Slashdot Mirror

← Back to Stories (view on slashdot.org)

Flash Mob Supercomputer?

Posted by timothy on from the all-the-cool-kids-are-doing-it dept.

dan of the north writes "The NY Times (free reg yyy bbb) is running an article on flash mob computing. More info on the first event in SF on April 3, 2004. The goal is to run Linpack and "build a home-brew computer powerful enough to be added to a list of the world's 500 fastest computers." Minimum requirements are 1.3 GHZ Pentium III/AMD equivalent or better with 256MB of RAM, a 100 Base-T network connection and a CD-ROM - laptops preferred. "After taking a shot at a speed record, the computer will be reorganized to serve as the host of a giant multiplayer video game tournament." Cool... a 2fer!"

5 of 259 comments (clear)

  1. Printer friendly flash-free link by Anonymous Coward · · Score: 5, Informative

    Link

  2. GO USF/DONS by Zero_K · · Score: 5, Informative

    As a USF Student, I will be there with my laptop...All I have to say is GO DONS... The computer science department already has a cluster called the "keck cluster". Basically 64 nodes of dual P3 at 1GHz, with 1 Gig ECC Ram. There is talk about throwing the keck cluster into the flash mob cluster, but the biggest hurdle is appearntly laying the lines. Harney Science center is about 200 hundred yards from the gym where this is going to happen. And just FYI, they wanted it to be done on the 1st of april, but that didn't work out for some reason.

  3. The first time ... NOT by kb · · Score: 5, Informative

    Maybe somebody should point out that is not the first time somebody has done such a thing... back in 1998 there was a quite similar event at the University of Paderborn where 512 normal home PCs brought by people were connected for one night (the event was even broadcasted live on German TV). I have to admit that the "flash mob" element here is more predominant (back then people knew about this two weeks in advance), but it's definitely not the first attempt to create a spontaneous supercomputer with home machines. The cluster even made it into the Top250 IIRC. :)

    More info...

  4. been there done this ... by thehosh · · Score: 5, Informative

    guys from the CLOWN '98 [http://www.tlachmann.de/linux-cluster/] already tried this (even it was not the main goal). it was a temporary cluster for only one night, but to get into top500 you have to build a durable cluster.

  5. Re:Erm, well good luck....you'll need it... by thesupraman · · Score: 5, Informative

    Exactly, speaking as someone who has run CFD code on smallish clusters, 100MB ethernet falls flat on it's face at about 20 machines, and those were dual 500MHz machines - it will be worse with faster ones! (to preempt some silly comments, CFD code and linpack have a LOT in common)

    And that was using specially tuned low latency ethernet drivers and TCP stack under linux.
    These guys have very very little chance of doing anything useful at all - which is a bit of a pity, but perhaps if they did just a little research first..

    I wonder if they even have network switches that will efficiently route 1200 nodes.. let alone a decent plan to interconnect them.

    The first step would be to use 1Gbit or faster concentration to some very smart switches to at least cut down the network blockage a little.. It won't help with the terrible latency, but will give them a little headroom at some vector lengths.

    They will also suffer terribly from the differing speeds of nodes - I've yet to see a solution for linpack that distributes efficiently over a wide speed range of machines.

    Of course, I bet in the end they just come up with a great SETI score, or something similar - something that would actually scale at all on a cluster like this.

    Oh well, I wish them luck anyway.