This is the voice of world control...

[R.O.A]

Article

WASHINGTON - Scientists unveiled the world's fastest supercomputer on Monday, a $100 million machine that for the first time has performed 1,000 trillion calculations per second in a sustained exercise.

ADVERTISEMENT

The technology breakthrough was accomplished by engineers from the Los Alamos National Laboratory and IBM Corp. on a computer to be used primarily on nuclear weapons work, including simulating nuclear explosions.

The computer, named Roadrunner, is twice as fast as IBM's Blue Gene system at Lawrence Livermore National Laboratory, which itself is three times faster than any of the world's other supercomputers, according to IBM.

"The computer is a speed demon. It will allow us to solve tremendous problems," said Thomas D'Agostino, head of the National Nuclear Security Administration, which oversees nuclear weapons research and maintains the warhead stockpile.

But officials said the computer also could have a wide range of other applications in civilian engineering, medicine and science, from developing biofuels and designing more fuel-efficient cars to finding drug therapies and providing services to the financial industry.

To put the computer's speed in perspective, it has roughly the computing power of 100,000 of today's most powerful laptops stacked 1.5 miles high, according to IBM. Or, if each of the world's 6 billion people worked on hand-held computers for 24 hours a day, it would take them 46 years to do what the Roadrunner computer can do in a single day.

The IBM and Los Alamos engineers worked six years on the computer technology.

Some elements of the Roadrunner can be traced back to popular video games, said David Turek, vice president of IBM's supercomputing programs. In some ways, he said, it's "a very souped-up Sony PlayStation 3."

"We took the basic chip design (of a PlayStation) and advanced its capability," said Turek.

But the Roadrunner supercomputer, named after the New Mexico state bird, is nothing like a video game.

The interconnecting system occupies 6,000 square feet with 57 miles of fiber optics and weighs 500,000 pounds. Although made from commercial parts, the computer consists of 6,948 dual-core computer chips and 12,960 cell engines, and it has 80 terabytes of memory housed in 288 connected refrigerator-sized racks.

The cost: $100 million.

Turek said the computer in a two-hour test on May 25 achieved a "petaflop" speed of sustained performance, something no other computer had ever done. It did so again in several real applications involving classified nuclear weapons work this past weekend.

"This is a huge and remarkable achievement," said Turek in a conference call with reporters.

A "flop" is an acronym meaning floating-point-operations per second. One petaflop is 1,000 trillion operations per second. Only two years ago, there were no actual applications where a computer achieved 100 teraflops — a tenth of Roadrunner's speed — said Turek, noting that the tenfold advancement came over a relatively short time.

The Roadrunner computer, now housed at the IBM research laboratory in Poughkeepsie, N.Y., will be moved next month to the Los Alamos National Laboratory in New Mexico.

Along with other supercomputers, it will be key "to assure the safety and security of our (weapons) stockpile," said D'Agostino. With its extraordinary speed it will be able to simulate the performances of a warhead and help weapons scientists track warhead aging, he said.

But the computer — and more so the technology that it represents — marks a future for a wide range of other research and uses. "The technology will be pronounced in its employment across industry in the years to come," predicted Turek, the IBM executive.

Michael Anastasio, director of the Los Alamos lab, said that for the first six months the computer will be used in unclassified work including activities not related to the weapons program. After that, about three-fourths of the work will involve weapons and other classified government activities.

Anastasio said the computer, in its unclassified applications, is expected to be used not only by Los Alamos scientists but others as well. He said there can be broad applications such as helping to develop a vaccine for the HIV virus, examine the chemistry in the production of cellulosic ethanol, or to understand the origins of the universe.

Turek said the computer represents still another breakthrough, particularly important in these days of expensive energy: It is an energy miser compared with other supercomputers, performing 376 million calculations for every watt of electricity used.

[The Vortex Empire]

Holy shit, I want one. The power of a 1.5 mile high stack of laptops, eh? Just need to find $100 million lying around...

The applications we could put this thing to are endless.

[Singular Intellect]

Finally, a computer system that can play the game Crysis in it's entire visual glory!

[Enigma]

Finally a computer I can game and not need to upgrade for another ahhhhhh, six months? lol

[Drooling Iguana]

Finally, a computer system that can play the game Crysis in it's entire visual glory! :P

On Vista, no less.

[18-Till-I-Die]

So they made Skynet, basically.

I cant wait till "he" opens the automated death camps, the combination of terror (at their foregone deaths) and admiration (at the L33t hardware) on the look of computer nerds everywhere will be a real sight.

[Resinence]

To put the computer's speed in perspective, it has roughly the computing power of 100,000 of today's most powerful laptops stacked 1.5 miles high, according to IBM. Or, if each of the world's 6 billion people worked on hand-held computers for 24 hours a day, it would take them 46 years to do what the Roadrunner computer can do in a single day.

As was said on /.

Yay for completely arbitrary and meaningless measurements? Dude, my car gets 5 football fields to the distance of fuel that you can smell a cow's shit from across the farm. I mean it really puts it into perspective right, it's just sad that they have to dumb down units into utterly meaningless bullshit for people to understand it. 1.5 miles of laptop? How fast are the laptops? What model? How thick? Why fucking bother?

And seriously, he goes on to explain what flops are and that this computer is the first to reach a petaflop, so it really is just there to put it into perspective for morons. Although he then fucks it up by linking the application being run to the achieved operations per second, for some reason (you could feed it completely useless operations and reach a petaflop, what do people think benchmarks do?) And always with the retarded "understand the origins of the universe" bullshit, in other words "will be used for quantum physics problems".

Building such a powerful supercomputer is seriously a great achievement, and it's a huge fuck-you to the idiots who say CELL is slow. But it's just going to be doing the same things all other current clusters are, faster. I'm so sick of journalists who scream how this super-new cluster is magically going to lead to some huge breakthrough in <<insert field here>>.

Yeah, nerd rant, chill out. I'm making something over nothing etc etc. Well you know I'm right so fuck you. *puts on flame-suit*

[K. A. Pital]

So they made Skynet, basically.

Skynet?

Think again: Blue Gene/L (post-upgrade) was powerful enough to sustain a neural network simluation half as complex as a brain of a mouse for 10 seconds. A mouse.

It's a breakthrough, but it's far below of what is required for a true AI in my view

[Sea Skimmer]

And always with the retarded "understand the origins of the universe" bullshit, in other words "will be used for quantum physics problems".

Actually the main use and the only reason why they got 100 million dollars for it is to simulate the initiation of a nuclear device. Its still probably not fast enough for the job… resuming nuclear testing would be absurdly more practical.

[K. A. Pital]

Hypersonic research, SSTO and high overpressure research also requires supercomputing.

It's far more useful and versatile than just for simulating nukes going off.

[The Grim Squeaker]

And seriously, he goes on to explain what flops are and that this computer is the first to reach a petaflop, so it really is just there to put it into perspective for morons. Although he then fucks it up by linking the application being run to the achieved operations per second, for some reason (you could feed it completely useless operations and reach a petaflop, what do people think benchmarks do?) And always with the retarded "understand the origins of the universe" bullshit, in other words "will be used for quantum physics problems".

Yeah, nerd rant, chill out. I'm making something over nothing etc etc. Well you know I'm right so fuck you. *puts on flame-suit*

Actually, it's going to be used for simulating the first second of a nuclear explosion, as part of checking the effects of America's deteriorating nuclear weapons stockpiles.
And it cost 133 million .
Slightly less dumbed down source

[Pu-239]

IIRC the Cell (it's supposed to be capitalized, proper noun, idiot journalists) only good for number crunching, losing its advantage for stuff with branches like AI code?

[Resinence]

Yeah, I'm pretty sure most of the power of the Cell is in the FPU, it's worth noting this machine uses the 2nd specification of Cell which fixes several issues that made the original unsuited to being used as a cluster chip (lack of floating point precision). I've never seen anything impressive from IBM involving it that wasn't number crunching in some form, so your probably right.

Frankly I'm more interested in what can be done with high end GPU's, they have hundreds of shader units which are all very well suited to doing floating point calculations. I'm betting they will be used more and more in the future.

I'm quite surprised they were able to build it so cheap, 133mil seems like a lot but for a bloody petaflop super-cluster they did pretty well.

Actually the main use and the only reason why they got 100 million dollars for it is to simulate the initiation of a nuclear device. Its still probably not fast enough for the job… resuming nuclear testing would be absurdly more practical.

How easy is it to get precise data on the first second of a nuclear initiation in real life with sensors? (honest question, I'm curious)

[Sea Skimmer]

How easy is it to get precise data on the first second of a nuclear initiation in real life with sensors? (honest question, I'm curious)

Not easy, but if we could set off an old nuke to measure, we wouldn’t need that hyper detailed data in the first place. The whole problem is the US nuclear stockpile consists mostly of 20-30 year old devices (with designed lives of typically only 10-15 years) and we aren’t sure they’ll actually work anymore. We have to attempt to do computer simulations, several previous supercomputers have been unable to do the job, because politics has ruled out simply doing a few test shots with them, and its seen as too expensive to just remanufacture every last device into a new one. Plus even if we did build new nukes, which will eventually be necessary no matter what, we’d either need a test shot or the same kind of computer simulation to validate the new design.

It’s a very annoying situation when the solution is so easy… test the nukes. I don’t think people really realize it, but the last US nuclear test was not so long ago, September 1992 in fact. Every nuke we have now was proved with repeated live test shots.

[Shroom Man 777]

I distinctly remember Nastasha Romanenko of Metal Gear Solid talk about computer simulations rendering actual-factual nuclear testing obsolete due to hyper-accurate calculations and such.

Mang, talk about the foresight of 10 year old video games...

[Sarevok]

I distinctly remember Nastasha Romanenko of Metal Gear Solid talk about computer simulations rendering actual-factual nuclear testing obsolete due to hyper-accurate calculations and such.

Mang, talk about the foresight of 10 year old video games...

And a playstation videogame nonetheless.

I wonder if the playstation derived technology has rootkits in it.

[Solauren]

Wow, a computer I can play Homeworld 2 on at full theoritical capacity.

(Using the Star Wars mod and building all possible ships)

[Starglider]

Holy shit, I want one. The power of a 1.5 mile high stack of laptops, eh? Just need to find $100 million lying around...

The applications we could put this thing to are endless.

Actually since it relies on massively parallel processing and DSP-like hardware for most of the compute power, there are major classes of application that it is not particularly good at, and a few that it actually sucks at. For example an overclocked desktop will do stream encryption faster. But for most simulation applications (save for the most tightly interconnected) it would be great - while it's not a completely general design, it's a lot more general than algorithm-specific machines such as MDGRAPE.

IIRC the Cell (it's supposed to be capitalized, proper noun, idiot journalists) only good for number crunching, losing its advantage for stuff with branches like AI code?

This is true for many types of AI code, including the state-machine and script heavy AI you typically get in video games. However the Cell architecture is well suited to low-level connectionist designs (i.e. neural nets and in particular attempts to realistically simulate human neuroarchitecture at the chemical/microstructural level - a major research effort these days).

So they made Skynet, basically.

Some wag came out with that comment for every 'new supercomputer' story for the last couple of decades and probably will for the next couple of decades too. Similarly, I've seen quite a few people say 'oh, Google will one day spontaneously become sentient!' (hopefully you weren't serious but they were). Intelligence is more about software architecture than compute power.

It's a breakthrough, but it's far below of what is required for a true AI in my view Very Happy

In my view standard quad-core desktops are now getting powerful enough to support 'true AI' relatively easily - if we had the appropriate software, which of course we don't. Excess amounts of computing power let you substitute brute force for intelligent design in building an AI; it isn't strictly necessary and is in fact arguably harmful.

[dr. what]

So they're building the most powerful computer ever and then letting it loose on "Nuclear weapons work"?

btw, Britain is launching its fifth Skynet satellite this week...

[Darth Wong]

I distinctly remember Nastasha Romanenko of Metal Gear Solid talk about computer simulations rendering actual-factual nuclear testing obsolete due to hyper-accurate calculations and such.

Mang, talk about the foresight of 10 year old video games...

The idea of making real testing obsolete with simulations has been around for decades, and it's just as stupid now as it was then. Part of the reason for real experiments is that we do not have absolute certainty about our theories and assumptions, and the fact that we might have neglected some contributing factor. None of those issues are dealt with in simulations.