Slashdot Mirror

← Back to Stories (view on slashdot.org)

First 96-Node Desktop Cluster Ships

Posted by timothy on from the yes-please dept.

Panaphonix writes "The Register reports that Orion Multisystems is shipping the first 96-node desktop cluster. 'With the new, larger system, customers get pretty much the most powerful computer around that can plug into a standard electrical socket.' According to the spec sheet, the DS-96 runs Fedora Core 2 and gets 110 GFlops sustained, 230 GFlops peak."

24 of 323 comments (clear)

  1. Imagine a Beowulf... by Anonymous Coward · · Score: 4, Funny
    Oh, never mind.

    I FAIL IT!

  2. Question by ErichTheWebGuy · · Score: 3, Interesting

    I am not familiar with the architecture of clusters, so I am a little surprised by the more than 100% difference between sustained and peak GFlops. I know what a GFlop is and all that, I just don't immediately see why there is such a huge difference.

    Can someone summarize why there is such a huge difference?

    --
    bash: rtfm: command not found
    1. Re:Question by bobbozzo · · Score: 4, Informative

      finite bandwidth between processors makes it impossible to sustain anywhere near peak performance for most real-world applications.

      Linpack is what is usually used to measure sustained performance on HPC systems.

      --
      Nothing to see here; Move along.
    2. Re:Question by katana · · Score: 4, Funny

      Sustained: Riding down the street.
      Peak: Taking it off some sweet jumps.

    3. Re:Question by brsmith4 · · Score: 5, Informative

      The theoretical max gives a rough estimate of the raw Floating-point power for all of the processors on the system. You pretty much add up the GFlops potental for each node (not exactly, but pretty much). The sustained and demonstrated GFlops of the cluster is based on the Linpack benchmark. The reason there is such a huge difference between the two numbers can be a result of a few factors. 1) The interconnect is GigE and Linpack tends to make use of Message passing comms which are affected adversely by the latencies produced by the GigE connections (myrinet would have been a good choice, but I suppose it was probably impossible to squeeze that into that case) 2) Memory speeds also are a factor as pushing floating point numbers around involves memory. This cluster isn't using anything fancy when it comes to the memory and I suspect this may be another cause for this.

      When they say that this line of clusters can "make or break" Orion, I am right now, leaning for broke. For the cost of this machine, one can get a real cluster with a lot more performance. I know this thing is nice because of the power requirements and the fact that you don't need a dedicated server room to store it, but for $100,000, you can get Microway to build you a pimptacular cluster with Dual-Opteron nodes, high-speed memory and a phat interconnect with either myrinet or infiniband. You will get a lot more work done for the same price.

    4. Re:Question by Daniel+Phillips · · Score: 4, Insightful

      I know this thing is nice because of the power requirements and the fact that you don't need a dedicated server room to store it, but for $100,000, you can get Microway to build you a pimptacular cluster with Dual-Opteron nodes, high-speed memory and a phat interconnect with either myrinet or infiniband. You will get a lot more work done for the same price.

      You forgot a couple of things:

      * HVAC costs
      * Realestate costs

      Remember, this is a deskside cluster. Try that with your dual-opteron cluster. And try adding up all the costs.

      --
      Have you got your LWN subscription yet?
    5. Re:Question by glesga_kiss · · Score: 4, Insightful

      He's also forgotten about hiring an extra 2 guys at least to maintain the extra machines. A selling point of this box will be that "it just works". Pay for a support contract and wammo, you've got a cheap low maintainence cluster. For people working on top-secret stuff (who else needs clusters? ;-), hiring people is a risk and the vetting process is expensive.

    6. Re:Question by femtoguy · · Score: 3, Interesting

      This point is a biggy in the scientific computing world. It is easier to get capitol equipment money than it is to get salary money. (This is because equipment money is overhead free, while salary money incurrs a 50% overhead rate). We just had a donor give us many millions to guy a cluster, but he would commit to long term money for sys-admin support. We ended up including a lot of vendor support into the bid for the contract in order to turn support money into capitol equipment money. Considering that a sys-admin can easily cost $100K per year, this isn't such a bad deal.

  3. Wrong term by AndrewStephens · · Score: 5, Funny

    Its not duped, this article has been clustered.

    --
    sheep.horse - does not contain information on sheep or horses.
  4. Re:Give it to me in terms I can understand. by spectrokid · · Score: 5, Funny

    You want to be impressed heh? Well, it is powerfull enough to play solitaire on Longhorn!!!!

    --

    10 ?"Hello World" life was simple then

  5. Practical measures... by crottsma · · Score: 3, Funny

    With power requirements quintupling that of a standard desktop computer, I'd probably have to use it at my local coffee shop, or only turn it on briefly to scare away song birds.

  6. More Importantly... by mtrisk · · Score: 3, Informative

    Will this make or break Transmeta? It uses their processors (Transmeta Tinside as the Register calls it). Slashdot already pronounced the death of Transmeta though (it has no more niches!), maybe this could revive interest?

    --

    Without a proper flamewar, Anonymous was undecided on what shell to run.
  7. Re:$100,000 by raehl · · Score: 4, Insightful

    $100,000/96 = ~$1,000.

    Not a bad deal.

    --
    paintball
  8. What?! by Anonymous Coward · · Score: 5, Funny

    I can't hear you, the guy two cubicles over just fired up his new Opteron cluster. I'm just trying to hold on to my desk!

  9. Only on /. (Re:$100,000) by po8 · · Score: 5, Funny

    You can tell you're on /. when dividing 1e5 by 1e2 to get 1e3 gets modded up to +5 insightful. :-)

  10. Too expensive by Buzh · · Score: 4, Interesting

    Although having 96 nodes in a single box makes it quite cute, from what I can interpret from the specs, you would get more bang for your $100K by getting what the beowulf crowd like to call MMCOTS (Mass-Market-Common-Off-The-Shelf, i.e. mass produced computers from Dell or the like), hooked toghether with a specialty high-bandwidth low-latency interconnect like Infiniband, Myrinet or SCI. Running a free beowulf cluster OS like for instance ROCKS would mean that a normal linux admin could maintain it quite effectively.

    I expect this thing to be marketed towards scientists in small or medium businesses that aren't employing many/any IT staff, who use commercial computer models to do things like theoretical chemistry (Gaussian, ADF etc), bioinformatics (Phase, BLAS, Paralign etc), fluid dynamics, statistics, crypto, you name it. I don't expect to see any of these types of systems used in normal supercomputing sites, where people write their own (parallel) code and skilled staff maintain the cluster.

    --
    -- Buzh
  11. Re:Apple? by EvilAndrew · · Score: 4, Informative

    Parent poster messed up on their calcs. Current XServes are 18.4 GFlops peak, not 35 eg Virginia Tech currently at #7/500 is 20240 GFlops peak for 1100 XServes. So 7 would only be ~129 GFlops peak, and 33 would be 607 GFlops peak. But not exactly fitting in a single tower case - though 8 would fit nicely one of those mini sound-padded racks which would be almost as good. And at least the last time I saw a price comparison made, the G5s were far cheaper than comparable rack P4s. (The G5 has 2x the FP hardware).

  12. Altitude by hey · · Score: 3, Interesting
    It has Altitude restrictions:

    Altitude -300 meters to +3000 meters
    -1000 feet to +10,000 feet

    I've never seen that before.
    So you cann't use it on a plane.
  13. And? by daVinci1980 · · Score: 3, Interesting

    Who cares? Modern graphics cards are capable of (sorry it's a PDF, it was all I could find) 40 GFLOPS. That's not even in SLI mode, which actually does push you to about a 98% over a single card (in terms of raw processing power).

    Why would you buy a 96-CPU setup when you could buy a 6-GPU setup and match the same theoretical performance? (All jokes aside about the costs being roughly equivalent, they're nowhere near the same.) 6 top of the line 6800s would run you about $3600. Even if you added top of the line parts for the rest of the system, you'd be looking at about $1600 per system. Add $0 for the linux distribution to drive the whole thing, and you're at a grand total of $10K.

    I'm not impressed.

    --
    I currently have no clever signature witicism to add here.
    1. Re:And? by sjwaste · · Score: 4, Interesting

      Serious question here. Does production software exist to drive arbitrary computation across a GPU? I've seen articles about software on its way, etc. Does it exist, either as an application or integrated into some OS? Man, if I could push some of my statistical computing off to the GPU...

  14. The Islands of Patmos Super Computing by Danzigism · · Score: 3, Interesting

    Yea, I agree that Fedora was definitely an odd choice.. Well, I can trust that the kind of person whom can build a 96 node super computer, makes very educated decisions.. I'm glad to hear that company's are still involved in making these clusters. Its a great way to build something powerful for a cheap price, and not having to lean towards Crays etc.. I worked for a company called Patmos International for the longest time, and we never shipped a single cluster.. We had tons of investors that seemed interested of course, but after 2 years of contiuous development, and no sales, the investors simply stopped investing, therefore my job was done for.. We advertised the "$99,000" super computer that would supposedly be in "everyones" garage one day.. Of course that was just a saying because of how cheap we could offer a 32 node system with all of our custom applications and linux operating system. Pretty sweet setup.. it sucks to see the big guys go down sometimes.. to this day, it was the best job I've ever had.. You can still read about Patmos if you search for James Gatzka on google.. They tried their hardest to bring some damn technology and culture to the podunk town of the Eastern Shore of Maryland..

    --
    *plays the Apogee theme song music*
  15. Multichannel EEG processing in Canada vs Chile by mindpixel · · Score: 5, Interesting

    One of my best friends just bought a tiny little house in downtown Toronto for $377,000. I left Toronto last November and moved to Santiago, Chile and live downtwon where my rent is $260/month, for quite a nice, though small place, in an excellent area.

    So, if I spend $100K on the Orion DS-96, that leaves me more than enough for a 250 channel geodesic EEG system which would allow me to compute self-organizing maps of the human mind based on flashing the 1.6 million mindpixels I have collected over the past five years to various volunteers [english teachers], AND still have 56.73 years worth of rent left!

    Too bad no bank will loan me $377,000 for a computer and an EEG system and the time to play with it...

  16. Slashdotted! by 3Suns · · Score: 3, Funny

    Warning: mysql_pconnect(): Too many connections in /home/www/php/functions/executequery.php on line 21

    How blissfully ironic!

    --

    -3Suns

    ~~~~
    The Revolution will be Slashdotted
  17. Re:$100,000 by Grayputer · · Score: 3, Insightful

    How about "plugs into a single 110 volt US outlet". This thing draws 1500 watts PEAK. That's about 14 amps for the math challenged or well under the recommended max for a 20-amp circuit from your household panel (think coffee maker on steroids). Let's try your barebones system approach and let's say you tweak it to use 50 watts per system (good luck). That's 50 watts per system multiplied by roughly 100 systems => 5Kwatts or 3+ times the power consumption. PLUS as an added bonus you get 96 cases, external cabling, at least 96 fans, a dozen power strips, and assorted other toys to trip over. Oh yeah and a few weeks of setup, integration, burn in, and testing.