ROADRUNNER: World's Fastest Supercomputer

[Chess Gator]

http://www.cbc.ca/cp/technology/080609/z060917A.html

U.S. scientists announce world's fastest supercomputer; dub it Roadrunner
Published: Monday, June 9, 2008 | 6:28 PM ET
Canadian Press: Josef Hebert, THE ASSOCIATED PRESS

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

To put its speed in perspective, if every one of the six billion people on Earth used a hand-held computer and worked 24 hours a day it would take them 46 years to do what the Roadrunner can do in a single day. Or, it has roughly the computing power of 100,000 of today's most powerful laptops stacked 2.5 kilometres high.

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 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.

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 nearly 560 square metres with 92 kilometres of fibre optics and weighs almost 227,000 kilograms. 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 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, 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.

[Terry McCracken]

This is very interesting and I know of Roadrunner, but it can't sustain a petaflop for real world applications AFAIK. That's it's peak performance.

OTOH, Blue Gene L will be sold before the end of the year with a peak performance over a petaflop and do real-time scientific calculations at one quadrillion floating point operations. The Blue Gene L project goal is to reach 3 petaflops before work on that project is over and IBM goes on to even bigger and better machines!

Terry

[Chess Gator]

Some related links

http://en.wikipedia.org/wiki/IBM_Roadrunner

http://uk.reuters.com/article/technolog ... 7420080610

http://www.doe.gov/news/6321.htm

[Terry McCracken]

Some related links

http://en.wikipedia.org/wiki/IBM_Roadrunner

http://uk.reuters.com/article/technolog ... 7420080610

http://www.doe.gov/news/6321.htm

Roadrunner is all the hype for the moment as it was first to break a petaflop.

I don't have all the old links, but Roadrunner isn't slated to replace the Blue Gene Project.

Both are IBM.

[M ANSARI]

Well if you think that this system uses dual core processors ... about $1 million of investment and you could double that by using today's low watt Quadcores ... and pretty soon the 8 cores. These things are remarkably useful in predicting the usefulness of a nuclear weapon ... since half life of plutonium and other esoteric materials need to be accurately predicted to make sure a critical mass is reached for the warhead to detonate. Unfortuantely software is still behind when it comes to such hardware ... there are very few applications that can truly use all this power.

[Chess Gator]

I wonder what the ELO of Shredder would be on this machine??

[swami]

I wonder what the ELO of Shredder would be on this machine??

I suspect it would only be 3100 in comp chess ratings. Besides hardware, software is equally important as there are certain positions that some engines do not yet understand even if it searches more. There are still flaws even in stronger engines in faster hardwares and several different kind of bugs vary from engine to engine.

[swami]

A chess engine could search all the possible moves on the board even two moves ahead of every single move, but evaluation varies.

Just searching all the possible moves 2 or 3 plies ahead in fraction of seconds wouldn't tell you much, the engine has to differentiate between "absolute best move" from "really good move", "good move" and a "mediocre move" and choose it.

Also, I believe that there's no "only move" in the middle game or strategy. You take one plan or you could take other route.

Some engines rate one move higher in its evaluation and go on with it even though some other move that leads to different plan is also Ok.

For that you need perfect evluation and understanding. No engine has fine tuned perfectly designed evaluation. Evaluation is the trickiest part.

So, the hardware doesn't play the important role, it just makes the search faster and lets engine that runs on it win in time limited touraments.

Chess cannot be solved, IMO.

[Terry McCracken]

A chess engine could search all the possible moves on the board even two moves ahead of every single move, but evaluation varies.

Just searching all the possible moves 2 or 3 plies ahead in fraction of seconds wouldn't tell you much, the engine has to differentiate between "absolute best move" from "really good move", "good move" and a "mediocre move" and choose it.

Also, I believe that there's no "only move" in the middle game or strategy. You take one plan or you could take other route.

Some engines rate one move higher in its evaluation and go on with it even though some other move that leads to different plan is also Ok.

For that you need perfect evluation and understanding. No engine has fine tuned perfectly designed evaluation. Evaluation is the trickiest part.

So, the hardware doesn't play the important role, it just makes the search faster and lets engine that runs on it win in time limited touraments.

Chess cannot be solved, IMO.

Well most believe that chess can't be solved, but in time people will say that simply isn't so.

One reason is you don't have to work out every legal possible move for chess.

Checkers couldn't be solved either and it has been for all practical purposes. We know it's draw and we suspected that all along. This could be applied to chess even though the task is far more daunting. Machines in the next few decades could actually make this prohibitive task a reality.
There are some great new concepts in the works and they will be reality in time, sooner than people think.
We'll just have to wait and see.

[wgarvin]

A chess engine could search all the possible moves on the board even two moves ahead of every single move, but evaluation varies.

Just searching all the possible moves 2 or 3 plies ahead in fraction of seconds wouldn't tell you much, the engine has to differentiate between "absolute best move" from "really good move", "good move" and a "mediocre move" and choose it.

Also, I believe that there's no "only move" in the middle game or strategy. You take one plan or you could take other route.

Some engines rate one move higher in its evaluation and go on with it even though some other move that leads to different plan is also Ok.

For that you need perfect evluation and understanding. No engine has fine tuned perfectly designed evaluation. Evaluation is the trickiest part.

So, the hardware doesn't play the important role, it just makes the search faster and lets engine that runs on it win in time limited touraments.

Chess cannot be solved, IMO.

Well most believe that chess can't be solved, but in time people will say that simply isn't so.

One reason is you don't have to work out every legal possible move for chess.

Checkers couldn't be solved either and it has been for all practical purposes. We know it's draw and we suspected that all along. This could be applied to chess even though the task is far more daunting. Machines in the next few decades could actually make this prohibitive task a reality.
There are some great new concepts in the works and they will be reality in time, sooner than people think.
We'll just have to wait and see.

Checkers has fewer piece types and a much lower branching factor. It was only solved with the aid of extensive databases for both the beginning and the end of the game (far more extensive than are possible for chess, due to the number of piece types and higher branching factor).

I don't think we'll see Chess "solved" (in the sense of proof of GTV for the starting position) anytime in the next 100 years. I expect it will take exponentially more resources than we currently have to throw at the problem. I would love to be proved wrong, though!