• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Supercomputer sets record: 12,960 chips that are an improved version of the IBM Cell

FrankT

Member
SAN FRANCISCO: An American military supercomputer, assembled from components originally designed for video game machines, has reached a long-sought-after computing milestone by processing more than 1.026 quadrillion calculations per second.

The new machine is more than twice as fast as the previous fastest supercomputer, the IBM BlueGene/L, which is based at Lawrence Livermore National Laboratory in California.

The new $133 million supercomputer, called Roadrunner in a reference to the state bird of New Mexico, was devised and built by engineers and scientists at IBM and Los Alamos National Laboratory, based in Los Alamos, New Mexico. It will be used principally to solve classified military problems to ensure that the nation's stockpile of nuclear weapons will continue to work correctly as they age. The Roadrunner will simulate the behavior of the weapons in the first fraction of a second during an explosion.

Before it is placed in a classified environment, it will also be used to explore scientific problems like climate change. The greater speed of the Roadrunner will make it possible for scientists to test global climate models with higher accuracy.

To put the performance of the machine in perspective, Thomas D'Agostino, the administrator of the National Nuclear Security Administration, said that if all six billion people on earth used hand calculators and performed calculations 24 hours a day and seven days a week, it would take them 46 years to do what the Roadrunner can in one day.


The machine is an unusual blend of chips used in consumer products and advanced parallel computing technologies. The lessons that computer scientists learn by making it calculate even faster are seen as essential to the future of both personal and mobile consumer computing.

The high-performance computing goal, known as a petaflop — one thousand trillion calculations per second — has long been viewed as a crucial milestone by military, technical and scientific organizations in the United States, as well as a growing group including Japan, China and the European Union. All view supercomputing technology as a symbol of national economic competitiveness.

By running programs that find a solution in hours or even less time — compared with as long as three months on older generations of computers — petaflop machines like Roadrunner have the potential to fundamentally alter science and engineering, supercomputer experts say. Researchers can ask questions and receive answers virtually interactively and can perform experiments that would previously have been impractical.

"This is equivalent to the four-minute mile of supercomputing," said Jack Dongarra, a computer scientist at the University of Tennessee who for several decades has tracked the performance of the fastest computers.

Each new supercomputing generation has brought scientists a step closer to faithfully simulating physical reality. It has also produced software and hardware technologies that have rapidly spilled out into the rest of the computer industry for consumer and business products.

Technology is flowing in the opposite direction as well. Consumer-oriented computing began dominating research and development spending on technology shortly after the cold war ended in the late 1980s, and that trend is evident in the design of the world's fastest computers.

The Roadrunner is based on a radical design that includes 12,960 chips that are an improved version of an IBM Cell microprocessor, a parallel processing chip originally created for Sony's PlayStation 3 video-game machine. The Sony chips are used as accelerators, or turbochargers, for portions of calculations.

The Roadrunner also includes a smaller number of more conventional Opteron processors, made by Advanced Micro Devices, which are already widely used in corporate servers.

"Roadrunner tells us about what will happen in the next decade," said Horst Simon, associate laboratory director for computer science at the Lawrence Berkeley National Laboratory. "Technology is coming from the consumer electronics market and the innovation is happening first in terms of cellphones and embedded electronics."

The innovations flowing from this generation of high-speed computers will most likely result from the way computer scientists manage the complexity of the system's hardware.

Roadrunner, which consumes roughly three megawatts of power, or about the power required by a large suburban shopping center, requires three separate programming tools because it has three types of processors. Programmers have to figure out how to keep all of the 116,640 processor cores in the machine occupied simultaneously in order for it to run effectively.

"We've proved some skeptics wrong," said Michael Anastasio, a physicist who is director of the Los Alamos National Laboratory. "This gives us a window into a whole new way of computing. We can look at phenomena we have never seen before."

Solving that programming problem is important because in just a few years personal computers will have microprocessor chips with dozens or even hundreds of processor cores. The industry is now hunting for new techniques for making use of the new computing power. Some experts, however, are skeptical that the most powerful supercomputers will provide useful examples.

"If Chevy wins the Daytona 500, they try to convince you the Chevy Malibu you're driving will benefit from this," said Steve Wallach, a supercomputer designer who is chief scientist of Convey Computer, a start-up firm based in Richardson, Texas.

Those who work with weapons might not have much to offer the video gamers of the world, he suggested.

Many executives and scientists see Roadrunner as an example of the resurgence of the United States in supercomputing.

Although American companies had dominated the field since its inception in the 1960s, in 2002 the Japanese Earth Simulator briefly claimed the title of the world's fastest by executing more than 35 trillion mathematical calculations per second. Two years later, a supercomputer created by IBM reclaimed the speed record for the United States. The Japanese challenge, however, led Congress and the Bush administration to reinvest in high-performance computing.

"It's a sign that we are maintaining our position," said Peter Ungaro, chief executive of Cray, a maker of supercomputers. He noted, however, that "the real competitiveness is based on the discoveries that are based on the machines."

Having surpassed the petaflop barrier, IBM is already looking toward the next generation of supercomputing. "You do these record-setting things because you know that in the end we will push on to the next generation and the one who is there first will be the leader," said Nicholas Donofrio, an IBM executive vice president.

By breaking the petaflop barrier sooner than had been generally expected, the United States' supercomputer industry has been able to sustain a pace of continuous performance increases, improving a thousandfold in processing power in 11 years. The next thousandfold goal is the exaflop, which is a quintillion calculations per second, followed by the zettaflop, the yottaflop and the xeraflop.

http://www.iht.com/articles/2008/06/09/technology/09petaflops.php?page=1


Interesting read to say the very least and the possibilities just wow.
 

ToxicAdam

Member
I'm a dummy in this field, but does this have any kind of ramifications considering Moore's law? Does it shatter that .... or are "stunts" like this not taken into consideration?
 

zoukka

Member
Gives you some perspective on the "power" of games consoles. Remember PS3 marketed as a "supercomputer" :lol
 

bdouble

Member
It will be used principally to solve classified military problems to ensure that the nation's stockpile of nuclear weapons will continue to work correctly as they age. The Roadrunner will simulate the behavior of the weapons in the first fraction of a second during an explosion.

Glad to see it will be put to some good use /sarcasm.

GaimeGuy said:
Seriously.

Using it for the military and nuclear weapon simulations is a god damn waste.

Yeah well we know why. Basically its a GW Bush dick waving competition.

The Japanese challenge, however, led Congress and the Bush administration to reinvest in high-performance computing.
 

GaimeGuy

Volunteer Deputy Campaign Director, Obama for America '16
bdouble said:
Glad to see it will be put to some good use /sarcasm.
Seriously.

Using it for the military and nuclear weapon simulations is a god damn waste.
 

Yagharek

Member
I take it this doesnt count towards NPD?

/sorry

I want to see a photo of this setup. Must take up a pretty big volume I would expect. Dozens and dozens of racks worth.
 

gofreak

GAF's Bob Woodward
2le15zb.jpg


.
 

godhandiscen

There are millions of whiny 5-year olds on Earth, and I AM THEIR KING.
The Roadrunner also includes a smaller number of more conventional Opteron processors, made by Advanced Micro Devices, which are already widely used in corporate servers.

GO GO AMD GO GO!


Nahh jk, most likely the AMD processors here were used because of their low power to performance radio, or to control the other processors. Much akin to what a main core is to the SPE's.
 

SapientWolf

Trucker Sexologist
AlternativeUlster said:
Even with using all these god damned chips, I am still not going to get Toy Story graphics, am I?
Yeah, it probably gets 5 fps in Crysis.

I know render farms are set up for image quality and not speed, but is there a machine out there that could eat World in Conflict for lunch and Crysis for dinner? Maxed out settings at super insane resolutions at a butter smooth 60fps?
 

the_id

Member
It still can't simulate the biochemistry of 1 cell let alone a multi-cellular organism. When are we going to get those kind of super computers for protein engineering?
 
133 million dollars just to see if nuclear weapons stay operative? I wish I were american just to say: "My two cents" :lol
 

gofreak

GAF's Bob Woodward
Deus Ex Machina said:
Sounds like a parallel mess... Not good for video games at all..

Well, videogames better get 'good at parallelism'. 2/3 of the current console CPUs are parallel to varying degrees, and more massively parallel chips seem to be en vogue going forward.
 
gofreak said:
Well, videogames better get 'good at parallelism'. 2/3 of the current console CPUs are parallel to varying degrees, and more massively parallel chips seem to be en vogue going forward.
Game Developers seem not to like parallel CPUs.
 

gofreak

GAF's Bob Woodward
Deus Ex Machina said:
Game Developers seem not to like parallel CPUs.

I think it's a matter of opinion. They don't have much choice anyway currently (sans Wii), and they likely won't have any choice going forward, so it's a pretty moot point. Complaining about parallelism is like pissing in the wind at this point. Everyone will see you doing it, but you'll only end up messing yourself up.
 
I find the idea of supercomputers facinating. Well... more so how quickly they become outdated.

Could anyone tell me approximately what year's supercomputers were roughly equivalent to today's high-end desktop computers? How about iphone?
 

nli10

Member
42



Also this machine seems zetta slow compared to the ones in the future it's nice to see that they built a fast computer that they can't yet use all the speed for though as that will mean that it will last them a good while. Meanwhile a slower supercomputer with a pointer interface will probably get better results as people already know how to program for it.
 

SapientWolf

Trucker Sexologist
gofreak said:
Well, videogames better get 'good at parallelism'. 2/3 of the current console CPUs are parallel to varying degrees, and more massively parallel chips seem to be en vogue going forward.
It would be cake, if it wasn't for that pesky requirement that they offer interactivity. Everything has to be in sync with the main thread running on one of the CPUs.
 

gofreak

GAF's Bob Woodward
SapientWolf said:
It would be cake, if it wasn't for that pesky requirement that they offer interactivity. Everything has to be in sync with the main thread running on one of the CPUs.

I'm not saying there aren't challenges, all I'm saying is that videogame programmers will have to get used to parallelism. If you want more power, it's really the only clear way forward. I think most game developers do probably want more power.
 

patsu

Member
SapientWolf said:
It would be cake, if it wasn't for that pesky requirement that they offer interactivity. Everything has to be in sync with the main thread running on one of the CPUs.

You should check out Mike Acton's presentation in this year's GDC. It talks about the parallel PPU + SPU framework, and how they decoupled the thread logics to minimize syncing.
 

Killthee

helped a brotha out on multiple separate occasions!
RandomVince said:
I take it this doesnt count towards NPD?

/sorry

I want to see a photo of this setup. Must take up a pretty big volume I would expect. Dozens and dozens of racks worth.
09petaflop.enlarge.jpg
 

voltron

Member
Surprise surprise the first thing the Americans want to do with it is analyse the effectiveness of their oh-so-precious stock of nuclear weapons.

How about using it to help reverse engineer alien anti-grav & propulsion technology?
yes I did just say that
 

FirewalkR

Member
voltron said:
How about using it to help reverse engineer alien anti-grav & propulsion technology?

That was done since the fifties.

Now, if you please would look at this little red light...
 

wsippel

Banned
Very badly researched article.

- The PowerXCell 8i processors are not really the main processors of Roadrunner, they are basically coprocessors for the Opteron cores. There's one PowerXCell accellerator per Opteron core, connected via PCIe 8x.

- Roadrunner uses more Opteron cores than CELL processors, as each CELL node contains four Opteron cores, but there are also Opteron-only nodes.

- CELL wasn't just developed for PS3, IBM planned versions for supercomputers all along.
 

gofreak

GAF's Bob Woodward
wsippel said:
Very badly researched article.

- The PowerXCell 8i processors are not really the main processors of Roadrunner, they are basically coprocessors for the Opteron cores. There's one PowerXCell accellerator per Opteron core, connected via PCIe 8x.

- Roadrunner uses more Opteron cores than CELL processors, as each CELL node contains four Opteron cores, but there are also Opteron-only nodes.

Where are you getting your info from? I'd be interested to read more. Wiki says basically the opposite of what you're saying.

i.e. the base unit in Roadrunner is a triblade, which consists of 2 dual-core Opterons and 4 Cells. The sum total across all triblade is 6,480 Opterons and 12,960 Cells. If that's true, Cell is indeed the main player here (and lets face it, it is what gives it its vast computing power..it is the defining element of the system).
 

wsippel

Banned
gofreak said:
Where are you getting your info from? I'd be interested to read more. Wiki says basically the opposite of what you're saying.

i.e. the base unit in Roadrunner is a triblade, which consists of 2 dual-core Opterons and 4 Cells. The sum total across all triblade is 6,480 Opterons and 12,960 Cells. If that's true, Cell is indeed the main player here (and lets face it, it is what gives it its vast computing power..it is the defining element of the system).
What you said doesn't contradict my statement in any way. There are 6,480 dual core Opterons connected to 12,960 PowerXCell "coprocessors" (but like I said, there are also some Opteron-only nodes as far as I know). Cell definitely is the defining element, though - no doubt about it. It's just somewhat funny the article downplays the importance of the Opteron nodes like that.
 

a.wd

Member
voltron said:
Surprise surprise the first thing the Americans want to do with it is analyse the effectiveness of their oh-so-precious stock of nuclear weapons.

How about using it to help reverse engineer alien anti-grav & propulsion technology?
yes I did just say that

Yeah I like this...
 

gofreak

GAF's Bob Woodward
wsippel said:
What you said doesn't contradict my statement in any way. There are 6,480 dual core Opterons connected to 12,960 PowerXCell "coprocessors" (but like I said, there are also some Opteron-only nodes as far as I know). Cell definitely is the defining element, though - no doubt about it. It's just somewhat funny the article downplays the importance of the Opteron nodes like that.

The opteron is a host processor. I guess they don't concentrate on it so much because it's probably expected that the bulk of the work will be done on the Cells.

Los Alamos has their own presentation here - if you like, it also downplays the importance of the Opterons in the system, so I'd forgive the media if it does too.

http://www.lanl.gov/orgs/hpc/roadru...ner Overview/RR Seminar - System Overview.pdf
 
Top Bottom