PS2 Supercomputer

[Boltzmann]

Greetings,

Tonight I was reading the articles as usual at KurzweilAI.net, and saw this interesting news:

New York Times, May 26, 2003


The National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign has assembled a supercomputer from an army of Sony PlayStation 2's for about $50,000.

The center's researchers believe the system may be capable of .5 teraflops. It uses the PlayStation's graphics co-processor, the Emotion Engine, which is capable of producing up to 6.5 gigaflops, rather than its microprocessor.


The original article can be found here (http://www.kurzweilai.net/news/frame.html?main=/news/news_single.html?id%3D1986)

The original NY Times article is located here (http://www.nytimes.com/2003/05/26/technology/26XSUPE.html), but I couldn't read it since I'm not registered.

Interesting news anyway. Maybe I'll try to make a nuclear bomb using the PS2 for doing the calculations :D :D

Live long and prosper.

[Player-X]

Stick a few together and you'll have a missile tracking system

[Unicron]

I wonder how many flops the human brain is capable of?

edit:..or if there is a way to measure that...

[Hanamichi]

Of course you wont, unless the PS2 has some sort of heat tracking system for evil purposes like tracking the user to use gamma rays on him. :evil:

[Unicron]

Originally posted by Hanamichi
Of course you wont, unless the PS2 has some sort of heat tracking system for evil purposes like tracking the user to use gamma rays on him. :evil:

you COULD expect :evil:Sony:evil: to do something like that.

Concerning the brain I wonder if this is actually accurate??? http://www.tmeg.com/ai/brain.htm

[Rhombus]

It's an interesting article... however there's one thing it doesn't really factor in, and that's learning and evolution. The human brain hasn't really physically changed since the rise of Homo Sapien Sapien... that was what? About 15 to 20 thousand years ago? But it's taken us this long to learn and evolve to use our brains. Even if an emergent intelligent DOES exist in 16 years (or whatever it said), it certainly wouldn't instantly become a doctor or a lawyer or whatever. It would be like a caveman robot.

[Tori-Yama]

when games become reality.

[Boltzmann]

Originally posted by Unicron
I wonder how many flops the human brain is capable of?

edit:..or if there is a way to measure that...

All we can have are estimatives. Like the one found in Robert Bradbury "Matrioshka Brains (http://www.aeiveos.com/%7Ebradbury/MatrioshkaBrains/MatrioshkaBrains.html)":

Computer and Human Operations Equivalence.
At the simplest level of abstraction, neurons can be considered to be multiplication and adding machines. Neurons multiply the "strength" of a synaptic connection times the "weight" of an incoming signal and sum these values across a number of input synapses. If the result exceeds a certain threshold, the neuron fires and transmits a signal to other neurons connected to its network. Neurons fire very slowly, < 100 times per second. The immense power found in the human brain is due to neuron features other than speed. These include their small size, low power consumption, high interconnection levels (100-10,000 per neuron) and to a large degree sheer numbers. The human neocortex, which is the most highly developed portion of the human brain, and that part which is thought to be responsible for "higher thought", contains ~21 billion (2.1x1010) neurons [Pakkenberg, 1997]. The total number of neurons in the brain is less certain, but since the neocortex contains roughly 1/3 of the brain volume, unless neurons density is much higher in other brain regions, extrapolations from Pakkenberg's data would imply there is a total of 60 billion (6x1010) neurons in the brain. To provide a proper perspective, if current SIA projected trends continue, microprocessors would not have 60 billion transistors circa 2025. Even then, a single transistor does not possess the computational capacity of a neuron. On the other side of the coin, a microprocessor with 60 billion transistors would occupy a volume much smaller than that of the human brain.
If we assume 6x1010 neurons x 5x101 firings per second x 103 operations per neuron firing , we end up with a result of 3x1015 operations per second (300 Trillion operations per second or 300 TeraOps). This is likely to be at the high end of possible computational capacities since it is assuming that all neurons are being used simultaneously. This is unlikely to be true since the brain clearly has specialized structures for visual, auditory and odor input; speech output; physical sensation and control; memory storage and recall; language analysis and comprehension; and left-right brain communication. It is unlikely that all of these structures will be optimally utilized at any point in time.

A high end estimate of 300 TeraOps for human thought capacity does not significantly differ from those found in the literature.


Notice that TeraOps aren't the same thing as teraflops (which is the measure given in the article that you found).

>>Even if an emergent intelligent DOES exist in 16 years (or whatever it said), it certainly wouldn't instantly become a doctor or a lawyer or whatever. It would be like a caveman robot.

It would be true only for a raw, unlearned AI. And given the database capabilities of the new machine, the possibility given by evolutionary algorithms and all, I believe that a better-than-humans AI is highly probable (once we have the needed computing power).

[ready2rumbelX]

I don't know this for sure, but wasn't there a problem, before the PS2 was released, that the PS2 could set off or control missiles? I remember hearing something like that before it came out...

ready2rumbelX :D

[Galway no Sora]

Not the PS2, but the memcard IIRC. I myself didn't belived that article, but it's an overwhelming issue. Dunno the reality.

[ready2rumbelX]

The memcard?.....strange. :eek:

Oh well, PS3 will probably be able to launch nukes. ;)

ready2rumbelX :)

[Syed Fawad]

How many did they use?

[Razor Blade]

Originally posted by ready2rumbelX
I don't know this for sure, but wasn't there a problem, before the PS2 was released, that the PS2 could set off or control missiles? I remember hearing something like that before it came out...

ready2rumbelX :D

Actually I read that in OPM when the PS2 first came out. It was Saddam Houssein bought 60 PS2's and North America and other parts of the world were scared that he was going to use them to set off some sort of missiles or use as a supercomputer. This is no lie. :dead:

[Demigod]

60 PS2s could potentially add up to 390 gigaflops per second, from the estimations Boltzmann posted. Still, a 32-way Itanium-2 supercomputer could do the same thing.

[Syed Fawad]

Originally posted by Demigod
60 PS2s could potentially add up to 390 gigaflops per second, from the estimations Boltzmann posted. Still, a 32-way Itanium-2 supercomputer could do the same thing.

Don't you think that the 60 PS2s will be cheaper? :p

[Boltzmann]

Greetings,

How much would cost a "32-way Itanium-2 supercomputer"? :)

Live long and prosper.

[Betamax]

Well IIRC the one I use for my research cost the department something like £1.1 Million ($1.78922 Million). Of course that included software, support, etc...

[Syed Fawad]

Nah, I'd prefer the 60 PS2s :p

It'll also allow you to play FFX :eek:

[Boltzmann]

Originally posted by Betamax
Well IIRC the one I use for my research cost the department something like £1.1 Million ($1.78922 Million). Of course that included software, support, etc...

:bigeek: So much?
Then I really would rather like 60 PS2s :)

Besides having the computing power of a supercomputer I would play Silent Hill 3 ;)

[Demigod]

Originally posted by Boltzmann
So much?
Then I really would rather like 60 PS2s :)

Besides having the computing power of a supercomputer I would play Silent Hill 3 ;) It's not unusual for supercomputers to cost that much. Consoles are only cheaper because the games are what make most of the money. They intentionally sell consoles at a loss and rely on the sell of games to make up for it. Thus, PS2 machines for a supercomputer should cost more per unit, since there won't be any game purchases to balance out the books.

Hey, playing games on a supercomputer, what a waste of precious CPU cycles... current games (at least PC games) are too primitive to take advantage of the power of supercomputers or even today's consumer computers. Tests done by Kyle Bennett of HardOCP has concluded that Intel has reached one of the "Holy Grails" of enthusiasts with their 800 Mhz FSB. Increasing the FSB to 1000 Mhz yielded absolutely no benefits in single-gaming situations (traditionally where large performance gains are shown with FSB overclocking), which suggests that current games (as in today's latest PC game technologies) do not stress the CPU bus enough. It's only when running multiple apps that benefits are shown. Of course more CPU power will help but not many games are programmed for multi-thread usage anyways. PS2 games certainly won't benefit from it, as the games are designed to run on a single PS2 console (adding more CPU power could also throw the timing off on games).

[Boltzmann]

Originally posted by Demigod
It's not unusual for supercomputers to cost that much. Consoles are only cheaper because the games are what make most of the money. They intentionally sell consoles at a loss and rely on the sell of games to make up for it. Thus, PS2 machines for a supercomputer should cost more per unit, since there won't be any game purchases to balance out the books.

Hey, playing games on a supercomputer, what a waste of precious CPU cycles... current games (at least PC games) are too primitive to take advantage of the power of supercomputers or even today's consumer computers. Tests done by Kyle Bennett of HardOCP has concluded that Intel has reached one of the "Holy Grails" of enthusiasts with their 800 Mhz FSB. Increasing the FSB to 1000 Mhz yielded absolutely no benefits in single-gaming situations (traditionally where large performance gains are shown with FSB overclocking), which suggests that current games (as in today's latest PC game technologies) do not stress the CPU bus enough. It's only when running multiple apps that benefits are shown. Of course more CPU power will help but not many games are programmed for multi-thread usage anyways. PS2 games certainly won't benefit from it, as the games are designed to run on a single PS2 console (adding more CPU power could also throw the timing off on games).

I was just kidding about playing games in the supercomputer, though :)
But you gave a great explanation about gaming and CPU power, anyway. I should increase my own hardware knowledge (mine is very limited).

[roxor]

Well one thing that cou;d be done with games to make more use of the CPU would be to implement raytraced graphics for them. But given current hardware, for photorealistic graphics, it would still be too slow. I suppose yoiu could always just use basic Wolf3d style graphics, though.

[Demigod]

But why would you want to make more use of the CPU? The PC CPU has more than enough things to do (aside from 3D acceleration it pretty much does everything). Besides, why do raytracing when you can have polygon-based 3D accelerators?