FlashMobComputing

I was checking this out and I thought it might be cool to kill an hour or two.

http://www.flashmobcomputing.org

The idea is pretty simple. Users link their computers up on a network, boot with FMC's bootable cdrom, it loads up Knopix with Linpack (used to test supercomputers), and you determine the power of your "LAN Supercomputer". All that is needed is a 1.3 GHZ Pentium III Celeron/AMD equivalent or better with at least 256MB of RAM and a 100base network card.

I'm going to try to play with it some before the party if people are interested in seeing how much computing power we can generate as a group.

Can we hack the gibson after that!?


_________________
I am your mom and I just took you to school in the car of pain.

http://www.pimpedoutcases.net/article.php?50.0

As luck (?) would have it, AMD machines ended up being quarantined from the rest of the group because of some trouble with the software. Apparently, the AMD machines tore through the benchmark quite a bit faster than most of their Intel counterparts, but had trouble communicating with the Intel boxes, which were in the vast majority.

LOL. We need Maxvla to bring all his uber'clocked rigs along to make sure we post some good numbers.

Heard about this the other day on TechTV. Pretty interresting. Only problem is when they did it in CA they had like 700 computers and they only were able to get like 70Gflops. Not so great if you ask me concidering "super" computers in the top ones have ~ 2000 or 35,000 if your the No. 1.

Still a neat idea, but impracticle as you can read from the article.
_________________
For with what measure you measure it will be measured to you.

I read the TechTV article then more info on their site. It looks like they hit 180 GFlops with 256 computers but it crapped out when one crashed (I'm suprised the benchmark didn't continue working). When they hit 77 GFlops they were only running 150 computers. I heard somewhere that they had alot of problems with the network and thats why they couldn't get better results.

I'm not saying that we are going to top something like the 2000 G5 Mac Servers but it could still be alot of fun to see what we can accomplish.

It sounds really cool... when I was watching the TechTV news about it I was wondering why (via the internet) everyone in the world couldnt just make one giant supercomputer. But then I figured the internet bandwidth was probably the limmiting factor. But, it raises the question of how long will it be till bandwidth is high enough and computers are relyable enough that we could all join togeather and make the fastest computations ever seen.
_________________
For with what measure you measure it will be measured to you.

They have: SETI, Distributed.net, Folding@Home, etc

The flashmob idea is a different approach though. Yes, bandwidth would definitely be a limiting factor in this approach.

I'm hoping that I can get my hands on our new work server (when it arrives) to test it out. Its a dual opteron with 8 gig of ram. When we order the db server it'll be a quad opteron with 32 gig of ram (assuming the dual works as well as its supposed to). I'm watching a couple of quad itaniumI's on ebay now but they are becoming old technology.

I would love to be able to work with that kind of hardware. Working for the state sucks. We have no money . All you go out and pay your taxes so I can go buy some sweet hardware!

(or maybe I should just look for a job elswhere) heh.
_________________
For with what measure you measure it will be measured to you.

ya same i would love to work with that kind of hardware!!!
_________________
mmm, mmmm, mmmmm, um ,mmmmm

On a side note, the 85gigaflop score produced by the AMD subnet of a mere 30 machines was not combined at any point in the benchmarking.

Corporate sponsors such as HP and E-Loan also donated about 100 machines a piece, and many USF offices and classrooms also had their computers unknowingly plundered for the sake of more raw power.

I'll particerpate

Shinare wrote: I would love to be able to work with that kind of hardware. Working for the state sucks. We have no money . All you go out and pay your taxes so I can go buy some sweet hardware!

(or maybe I should just look for a job elswhere) heh.

Shinare wrote: I would love to be able to work with that kind of hardware. Working for the state sucks. We have no money . All you go out and pay your taxes so I can go buy some sweet hardware!

(or maybe I should just look for a job elswhere) heh.

Blu, your 3.2EE should do pretty well, especially if we can shortterm O/C it.

My XP 2700+ owns you bitches!

riiiiiiighhttt

Well, if AMD's "tore through" the benchmarks compaired to Intels, then I would venture to say that whatever they were using these computers for favored the AMDs. I have both my Intel P4a [email protected] AND my wife's Barton 2400+ at 2.2GHz on air (the barton can go much higher but waiting on better cooling, wanting 2.6). I'm really interrested in seeing which one works better at this. Perhapse if we all ever get togeather and do this I'll bring both as long as they get the wrinkles with AMD CPUs ironed out in the software.
_________________
For with what measure you measure it will be measured to you.