free computing power. If nothing else, Linux can help countries like Niger and Djibouti keep up in the technology game. You think they'd have that chance if they had to pay per seat to get Windows servers up? They'd have a much, much worse chance wihtout Linux around.

2. Linux is helping to sure disease. I happen to know for a fact that UCLA (among many other places) runs a rather large Linux cluster for doing things like chemical analysis for medical research. They do way more with Linux than they could ever do without Linux. Big hardware isn't cheap, and there's only so much grant money to go around -- a Linux RAIC-ish system stretches a budget to the point that the previously impossible becomes possible.

   Take a look at the Linux Medical Research HOWTO and then tell me Linux isn't helping cure disease. I'm sure that someone else here can give you first-hand experience on what Linux is doing in the medical community.

3. Linux isn't going to put a man on Mars?!? Now I know you're joking. Do you know how much Linux junk NASA runs? Hell, the driver for the NIC in every Linux box I have was written by a guy at NASA (it's the Tulip driver). And there's always Beowulf, but that probably won't do anything to help get a man on Mars.

I'm not saying Linux is a panacea, and anyone that does so is a certifiable moron who has no idea what (s)he's talking about (at least they'll be easy to identify and avoid). But to say that Linux isn't helping make some really wonderful things happen is extremely short-sighted. In fact, I'm still not convinced that you weren't joking. If I hadn't replied in this thread, I probably would have moderated it up as "funny".

-B

While I agree that vi doesn't use many system resources, I would argue that it doesn't because it's because these programs are more efficient.

--

And to JDax, since I didn't get a chance to reply last night: Hell, I wasn't going to blame Linux for bad weather forecasting, I was going to blame it for the bad weather itself! Global warming, the recent spate of droughts, floods, and natural disasters: all can be traced back to Torvalds and Cox. It's true! :)

And I was gonna point out NASA's Beowulf cluster but then you'd say... ohhhhh... &nbsp fsck it! &nbsp ;-)

Actually want to make a point that hasn't been touched here yet and that is this: &nbsp Remember learning in U.S. History about what happened, oh... &nbsp around 1849 in San Fran? &nbsp A little thing called the "gold rush"? &nbsp Looks like we're seeing a repeat in history here some 150 years later...

And for those who are new to that area and/or are considering the latest dot com gold rush, remember that the last sortof big earthquake occurred when? In 1989? &nbsp And they're looonnngggg overdue. &nbsp I remember watching the World Series when Candlestick Park (OOPPPSSS!!! &nbsp WRONG NAME! &nbsp Ahem. &nbsp "3COM Park", I believe it was later called) was literally being shaken apart. &nbsp If you like San Andreas fault living, be my guest. &nbsp Zico and me will stick with the Linux-caused hurricaines and blizzards and tornadoes on the east coast!

And by the way, regarding the real estate there? &nbsp ALL that stuff has to go - the Presidio included. &nbsp I mean come on. &nbsp How the hell can they start building Star Fleet Academy with those damn dot coms in the way, huh? &nbsp Let's start moving to the REAL technology!

;-)

Is anyone out there actively lobbying the government to officially endorse OSS solutions rather than proprietary software? It seems to me as a taxpayer that I would like to see the vast number of government projects out there actively evaluate Linux as well as Solaris and NT as platforms.

Alot of us are trying, believe me. &nbsp Most of us have OSS snuck in the background but contributing none the less... &nbsp The more "public" uses are seen at NASA with their Beowulf cluster and at NIST.

There's a FOSE conference coming up next month and one of the seminar tracks includes a session on Linux. &nbsp However I wish that more on OSS was scheduled to be discussed at this. &nbsp The timing of FOSE is really good considering all that's going on right now, but seems no one in the D.C./MD/VA area appeared to push it for this year's conference...

Recently I set out to finally install RedHat. However, I could not get the Cogent eMaster 110 tx nic to work. The installer correctly identified it as tulip compliant, but something was wrong with autosensing between 10 and 100. I read the relevant ethernet howtos, but there was no clue. Then I discovered the tulip driver page with a grid of options to try, but none worked. Finally, I found posts on tux.org and beowulf.org from people who were having the same problem, but there weren't any follow-up posts from the Linux community. I tried asking for some advice on #linuxhelp and #redhat, but people mostly ignored me. This is the crux of the problem; how does the Linux community expect their OS to be widely adopted by consumers when hardware drivers are still so buggy, no one will help you diagnose or solve the problem even when you say pretty-please, and you're forced to buy a new nic out of desperation? I'm a reasonably skilled computer nerd, so I didn't panic when the install repeatedly didn't work. However, I can easily imagine a less skilled person encountering this aggravating problem and just giving up. It's time for the Linux community to stop ignoring and antagonizing potential "newbies" or else your OS will fade into sysadmin oblivion once again.

How about a Beowulf cluster of these? :-)

The Acard Ultra ATA-66 PCI PnP Adapter (AEC6260) has on-board upgradable flash BIOS. Would it be possible to do the same with this card?

I wonder how the guy who did this first with the Promise card figured it out? Serendipity?

Cheap is good.

There is a mailing list full of people who know what they're talking about with regard to this topic. Subscribe with info from www.beowulf.org for these sorts of questions. Don't trust these comments because they're mostly from clueless slashdot users who misuse the Beowulf name and have no idea about any of this. Hell, somebody is posting links to the extreme linux CD... do NOT buy the Extreme Linux CD if you are trying to get a real cluster together - it is an outdated POS

Oh-my-gawd even the Germans have one.

Oh-my-gawd even the Germans have one.

Beowulf is a linux clustering system which allows you to use stock PCs in parallel to get power equivilent to a supercomputer.

Check out http://www.beowulf.org/

~~~~~~~~~
auntfloyd