Tuesday, February 3rd 2009
IBM to Supply US Government with 20 Petaflops Supercomputer
IBM will be supplying the US Government with two new supercomputers for the Lawrence Livermore National Laboratory to handle analysis of the U.S. nuclear stockpile. The first is a 500 teraflops supercomputer called the BlueGene/P which the lab will receive by April, the second being the 20 Petaflops supercomputer which is due by 2012. It is estimated to perform up to 10 times greater than the current most powerful systems. More information follows:
Source:
EETimes
BlueGene/P uses a modified PowerPC 450 processor running at 850 MHz with four cores per chip and as many as 4,096 processors in a rack. The Sequoia system will use 45nm processors with as many as 16 cores per chip running at a significantly faster data rate.
Both BlueGene/P and Sequoia consist of clusters built up from 96 racks of systems. Sequoia will have 1.6 petabytes of memory feeding its 1.6 million cores, but many details of its design have not yet been disclosed.
"The Sequoia system will be 15 times faster than BlueGene/P with roughly the same footprint and a modest increase in power consumption," said Herb Schultz, manager in IBM's deep computing group.
"IBM's BlueGene proposal exceeded our requirements while consuming less than half the power of its closest competitor and less than a third of the most power hungry," said Mark Seager, a principal investigator for supercomputers at Lawrence Livermore. "That is a savings of more than 35 megawatts for the lifetime of the machine and translates to a savings of $24.5 million and 108 kilotons less carbon dioxide expended," he said in an email exchange.
"These powerful machines will provide NNSA with the capabilities needed to resolve time-urgent and complex scientific problems, ensuring the viability of the nation's nuclear deterrent into the future," said NNSA administrator Thomas D'Agostino in a press statement. "This endeavor will also help maintain U.S. leadership in high performance computing and promote scientific discovery," he added.
42 Comments on IBM to Supply US Government with 20 Petaflops Supercomputer
For the record, Crysis is one of the worlds most poorly coded games, please refrain from comparing hardware potential with Crytek's poor excuse for a game.
Kei
(down with skynet)
The Mayans probably just ran out of rock. That, or they grossly estimated the amount of time that their empire would last, and just wanted to be prepared.
:)
either the fact that it can handle all of the US's nuclear stockpile
OR
That we make crysis jokes about it
:S
makes u think......
*has dramatic music in background*
So yes, its very vital that both of them scenarios dont occur and if having to use supercomputers as an alternative to underground testing is all these ' superbrains ' can come up with then sure . . . BUT if these people really had half a brain, they wouldnt have to stockpile nukes to begin with.
We have all seen movies such as Defcon and i wouldnt doubt that some of you even named your computer Joshua . . . But i have no doubt in my mind that if you ran a simulation of life and its endless variables on this supercomputer, the outcome would simply be along the lines of;
Mankind is f'd in the head for having nukes. No self respect let alone respect unto others.
When's Terminator Salvation comming out ?
I believe it is a waste of taxpayer money. The only time DOE should have to buy a new super computer for the purpose of stockpile longevity is if the facility the old one was in got bombed, hit by lightning, etc. where it would be cheaper to replace the system than attempt to repair it; moreover, the replacement system should be no more than 20% faster than what is needed to complete its assigned duty (operate at 80% load).
Get what I'm saying? As far as I can tell, a 20 petaflop machine is a waste of money for a 1 teraflop job (or whatever the figure is). Taxpayers shouldn't be paying out to have the fastest super computer in the world for the sole purpose of having the fastest super computer in the world.
Now, if they were going to use that 20 petaflop machine to figure out the problems and/or make fusion power more efficient, I'd be all for it. It would pay for itself rapidly. It's kind of like NASA where every taxpayer dollar spent on the program creates at least two dollars of return in new products and scientific advancement. It's no surprise that scientific advancment as a whole has wanned considerably with spending cuts directed at NASA.
Let's remember that just 10yrs ago we had some p3-500 w/ 500MFlops more or less while nowadays we had quad core i7 w/ 100GFlops+ of power, and let's say reading of pdfs w/ that time actual Acrobat 4 (or never forgotten V.5 two yeras later) was way much comfortable at that computer then nowadays reading w/ V.9 :) So in the 4 years when this rig farm become functional it'll be just enough for the current jobs they intend to do on it :/
On the other hand we use to say as more they're of us we become more stupid which is in fact induced lie. Why? Cause we more and more depend on that Tech That FordGT90 praise for. And we're just unfit to catch up with our fluffy brains so ofc we need implants that someone will want to sell us :). The same pieces of tech that coudn't even be SFX-ed for movies 30yrs ago.
And that tech advancement nus-produced by NASA that became useful in our everyday life. It would be great that they're sold it as original products an not as merely formula or an idea an not by them but by some third party that bought off something that seemed interesting for mass production (aka. profitable). The original claims should belong to that tax payers ... tax payer consortium :)) and to rightfully split claims between all of them it would need an SF 3 zeta Flops computer that might be produced in 2050. Of course anyone who claims rights for that products should be considered responsible for all bad quality implementations and pollution of the same when they end their life cycle.
And about formula-product-implementation it certainly shouldn't be allowed to sell someone in third world an knowingly dangerous formula just cause they don't have enough expertise or laws to consider you responsible. So the only proven products should be sold and the right owner must and should have rights to oversight that all procedures during production of that products are not wrongly implemented.
Of course that not even 1% of that procedure is done, and that praised tech are just been resold as bare patents, and all of possible misuse of the same lies in hands of end product inventor and their expertise and judgement (..sanity and morale) pressured by short as possible timeframe to invent it, and sold it while not been properly tested. So in the end there's no sci-tech advancement just bare degeneration w/ unproven half-products.
uhmmm ... well that 20PFlops sc could be used for that kind of pre-product process simulation and waste and recycling sims of the same cause bare warhead oversight and their course plots don't need more than 1PFlops.:D
20 Pflop = 20,000 Tflop = 20,000,000 Gflop = 20,000,000,000 Mflop
20,000,000,000 Mflop / 670 Mflop = 29850746.27 Core i7 920 processors
I think they are expecting a faster processor to come out by 2012 because that number of processors is simply ridiculous. That, or IBM PowerPC processors have higher flop performance than Core i7 does.