Slashdot Mirror


Coolest Cluster Ever

sw155kn1f3 writes "Scientists at Los Alamos National Laboratory built a cheap (less than $1k per unit) 294-unit Beowulf claster dedicated to run astrophysics calculations. According to their website it's 85th fastest computer in the world. Seems cool and promising as it made with cheap components and off the shelf hardware."

85 of 279 comments (clear)

  1. Imagine... by Anonymous Coward · · Score: 5, Funny

    Imagine a single unit of these...

  2. Obligatory by BSDevil · · Score: 4, Funny

    Imagine a Beowulf cluster of these...Beo...shit - I knew this joke would have to end at some point.

    --
    Cue The Sun...
    1. Re:Obligatory by paganizer · · Score: 3, Funny

      Oddly enough, I was just going to make the exact same post. Thank you for sparing me. Freenet IS the hope of the future

      --
      Why, yes, I AM a Pagan Libertarian.
  3. Uh oh by Eric(b0mb)Dennis · · Score: 3, Insightful

    Astrophysics.... 85th largest in the world... Call the department of homeland security, this could be terrorism! Yeah, okay.. it was stupid :P

    --
    Excuse me, I don't mean to impose, but I am the ocean
    1. Re:Uh oh by MoThugz · · Score: 3, Funny

      Yeah, it's stupid... as you can see from the article itself, everyone involved in that project has a @*.gov email.

      Sheesh...

    2. Re:Uh oh by Decimal · · Score: 2, Funny

      Astrophysics.... 85th largest in the world... Call the department of homeland security, this could be terrorism!

      We should actually be worried about the eithty-fourth fastest computer in the world... Saddam Hussein has built it out of a warehouse of imported PlayStation 2s.

      --

      Remember "Bring 'em on"? *sigh
  4. And they used Shuttle XPC SS51Gs! by Max+Romantschuk · · Score: 5, Informative

    They didn't even use a rack mount solution, they used regular Shuttle XPC SS51G Mini-PCs

    I thought Shuttles Mini-PCs were cool before but this really resets the scale... Now where is the HOWTO for this thing? ;)

    --
    .: Max Romantschuk :: http://max.romantschuk.fi/
    1. Re:And they used Shuttle XPC SS51Gs! by ryochiji · · Score: 5, Funny
      > They didn't even use a rack mount solution

      I was thinking that too when I saw the picture. I mean, how are they securing those boxes? The way it looks, I certainly wouldn't want to be standing near that rack when an earthquake hits...

      But then, imagine: Cause of death: crushed by Beowulf cluster. That's a geek's dream come true!

    2. Re:And they used Shuttle XPC SS51Gs! by kmellis · · Score: 2
      " But then, imagine: Cause of death: crushed by Beowulf cluster. That's a geek's dream come true!" - ryochiji
      Yeah, the kind of dream you wake from out of breath and with your heart racing. I, for one, am not yearning for a violent, ignoble death. But for those of you that are, it's probably easy to arrange.
    3. Re:And they used Shuttle XPC SS51Gs! by Tet · · Score: 2
      They didn't even use a rack mount solution, they used regular Shuttle XPC SS51G Mini-PCs

      The question has to be... why? I've never understood why so many people build clusters with essentially desktop PCs. Haven't these people ever heard of the 1U rack mount case? Yes, it's a slightly more expensive inital outlay. But surely the cost savings in terms of floorspace and power/cooling for such a large volume would outweigh that in no time flat? Plus the Shuttle, like most other desktop cases, don't have the option of hot swappable drives. With the number of machines in this sort of cluster, drive failure is a major problem. The ability to just pull the failed drive out of the front to replace it would be a huge win.

      --
      "The invisible and the non-existent look very much alike." -- Delos B. McKown
    4. Re:And they used Shuttle XPC SS51Gs! by Boiling_point_ · · Score: 2
      they used regular Shuttle XPC SS51G [shuttleonline.com] Mini-PCs
      Well, duh!! Of course they used shuttles - it's for a space simulation!
      http://space-simulator.lanl.gov/ ;-)
      --
      "If you create user accounts, by default, they will have an account type of Administrator with no password." KB Q293834
    5. Re:And they used Shuttle XPC SS51Gs! by aaarrrgggh · · Score: 2

      Actually, you should be careful about the benefits of making the servers more and more dense! 1U servers are a pain to deal with for cables, and keeping the things cool. If you have the space, there is almost no real benefit to fitting the equipment into as tight a footprint as possible.

  5. Um, how does that make it the 'coolest'? by autopr0n · · Score: 5, Insightful

    I don't really think you can really quantify coolness in general, but I fail to see how the fact that this thing is cheap makes it all that cool.

    Perhaps if it was going to run simulations of ultra-low temperature physics... get it? haha. I kill me.

    --
    autopr0n is like, down and stuff.
  6. For less than $300,000... by TWX_the_Linux_Zealot · · Score: 4, Interesting

    It would be possible for a group of people, not necessarily a small group, but not necessarily huge, either to repeat this. 100 people, each with $3,000, could do it. The group would need to find some space to house the thing, and would probably have to do it in a climate where it could be relatively naturally cooled, which definitely rules out Phoenix. The computer would then be one of the fastest machines in the world.

    Granted, I don't know what the hell they'd do with the computer, but it would be kind of cool to be on the list.

    --

    IBM had PL/1, with syntax worse than JOSS,
    And everywhere the language went, it was a total loss...
    1. Re:For less than $300,000... by buswolley · · Score: 2, Insightful
      actually. if these people are not after money, then this would be a great computer to work out simulations for social dynamics, alternatice economic system theory, etc.

      The point being, if you are creaticve, you could find quite a few interesting things to do.

      --

      A Good Troll is better than a Bad Human.

    2. Re:For less than $300,000... by Cheese+Cracker · · Score: 2

      "It would be possible for a group of people, not necessarily a small group, but not necessarily huge, either to repeat this. 100 people, each with $3,000, could do it. The group would need to find some space to house the thing, and would probably have to do it in a climate where it could be relatively naturally cooled, which definitely rules out Phoenix."

      Who'd pay the insurance and electric bills?

      "Granted, I don't know what the hell they'd do with the computer..."

      They'd spend another amount of cash on a OC-48 connection to the internet and then offer webhosting to website owners who think they might be next in line of getting linked in a story at slashdot.com ...

    3. Re:For less than $300,000... by jericho4.0 · · Score: 2, Interesting
      Good point. All the supercomputers in the world today are being used by big money programes. I can't wait until this is on my desktop (or on a network I can use).

      There are thousands of complex systems that could be modeled that might provide fasinating insights. What happens when you run the Game of Life for 10^100 generations? How about compiling the Linux kernel in a genetic algorithim? Who knows?

      --
      "A language that doesn't affect the way you think about programming, is not worth knowing" - Alan Perlis
    4. Re:For less than $300,000... by Captain+Large+Face · · Score: 2

      Granted, I don't know what the hell they'd do with the computer, but it would be kind of cool to be on the list.

      Duh! Doom III of course!

  7. Re:Wow by Max+Romantschuk · · Score: 5, Informative

    and a "heat pipe" instead of a fan!

    Actually, the heat pipe doesn't replace the fan, it just lowers the number of fans used in the system, since the case and processor fan can be combined.

    Tom's Hardware has a review of one of these things (not the same model though)... have a look.

    --
    .: Max Romantschuk :: http://max.romantschuk.fi/
  8. Re:Not that impressive by tgrotvedt · · Score: 3, Interesting
    C'mon....

    For $294,000US, this has to impress you.

    All of the top 10 fastest computers in the world are multi-multi-million dollar machines. This is a breakthrough because it represents another milestone in bringing supercomputing accessible.

    $264k one day, $100k the next. I sincerely hope that soon small-to-medium enterprises can own supercomputers. With all the low budget physics stuff going on at Universities around the world, cheap supercomputing can only be a good thing.

    --
    What makes a man want to be a mouse? (Python's Flying Circus)
  9. Shuttle by T-Kir · · Score: 2

    Now if there was ever an advert for Shuttle XPC systems, the image does that just nicely.

    If I only had 1 Shuttle XPC, that would be great. I suppose Shuttle ought to add this site to their news section. Hopefully their web server runs off these systems and a fat internet pipe... just to test the /. load.

    --
    Are you local? There's nothing for you here!
  10. 1000 machine beowulf cluster by SystematicPsycho · · Score: 4, Interesting

    Take a look at this room A 1000-Pentium Beowulf-Style Cluster Computer half way down the page.

    --
    Analytic & algebraic topology of locally Euclidean meterization of infinitely differentiable Riemmanian manifold
  11. Perhaps... by wd123 · · Score: 2, Funny

    Instead of astrophysics work they should use it to find all their radioactive trees.

    --
    "question = (to) ? be : !be;" --Shakespeare
  12. Re:I doubt this thing will run for long... by Roosey · · Score: 2

    (create much less heat than bulky p4's! I'm sure that lab is hotter than an african rainforest)

    So much for it being the coolest cluster ever.

  13. Whoah.. by Anonvmous+Coward · · Score: 3, Funny

    "According to their website it's 85th fastest computer in the world. Seems cool and promising as it made with cheap components and off the shelf hardware."

    I'm guessing the story submitters ran out of anti-MS ammo tonight. Heh.

    It's satire, laugh.

    1. Re:Whoah.. by enneff · · Score: 4, Funny

      I thought satire was supposed to be funny.

    2. Re:Whoah.. by Anonvmous+Coward · · Score: 2, Funny

      "I thought satire was supposed to be funny."

      That misperception disappeared when SNL inexplicably rose to popularity.

    3. Re:Whoah.. by Anonvmous+Coward · · Score: 2
      "You mean 'misconception'."

      No, I meant misperception:

      misperceive P Pronunciation Key (mspr-sv)
      tr.v. misperceived, misperceiving, misperceives

      To perceive incorrectly; misunderstand


      I had no intention of implying that he had ever concieved.
  14. Reminds me of.... by Drakonite · · Score: 2, Insightful
    Reminds me of when a computer company did the same thing... I think it was HP but I could be wrong.

    Although if I recall correctly they ended up quite a bit higher on list.

    --
    Shoot Pixels, Not People!
  15. Re:Bull by CanadaDave · · Score: 2

    Shit, am I on crack? They never claimed to have the cheapest Beowulf cluster... I'm sorry, I really must have been on crack. I read the website then came back to ./, started to write, got distracted with something else, and then must have forgotten the exact details. Ignore post, and mod down, please.

  16. Like all government news... by dethl · · Score: 3, Interesting

    Last modified: November 15, 2002

    We are the last to know about it :P

    --
    "Some fight for law. Some fight for justice. What will you fight for? One day, you will see."
  17. Hmm... by lvdrproject · · Score: 5, Funny
    1. Create Beowulf cluster
    2. ???
    3. SLASHDOT!

    Laugh, it's a joke.

    :Lav

  18. cheap clusters can also be bad by Stanley+Feinbaum · · Score: 2, Funny

    As much as I am a fan of cheap 'beowulf' custers, there is a certain issue which troubles me. In this case, the cluster in question cost less than 300,000$, a healthy sum, but much less than a large cray or sgi server would cost. Such clusters can be used for an array of activities, such as nucular bomb tests (one of the driving forces in pushing supercomputer technology is nucular weapons), or cracking encryption.

    Supercomputers are controlled by USA export laws, but powerful beowulf clusters can be made by anyone with a reasonable amount of money and knowlage. Since the software is free and of the shelf components can be used... wouldn't it be possible for terrorists to use open-source software to create their own supercomputers to test nucular weapons, crack American law and millitary encryption, ect... ?

    I believe this 'beowulf' techonology, as great as it is, could be possible dangerous to American interests. It is my hope that this software will soon be controlled by the American millitary and not be spread for free because I fear for the safety of my family and country (bless them both) if terrorists have access to supercomputing technology.

    --

    Stanley Feinbaum, professional journalist and master debater! God bless the USA!

    1. Re:cheap clusters can also be bad by Moses+Lawn · · Score: 3, Funny

      Ladies and gentlemen, we have a replacement for Jon Katz...

      --

      What if life is just a side effect of some other process and God has no idea we exist?

    2. Re:cheap clusters can also be bad by karlm · · Score: 2
      Are you a former lawmaker, or otherwise mentally deficient?

      • The bag is here, the cat is long gone. Taking beauwulf source code off the net is like taking pee out of a pool. (blatantly stolen from News Radio)
      • The physics behind nuclear weapons is well understood and widely published. It took a small town full of geniuses to figure it out the first time, (and determine that it's even possible) but no longer. They also designed a Uranium gun-type weapon and a Plutonium implosion-type weapon. The latter is muchmuch more complicated. Designing a gun-type Uranium weapon is trivial, and anly slightlymore difficult if you're restricted to pen, paper, and sliderule. (Granted I'm an MIT student, so I'm probably overestimating the amount of brain power available to a large terroist organization.) The difficulty lies in the tremendous amount of effort/electrical energy/sapce/time/money required to perform isotope seperation/smuggling without getting caught before you have the chance to detonate the weapon.
      • Cracking bulk ciphers is basically infinately parallelizable and does not require much communication bandwitdth at all. Beauwulf clusters don't gain you much. You'd get better performance running your cracking programs over V2OS, FreeDOS, or some other ultra-lightweight OS.
      • Bin Laden et al. are very smart. Sure they have some grandios long-term plans about getting the US out of Saudi Arabia, creatig an Islamic super-state, and acquiring a fission bomb. However, they are very smart and efficient. They use low-infrastructure, reliable attacks for the most part. Planes filled with fuel, boats filled with conventional explosives, and dirty bombs are all very simple and can be potentially acquired by a single autonomous cell. I'd be much more worried about poison gas attacks and coordinated massively distributed firebomb attacks. These are the kind of low-tech high-casualty attacks at thier disposal right now.
      --
      Copyright Violation:"theft, piracy"::Anti-Trust Violation:"thermonuclear price terrorism"<-Overly dramatic language.
  19. Or.... by raehl · · Score: 5, Funny

    They could spend their $3k on something actually needed by 100 people thinking about spending $3k to share a spot on a fastest computers list.

    Like a prosititute.

    You go to your high school reunion, what's, more impressive, the "Hugh Hefner" 100 $3k prostitutes that come with you, or the "Bill Gates" story about the 300 1k computers in your mom's basement?

  20. Re:I doubt this thing will run for long... by The+Wooden+Badger · · Score: 2, Funny
    I'm sure that lab is hotter than an african rainforest

    But it's a dry heat.

    --
    Heroscape, it's like legos combined with anachronistic wargames.
  21. Re:you have got to be kidding me by F2F · · Score: 2

    he's one of the backseat drivers -- never so much as gotten close to a supercomputer and he thinks he knows how to make one...

  22. Re:I doubt this thing will run for long... by Cryptnotic · · Score: 2, Insightful

    I do not believe in trolling, if you are a troll please do not respond to my posts.

    How can you have that as your sig when your entire post was one big troll against Intel, Maxtor, and DDR SDRAM? It's not ironic, it's hypocritical.

    --
    My other first post is car post.
  23. 85th Fastest in the World? by serutan · · Score: 4, Funny

    The photo alone is worth surfing over to the article. As Socrates once said, "what a rack!"

    But now that they've got the 85th fastest computer, what will they have to do to maintain that coveted position? I imagine the people who are running 86th are rushing out to buy more nodes. My own computer is the world's 27,385,422nd fastest, and I'm battling like crazy to get to 27,385,421. :-)

    1. Re:85th Fastest in the World? by The+Wooden+Badger · · Score: 3, Funny
      My own computer is the world's 27,385,422nd fastest, and I'm battling like crazy to get to 27,385,421.

      I'm not going down without a fight.

      --
      Heroscape, it's like legos combined with anachronistic wargames.
    2. Re:85th Fastest in the World? by Gumber · · Score: 2

      As Socrates once said, "what a rack!"

      Socrates? Hmmmm.

  24. The *real* reason for building this by GimmeFuel · · Score: 2, Funny

    Astrophysics? Pfft! How bout a LAN party? Does 10 billion fps on UT2003 sound good?

  25. You have to be kidding by thirty6hex · · Score: 2, Insightful

    Never forget that the most important part of a Beowulf cluster is that it relies on no single hunk of metal in its operation. A cluster is intended to break gracefully. A good RAID-5 solution is the same way, you expect at some point to lose a disk; thus the reason that RAID stands for Redundant Array of Inexpensive Disks.

    Also don't forget that you've got 3 years warranty on those Maxtors, and you can just reload the OS on the bare drive from a copy of one of the other 280 someodd. Sure they suck, but are you really ever even using them? I bet they just got a special on a whole box of em and 80's were all there were.

    RAM is exactly where these computers need in large, fast quantities. RDRAM, while arguably faster, is a money sucking wench, and 333 is just perfectly fine if you actually do your homework and buy the right chips. Sure, if you go and pick up the deal of the century at the lowest priced vendor online then you can expect to get some odd results. But if you are buying 300 gigs of the stuff you can get a pretty sweet deal out of a reputable manufacturer and get the nice chips to boot. And don't forget that those come with a warranty too; so you just send the dicey ones back....who cares if you lose a few boxes for a day or two?

    It's the coolest because it puts off less heat than most, using the head-pipe feature off the cpus. Run a big HVAC and hook it up to those pipes, and all of a sudden you have A/C cooling directly on your chips, and its more quiet to boot.

    This thing is built out of parts that are in *your* computer. It's built from the parts that are moving the fastest thru the vendors. Every single part of this cluster could be purchased in lot quantities at a very reduced cost due to slowdowns in the last 9 months.

    Not including the network backbone, you can build the very computer they are using for much less than a grand per node and have it rate; I think that was the point and I think that they made it.

  26. Re:Why build one anymore? by Cryptnotic · · Score: 3, Interesting

    Not to mention national security issues. LANL has had many problems with security in the last few years.

    --
    My other first post is car post.
  27. Nobody worded it the way I wanted, so... by Myco · · Score: 2

    Imagine one of these!

  28. Think again - EDA tools prove this wrong by StandardCell · · Score: 3, Interesting

    At the semiconductor company where I work, I'm finding most of the EDA tools are being ported from expensive Sun Ultrasparcs to P4/Athlon-based commodity platforms with Linux OSes. And guess what? The processor clock speed has a direct correlation with performance compared to the slow-poke Ultrasparc 3s. You can reach a memory limit for some operations, but tools like Magma and our internal tools that are ported are running at least twice as fast per processor. Particularly with hierarchical designs, the only time the Sun Servers become necessary is for all the back end physical verification, parasitic extraction and signal integrity analysis, where less users are interactively spending their time anyway versus the floorplan/place/route anyway. So, whether I go out and buy an E4500 with 6 processors and 20GB of memory and use LSF, or I buy a dual Xeon 2.4GHz with 4GB of ECC and a Seagate HD, I'm getting a hell of a lot more mileage out of the dual Xeon and a huge cost benefit too for 10% of the entry cost of the Sun.

    Sorry, but commodity PC hardware really does have a place in real computational work on the design of multimillion-gate standard-cell ASICs like the ones going into the latest Nvidia and ATI cards. The Suns are, for now, necessary, but it won't be long until commodity hardware usurps its place for a fraction of their overpriced niche monopoly in EDA tools.

  29. Re:Using Shuttle XPC SS51Gs no less... by Max+Romantschuk · · Score: 2

    Hey, what's with the copy-pasting of my comment? (See above).

    --
    .: Max Romantschuk :: http://max.romantschuk.fi/
  30. Ahh, Comic Book guy by jkitchel · · Score: 2

    Since the headline reminds me of the Comic Book Guy, I will dedicate this post to him. Besides, if this cluster, supercomputer, beo thingy is only 85th fastest in the world, you would think that they would use it to figure out the Comic Book Guy's rating scale (or something of equal importance) as opposed to some physics mumbo jumbo.

    An excerpt:

    [BABF01] Treehouse of Horror X: Desperately Xeeking Xena

    (The Collector, slowly, strikes a dramatic pose)

    Collector(CBG): Lucite hardening ... must end life in classic Lorne Greene pose from "Battlestar Galactica." Best ... death ... ever!

  31. Yes! by Konster · · Score: 2, Funny

    It must be running the website, too. It hasn't been /.ed yet.

  32. On a by Konster · · Score: 2

    On a side note, what's with the cheapo racks they have em on? This whole thing looks to be one seismic wiggle away from disaster.

    1. Re:On a by Meowing · · Score: 2

      That Nexel/Metro type of racking is a lot sturdier than it looks in the photo. All that open space lets air circulate and reduces the surface area for dust to build. Shelving like that beats finding out that you have just the right amount of rack space for the your new toy but the available holes don't line up.

  33. Sure, off the shelve cheap stuff... by dr.Flake · · Score: 3, Interesting

    I really like the picture on the sites frontpage.
    i can imagine the small size of the Shuttles being an advantage, not to mention the "coolness" factor looking at it. (i assume the "cool" in the intro refers to emotion and not teperature!!)

    But getting computation done cheap is no longer the challenge. It's getting the data from one node to the other. They still need "custom" expensive equipment for this.

    I see they use 3com gigabit ethernet. having this 300+ gigabit switch capability is not "cheap".

    Until one can buy this kind of networking equipment for really cheap, we shouldn't mention things like "of the shelve Beowulf super computer in the top 100".

    --
    Why are other peoples sig's always more witty ???
    1. Re:Sure, off the shelve cheap stuff... by tconnors · · Score: 5, Informative

      I really like the picture on the sites frontpage.
      i can imagine the small size of the Shuttles being an advantage, not to mention the "coolness" factor looking at it. (i assume the "cool" in the intro refers to emotion and not teperature!!)

      But getting computation done cheap is no longer the challenge. It's getting the data from one node to the other. They still need "custom" expensive equipment for this.

      I see they use 3com gigabit ethernet. having this 300+ gigabit switch capability is not "cheap".

      Until one can buy this kind of networking equipment for really cheap, we shouldn't mention things like "of the shelve Beowulf super computer in the top 100".


      Up until 3 hours ago, we said we were the only machine in the top500 dedicated solely to astrophysics. Now we are one of 2 :)

      But we use 180 processors, @ 2.15 GHz, and get about 0.90 Gflops/processor, whereas they get about 0.88 gflops/proc, getting us in the top 180th.

      The difference being they have faster memory, and we have a big badass switch. They have two switches, with something like only 10 gigabit between the switches! We have 250gbit within our one switch. a third of our nodes have 2 gigs RAM, and we also have room for upgrade to more nodes on our switch, and they don't. So, in the words of Nelson "Hee haw!" :)

      When they say $1000 per proc, they are not factoriing in their two switches. This will bring the price up to about $500,000, unless someone is donating a switch or 2 :)

      We have about the same cost ratio - something like 250,000 .au dollars for all our procs (and maybe switch - I don't know the details) - half the performance, half the cost.

  34. how about an Ask Slashdot with these guys? by hey · · Score: 2

    I'd ask how hot is the room. The heat pumps move the heat from the CP ... into the room.

  35. Re:Not that impressive by BWJones · · Score: 4, Informative

    I sincerely hope that soon small-to-medium enterprises can own supercomputers. With all the low budget physics stuff going on at Universities around the world, cheap supercomputing can only be a good thing.

    Actually they can with software like that from Dauger Research, Project Appleseed and Wolfram Research with gridMathematica

    The cool thing here is that this code can be run on all of the desktop computers that already occupy companies and universities world wide allowing for easy access to supercomputer level computational speed (for those problems that can be attacked using parallel computation of course) using the same computers normally used for productivity.

    Very cool.

    --
    Visit Jonesblog and say hello.
  36. Do you need a hard drive in each node? by Xenolith · · Score: 3, Interesting

    Disclaimer: I have never built a cluster before.

    What is the advantage of having a hard drive in each node. Can't you boot each node off of a networked image and load the OS and whatever "work" into memory.

    thanks

    --

    Journal
  37. Slashdot culture archive by 0x0d0a · · Score: 3, Interesting

    Has anyone ever written a Slashdot culture archive? Slashdot has a pretty rich and entertaining culture, and I hate to see things like Natalie Portman, Beowulf, and goats.cx...well, maybe not goats.cx....vanish forgotten into the mists of time. I've tremendously enjoyed cultural archives of USENET, where various trends or customs were explained, with links to example text.

    I'd love to read something like this, if anyone ever gets around to setting up a website to archive these.

    It's very difficult to identify trends (like, say, what the meaning of hot grits is) long after the fact -- you're looking at hundreds of thousands of old tech posts. But if someone is thoughtful enough to make a note that this is happening...well, five or ten years from now, it could be quite a fun to read little work.

  38. What would you use one for? by caluml · · Score: 2

    Seriously, assuming that you had the resources to build a large cluster, what would you do with it?

    And I'm hoping I won't get the obligatory "pr0n collection" jokes.

    1. Re:What would you use one for? by dsoltesz · · Score: 2

      PVM-Pov is probably the first thing I'd put on it. Then I'd probably hack around with image processing experiments - and (of course) data mining my erotica library and turning my pr0n collection into a massively parallel screensaver...

  39. Re:Last Post.... by saskboy · · Score: 2

    hey dude, that is cool, and soooooo true.
    Until I press Submit that is.

    --
    Saskboy's blog is good. 9 out of 10 dentists agree.
  40. Topology? by seanadams.com · · Score: 2

    The network switch is composed of a Foundry FastIron 1500 switch trunked to another FastIron 800 switch, which provides a total of 304 Gigabit Ethernet ports using the 16-port JetCore modules.

    What's the bandwidth of that trunk? Also, what's the capaity of the connections between each 16-port card and the backplane?

    Just curious... suppose all the units on a 16-port card have 1Gbps each, but only 8Gb total to the backplane. Then the backplane, in turn, has only 8Gb to the other switch. These are just made up numbers, but how would beowulf handle it? Can it group jobs requiring higher communication throughput onto the nodes which are closer to eah other? Does it have to be told the topology, or does it figure it out?

    1. Re:Topology? by tconnors · · Score: 4, Interesting

      What's the bandwidth of that trunk? Also, what's the capaity of the connections between each 16-port card and the backplane?

      Just curious... suppose all the units on a 16-port card have 1Gbps each, but only 8Gb total to the backplane. Then the backplane, in turn, has only 8Gb to the other switch. These are just made up numbers, but how would beowulf handle it? Can it group jobs requiring higher communication throughput onto the nodes which are closer to eah other? Does it have to be told the topology, or does it figure it out?


      Sounds like they have 10 ports @ 1Gbit each free, so get about 10Gbit between the switches. When we (position 180 on the latest top500) were investigating thin-tree connections, we thought that we might be able to effectively run one job on a third of the nodes, another on the second third, and miscellaneous jobs on the 3rd (since we have two or 3 researchers who like to use lots of resources, and people like me who only need a single proc at a time). So you just partition the nodes and only allow your mpi jobs to sit on one group at a time.

      Then there are things like openmosix which deal with topology automatically somehow. They will try to calculate the speed of the interconnect and the various nodes' procs (if heterogenous), and work out the distribution in the most efficient way. I am going to try to convince my sysadmin to try out openmosix on the lesser-used nodes of the cluster, because there is a feature of it that I think one of us might like - the combination of memory of the different nodes in one big contigous space, but right now we are busy cleaning up after the upgrade.

  41. Re:Did anyone else notice ? by Monkelectric · · Score: 2
    Most people probably didn't notice, given that 106996 bytes is actually only 0.11MB, roughly.

    I dont know what you downloaded, but the jpeg "ss1" on the front page is 1,160,721 bytes ... or 1.10MB

    --

    Religion is a gateway psychosis. -- Dave Foley

  42. rats by djupedal · · Score: 2

    So I spend the weekend tweaking my Shuttle w/Mandrake and my G4 w/OS X as a MySQL/PHP/Apache server and slave, and I feel pretty good that I have this tiny little image server straddling two boxes and working away...then this article about 294 Shuttles comes out and I've suddenly got a bad case of cluster envy...rats.

    Kind of like owning a hot rod Pinto (not), and taking heat from your friends until one shows up in the top ten at Daytona... :)

    1. Re:rats by djupedal · · Score: 2

      cable ties :) I ran out about the 5th time I had to redo the cables. I don't go back there anymore, I'm afraid I'd never make it out.

      I think the only thing that will help that rat's nest is an air brush :)

      Yes, Gallery is working good for me. I went thru a 1/2 dozen apps/script sets before I tried Gallery, and it is well above the rest.

  43. Hmmm..I'll bet they needed this to figure out.... by Newer+Guy · · Score: 2

    Hmmm..I'll bet they needed this to figure out how many radioactive trees there are nearby!

  44. Threepio is Faggy? by ArcSecond · · Score: 2

    I think you mean "effete". I don't think a translator droid has the equipment for that sort of thing. Must be the English accent. Still, you'd think a geek would cut a droid some slack for not coming off like a storm-trooper....

    --

    I've got a bad attitude and karma to burn. Go ahead. Mod me down.

  45. Wahaaaa, Now I know where those SS51Gs are by anonymous+cupboard · · Score: 2

    How did those bastards manage to get so many SS51Gs, especially as the rest of us are having to wait (probably after that /. article last week).

  46. Where do yo swap... by anonymous+cupboard · · Score: 3, Informative
    Back in the days when I was working with LAVCs (Local Area VAX Clusters) over 10-Mbit Lans, diskless systems were possible but any disk I/O hit the bandwidth of the entire network. Pageing could murder the system. Also for larger clusters, simply loading the system on a number of machines took time as a mass storage server can only serve one system per request.

    Ok, modern LANs, especially this one is a lot faster, but you still don't want to burden your cluster communications bus with disk I/O requests.

    Anyway, that 80 gig Maxtor does not add much to the cost of the node.

    1. Re:Where do yo swap... by anonymous+cupboard · · Score: 2
      SCSI is nice but you pay a lot extra for it. IDE fails more, but I suspect that a lot of this is connected with overheating. Someone who buys a SCSI drive is likely to put it into a better housing but the usual home box tends to be undercooled. I don't know how this one runs but from what people are saying, it seems to be ok.

      The main problem with file loads over a network is the lack of multicast protocols for file transfer (I know there are ways, but they are experimental).

      Interestingly enough, these boxes also have firewire which can be used as another form of cluster interconnect.

    2. Re:Where do yo swap... by jonbrewer · · Score: 2

      Anyway, that 80 gig Maxtor does not add much to the cost of the node.

      It's not the price of a hard disk in a cluster node that causes the trouble, it's the fact that you now have one more power-drawing, heat-generating, moving part - that's more likely to fail than anything else in the system!

  47. Re:I doubt this thing will run for long... by kasperd · · Score: 2

    It would have been wiser to use more reliable drives for cluster work such as seagates or ibm's.

    Yes, they should have bought IBM deskstars instead. They are cheap and they know exactly what they are going to get.

    Last year I bought a maxtor harddrive, only thing it ever did for me was crash my BIOS. So I got it replaced with a deskstar, at least I knew that one was going to work. It did.... for half a year. And BTW since then I have only bought seagate drives.

    --

    Do you care about the security of your wireless mouse?
  48. Re:What about the first 84? by shri · · Score: 3, Informative

    http://www.top500.org/

    (Dammit .. I have to wait 20 seconds before I hit submit!)

  49. Small... by BrookHarty · · Score: 2

    Looks like they fit 55 shuttles per rack, so 6 racks of space. Very small in terms of size. My Sun 10K's take up 3 racks each.

    Not sure who makes those plain metal racks, but I picked some up at costco(brown box), and they are sweet. They have big caster wheels so you can get to the cables. I use them in the closet, tv rack, and my server rack.

    Setting up the hardware is easy, I'm curious about the clustering software. Wonder if any 3d rendering packages exist (opensource/free) work on a linux cluster.
    -
    "Marijuana? Cocaine? I'm not going to talk about what I did as a child." - President Bush

  50. So M$ FUD partisans going bored? by Ektanoor · · Score: 2

    So let's have some flame here...

    When it will be the time M$ reaches the top 500 on supercomputing? Linux has been for long there. And it is getting nearer and nearer the first places. However, till now, Redmond couldn't manage to gather even a humble supercomputer made of crappy Windows. What is strangeas there are libraries for parallel computing on Windows.

    So it seems that Windows is not ready for the bleeding edge... And no one knows when it will be...

    1. Re:So M$ FUD partisans going bored? by dohcvtec · · Score: 2

      Good point; MCSEs always boast of Windows' clustering ability (needed for reliability more than anything,) although IIRC it's more high-availability clustering than parallel computational clustering. That said, as highly as MS touts Windows Datacenter server, it's surprising (well, OK not really) that a system running Datacenter Server hasn't even cracked the top 500.

      --
      -- Never hit a man with glasses. Hit him with a baseball bat.
  51. Jabberwocky cluster? by sporty · · Score: 2

    Imagine a raven-cluster of these.. er.. uh.. imagine a rhyme-of-the-ancient-mariner-cluster of these.

    Take all of our fun away by using a beowulf cluster... damned scientists. Everything else just sounds lame.

    The-tyger-cluster? Sounds more like some sorta lame attack. Bah, ferget it.

    --

    -
    ping -f 255.255.255.255 # if only

  52. 1. Xbox's 2. ??? 3. Profits!! by Kibo · · Score: 2

    When xbox live goes on line and MS secretly moves forward with phase 2, harvesting 5 to 10% of cpu cycles of their subscribers, which they then sell.

    Yeah, I know it's not true, but it definately seems like it might have been a good idea.

    --
    --Jimmy has fancy plans; and pants to match.
  53. Good and Good Again by 4of12 · · Score: 2

    I've followed Mike Warren's earlier Linux clusters with interest: Loki (x86), mid-90's, and Avalon (Alpha) a few years ago.

    The free software and low cost supercomputer are not so much news anymore since every intelligent consumer of compute cycles has at least one of these clusters available. No one has to "imagine" them anymore; they are real and commonplace.

    What's a nice development here is that the Los Alamos team has not only brought down the ratio of

    $/FLOP
    but they've started looking to bring down the ratio of
    Watt/FLOP
    as well.

    It represents an uncharastically appropriate use of resources at the Department of Energy and it also helps point the way for businesses looking to further minimize operational costs of racks of computers in air-conditioned rooms.

    --
    "Provided by the management for your protection."
  54. blah blah blah by fizban · · Score: 2

    Wonder what a Beowulf cluster of these things would be like...

    hahaha, I'm so funny. Laugh at my stupid played-out beowulf cluster joke...

    --

    +1 Insightful, -1 Troll. What can I say, I'm an Insightful Troll.

  55. Yes by Lxy · · Score: 2

    But does it run on radioactive trees?

    --

    There is no reasonable defense against an idiot with an agenda
    :wq
  56. Imagine if.. by RobPiano · · Score: 2

    A company made boxes designed to be cheap beowolf components. You could get very cheap, very cool boxes, running in a small space, and have only the min essenetials for clustering.

    Could be really useful for research with limited initial funding.

    -Rob

  57. Re:Not that impressive by Fizzlewhiff · · Score: 2

    What kind of environmental impact does a cluster like this have over a super computer. Hardware wise it is cheaper to build but what does it cost to power it, cool it, and house it?

    --

    'Same speed C but faster'
  58. ACK! by _ph1ux_ · · Score: 2

    "294-unit Beowulf claster..."

    GAHH! Finally a story where the beowolf cluster is an actual part of the story and you misspell it!