IBM Retakes Fastest Supercomputer Title

dshaw858 writes "BBC News reports that IBM has unveiled its new Blue Gene/L machine. The Blue Gene project already has two of the top ten supercomputers in the world. Big news for IBM! I wonder what great things they can calculate in just seconds now... maybe I should get a stronger PGP key."

Comments

Don't worry

[FiReaNGeL]

"IBM and its partners are currently exploring a growing list of applications including hydrodynamics, quantum chemistry, molecular dynamics, climate modeling and financial modeling."

So no PGP key cracking. At least officially.

I wonder how the Fold@Home total CPU power compare to this in terms of percentage?

Re:Don't worry

[Coryoth]

So no PGP key cracking. At least officially.

You really need something more than just a really fast/powerful computer to do PGP cracking. You're going to need something that can help you get your fingernails under the problem, because even this machine couldn't brute force PGP keys. There has been some papers written on theoretical weaknesses in RSA that, given a custom built machine, could be exploited. This is not a custom built RSA cracker. It may have enough raw power to make up for that of course, and that means you might manage 1024 bit RSA cracking if you are determined. Unfortunately any sane PGP/GPG users are using Diffie-Hellman/El-Gamal rather than RSA as their public key system, and for now there aren't any similar attacks for the discrete log problem as there are for factoring.

Your paranoia is misplaced. You should be worried that the NSA has come up with a serious break in RSA and Diffie-Hellman schemes that let them be cracked by a nice ordinary supercomputer, rather than worried about computer power overtaking key size. Most key sizes are chosen to have a fairly long lifespan even with massive increases in computing power. You aren't going to brute force 128bit symmetric systems any time soon, no matter how much computing power you stack up against it. No, the fear is in breaks to the encryption scheme.

Jedidiah.

Re:Don't worry

[JPriest]

If people want your key so bad they will build a supercomputer this big to crack it, you have plenty other things to worry about.

Re:Don't worry

[metlin]

Financial Modelling _is_ a big thing. I've worked on modelling stock and economic data using game theory and various analytic methods -- it's not as simple as that.

There's a lot of patterns, and a hell lot of data processing to be do.

However, that said, financial data modelling is not something which I think can be cracked using brute-force power. Although there has been a lot of fundamental progress in terms of using OR and GT algorithms and the like, it hasn't really had that "big breakthrough" to fundamentally determine the basis of financial data and market behaviour, and perhaps we never will.

Ofcourse, as always hope springs eternal - but that would also make markets a whole lot deterministic and bring about some serious differences in the way business is done.

Re:Don't worry

[Duncan3]

I wonder how the Fold@Home total CPU power compare to this in terms of percentage?

Folding@home has almost 3x the FLOPS, but we're all on the same side here. Slightly different problems can be tackled when you have local bandwidth.

Also keep in mind that Folding@home is not one project, but dozens of projects sharing the same CPU pool.

Years to go before we figure out how folding really happens...

Re:Don't worry

[mystran]

Ofcourse, as always hope springs eternal - but that would also make markets a whole lot deterministic and bring about some serious differences in the way business is done.

But hey, that'd immediately change the markets, so you'd probably need a new model.

I don't believe that predicting the markets on the big scale is ever possible, simply because the predictions affect the system. You would have to predict how your predictions affect the system, and if that was public, then how the predictions on effects of predictions... and so on..

Re:Don't worry

[Coryoth]

Aren't Diffie-Hellman and El Gamal just key exchange methods? I didn't know they had anything to do with the encryption itself...

Diffie-Hellman is just key exchange, El-Gamal is effectively using Diffie-Hellman style operations for encryption. The important thing to remember is that PGP/GPG only uses the public key aspect for key exchange. The message itself in encrypted with a symmetric cipher scheme, and the public key is simply used to exchange the one time key for the symmetric cipher for that particular message.

Jedidiah.

Re:Don't worry

[orangesquid]

Hmm.

But when do you think distributed.net will have a BlueGene/L port?

Re:Don't worry

[Tim C]

Good point.

What would you do - sink a few hundred million in building a supercomputer to crack some guy's PGP key, or kidnap him, hold a gun to his head and ask for the passphrase?

You'd build the computer if it was imperative that the guy not know you'd cracked his encryption, or if you wanted to do it on a large scale. If it's just one or two guys, and secrecy isn't necessarily an issue, there are other ways...

Re:Don't worry

[Gil-galad55]

As my number theory prof once famously said, "There are easier ways of finding out secrets than factoring large primes."

Well, I thought it was funny!

Re:Don't worry

[fatphil]

How can something that says "this machine couldn't brute force PGP keys." and "that means you might manage 1024 bit RSA cracking if you are determined" get moderated to +5?

You're gibbering, sir. You say one thing and then the opposite.

No-one "brute forces" PGP keys, that's not how you crack them. Exactly how you crack them depends on what the underlying algorithm is, it's either GNFS factoring or discrete logarithm, but _neither_ is brute force. So your first point is wrong.

With current algorithms, 1024 bits is completely out of reach. The algorithms are mostly "embarassingly parallel", and therefore there's little gain from a tighty-coupled supercomputer (except at the LA stage at the end, but that's a fraction of the total workload). So this machine is no greater than the sum of its parts (i.e. several thousand high speed processors). Such a setup cannot crack 1024-bit keys. So your second point is wrong.

FP.

Re:Don't worry

[JanPeterBalkenende]

I wonder how the Fold@Home total CPU power compare to this in terms of percentage?

They don't. This seems something that needs to be cleared up every supercomputing topic. Things you can calculate on a supercomputer, like the LINPACK benchmarks, are impossible using distributed network computing. Network computing is only useful for problems that can be chopped up in many pieces, each of which can be solved by one computer on its own.

In problems that supercomputers have to deal with, hydrodynamics, climate modelling, the work can be chopped up, but partial results of each processor need to be communicated to other processors and this many times during computation. Now if you think how many messages you can send over a network in a second and compare this with the number of computations that can be done in a second, it is easy to see that this kind of communication easily becomes a bottleneck. Even a processor accessing its own memory experiences this kind of bottlenecks, now imagine a system in which a few thousand processors have to share some information. As an extra handycap the problems of cooling an overclocked pentium is child's play compared to cooling thousands of processors crammed together.

If it weren't for these practical problems, a little math shows that if you would tie some 20,000 Pentium-4, 3Ghz 's together (each giving you about 3800 MFlop/s), you would have yourself a system of 76 TeraFlop/s for only $20 milion. And you think you can buy a Blue Gene for that money....

Re:Don't worry

[NanoGator]

"If people want your key so bad they will build a supercomputer this big to crack it, you have plenty other things to worry about."

If somebody actually gained access to one of these machines for that specific purpose, I'd just hand them the key and say "You win!"

Re:Don't worry

[Vicsun]

I wonder how the Fold@Home total CPU power compare to this in terms of percentage?

F@H is currently at 196 TFlops, while the IBM machine is at 70.72 TFlops. So Folding@Home runs at 270% the speed of Blue Gene/L. Pretty neat =)

Re:Don't worry

[spitefulcrow]

Only for now. BlueGene/L has 70 teraflops while still under construction. When finished, as TFA said, it'll do 360. I know one of the guys working on this, he said the architecture can scale to 4 million processors before it becomes inefficient.

Re:Don't worry

[spitefulcrow]

BlueGene/L has some neat solutions to those problems. The interconnects are done in a 3-dimensional torus where each processor communicates (over a custom bus) with those around it in 6 directions. Memory is on the same card as the processors. 16 processor cards (with 4 processors on each one) make up one board. The boards are mounted sixteen to a frame, two frames to a rack (2048 CPUs per rack) and talk to each other and the storage system over GigE.

Re:Don't worry

[ezzzD55J]

" As my number theory prof once famously said, "There are easier ways of finding out secrets than factoring large primes."

Nitpick: Factoring primes eh?

The technical term for one such method is rubber-hose cryptanalysis..

I was a bit relieved to find out that, according to a book (Ross Anderson's Security Engineering I think) I've read, the implied method is slapping a rubber hose against the subject's feet.. I was worried about more awful forms of rubber hose application to the human body.

I suppose this is plenty awful soon enough though.

Re:Don't worry

[Gil-galad55]

Yeah, I deserve that ;) My excuse is that it was early morning... but here's hoping my professor doesn't read this!

Re:Don't worry

[EngMedic]

to quote from the jargon file

rubber-hose cryptanalysis: n.

[sci.crypt newsgroup] The technique of breaking a code or cipher by finding someone who has the key and applying a rubber hose vigorously and repeatedly to the soles of that luckless person's feet until the key is discovered. Shorthand for any method of coercion: the originator of the term drily noted that it "can take a surprisingly short time and is quite computationally inexpensive" relative to other cryptanalysis methods.

Re:Don't worry

[PingPongBoy]

Unlikely to target just one person - an attack would be a demonstration of a crack. In a way of speaking supercomputers are cracking the secrets of nature - to find some reliable pattern of prediction without having to do the massive calculations required by fundamental models, or if all else fails to demonstrate that massive calculations are still doable for problems at the scale of humanity.

Bah

[zaxios]

IBM Retakes Fastest Supercomputer Title

If their supercomputers really were that fast, they would have taken the title back earlier.

Re:Bah

[RealProgrammer]

That's comedy gold!

Uh oh

[paul248]

So, IBM is taking the "Fastest Supercomputer" title away from NEC's Earth Simulator. How can NEC stand for this obvious theft of intellectual property? I sense a lawsuit brewing...

Re:Uh oh

[LiquidCoooled]

Yes, the RIAA are going to sue IBM for billions of dollars.

They played one illicit mp3 at 70 teraflops.

An RIAA spokesperson said "Playing a song at those astronomical speeds is highly illegal, it almost burnt our accountants fingers just counting the zeros!"

Re:Uh oh

[fbform]

They played one illicit mp3 at 70 teraflops.

Knowing the RIAA's history, they'd probably claim that it was equivalent to playing millions of MP3s at consumer PC speeds. :-)

"has two of the top ten supercomputers"

[Anonymous Coward]

Ah...

Must be those 2 guys I always see playing Quake with 1ms pings.

Re:"has two of the top ten supercomputers"

[dshaw858]

No, those are the guys with the OC-128 lines.

- dshaw

Re:"has two of the top ten supercomputers"

[adler187]

Don't you mean Doom 3?

Re:"has two of the top ten supercomputers"

[26199]

...who are also very close to the server. The speed of light is only 300,000km/s, so if you want a 1ms ping, you can only be 150km from the server, at the very most.

Re:"has two of the top ten supercomputers"

[100_Monkeys_Typing]

electricty moves slower than the speed of light in copper. So unless everything is fiber... it would have to be a lot closer.

Re:"has two of the top ten supercomputers"

[Anonymous Coward]

this also applies to fiber because the rays actually travel like /\/\/\/\/\/\/\/\

Re:"has two of the top ten supercomputers"

[26199]

True. But I believe it's something like .7 times the speed of light. So in fact it's not that big a difference.

Remember, it's not the electrons which transfer information (they travel very slowly indeed, on average). It's waves in the electric field.

Re:Again?

[SonicBurst]

No, that record was set by a previous machine. This one is just a prototype for a much larger/faster version, and still managed to hit 70 teraflops...

Re:Again?

[Anonymous Coward]

The machine has more nodes this time. It clocked in at over 70 teraflops, instead of the "mere" 36 that they had last time.

They probably did this because NASA/SGI's Columbia machine did over 40 teraflops a few weeks ago and the Top 500 list is coming out this Monday. They wanted to be on top, I think. :)

What about SGI?

[enigma32]

I recently read that SGI was to be claiming the world's most powerful supercomputer record from the Earth Simluator...

Does this mean that IBM leapfrogged SGI or does this mean that the SGI machine (to be built for NASA) wasn't all that exciting?

http://www.sgi.com/features/2004/oct/columbia/

Re:What about SGI?

[LnxAddct]

It means that the SGI announcement was of theoretical performance. In theory it is the fastest machine on earth... they are in the process of verifying that with tests right now. Once its proven you'll see the title taken by SGI again. Then the Earth Simulator will be 3rd on the list.
Regards,
steve

Re:What about SGI?

[Anonymous Coward]

There will always be a competitor claiming the fastest supercomputer. It's always a game of leapfrog.

The rankings used in supercomputing measure very specific benchmarks and have very specific deadlines.

In this case, SGI has a computer reportedly faster than Blue Gene/L, but it is neither 1) in production by the deadline nor 2) independently verified results.

Sure, they might do this, but by then the next supercomputer will shame SGI's new baby. Like I said, leapfrog.

Rankings are all about a fairly arbitrary snapshot in time.

Re:What about SGI?

[Anonymous Coward]

No, it just means that IBM (this time, at least) are full of PR-shit. The supercomputer they're talking about has one big difference from NASA's new system - the IBM box hasn't even left development yet, whereas the SGI system has been shipped and installed at a paying customer's site.

Need proof? Here's one way you could go about getting it:

(you) Hi IBM, I'm thinking of buying a BlueGene system for my lab, but I'm wondering - what operating system can I install on it?
(IBM) It runs Linux!
(you) But I looked carefully through the Linux source code, and couldn't find any mention of BlueGene systems. Are you sure it'll work?
(IBM) Actually, you need to use our modified version of Linux.
(you) Oh okay, no problem. Where can I get the source?

At this point, the IBM sales rep's head will implode as he realises he has only two options: he can either stall (and the longer he stalls, the more it looks like he might be violating the GPL!), or he can admit the truth: the operating system doesn't _actually_ exist yet in any sort of finished form yet.

On the other hand, SGI's version of Linux that runs on the 256-way Altix BX2 systems is a free download, as it should be. Start here.

Hope this helps! Your post does raise the interesting question of where the Top500 group (or benchmarking people more generally) should drawn the line: it's nice to know about cutting edge machines, which may even be "one of a kind" installations, but somehow it's not so helpful when companies announce performance results of things they haven't even finished building it. It would not be out of the realms of possibility that the final BlueGene machine shares little (or nothing) in common with the machine they're talking about today, which would mean that this PR is basically nothing but a fairly baseless grab at mindshare.

Re:What about SGI?

[Anonymous Coward]

SGI are NUMA. This is a cluster.

SGI NUMA boxes are the fastest *single system image* machines around. No question.

Re:What about SGI?

[NanoGator]

"Does this mean that IBM leapfrogged SGI or does this mean that the SGI machine (to be built for NASA) wasn't all that exciting?"

That depends: Are we talking today or yesterday?

Re:What about SGI?

[eXtro]

Bullshit. The sgi machine was up at NASA doing real work. The IBM machine's benchmarks have only been repeated in their manufacturing floor. I'm not saying the results are bogus but I'm saying that sgi is the one with the production ready supercomputer.

The bigger deal is that this sucker was put together and running customer code in 15 weeks. The build went so smoothly that people were surprised when a freshly installed machine didn't just run by the end of the day.

Re:What about SGI?

[Johnny Mnemonic]

(you) Oh okay, no problem. Where can I get the source?

(IBM) When you purchase (and take delivery of) the system, we'll be happy to give you the source upon request. Of course, there's nothing in the GPL that requires that we give source to everyone! Only to those who purchase and take delivery of our binaries.

Re:What about SGI?

[Wesley Felter]

Wow, nice FUD. It is true that BG/L is not finished, but it's more mature than you make it out to be. The hardware is finalized; the only difference between the current system and the final LLNL system is the number of racks. The software is working well enough for people to run real apps; all that's left is fixing bugs and tweaking for better scalability.

Re:What about SGI?

[thatguy[tc]]

Actually... The Top500 rankings system does establish some limits to prevent vendors from building, benchmarking, and publishing results on systems which are not available to customers. The restriction is that only 5% of a single vendor's total listed processing power can be at vendor owned facilities. See the Call for Participation.

We're not sure if BlueGene will be included on the list to be published this week, as it's an engineering system on IBM's manufacturing floor, and does not (yet) belong in a comparison of production systems. IBM needs 1,330 Terraflops of non-internal systems on the list to qualify BlueGene. The current list has ~400 TF worth of IBM systems, so BlueGene is unlikely to be on Tuesday's list unless the list maintainer decides to include it anyway.

What's certain is the the Earth Simulator will no longer be #1: NASA's SGI cluster was installed on site and in production even before being benchmarked, and the announced results of 42TF will push it down a notch just as easily as the "quietly submitted" results of 51.

IBM's system will certainly be high on the list when it enters, but subject to the list maintainer's discretion, they may have to wait until actually shipping it before they can have it included on the list. If they can manufacture, ship, and benchmark a 70TF system in the next 6 months, they likely will be #1 when the next list is published. Unless of course somebody buys an even larger Altix cluster or another vendor releases something competitive.

The restriction on submissions was designed for exactly this type of situation, and there's two parts to it:
What can be built in a lab is a function of money, and the larger vendors with other revenue streams can afford to build large systems for no purpose other than to post a benchmark - while smaller vendors can only afford to build what they can sell.
Posting more flops than anyone else is good work, but it really doesn't mean much to the HPC industry or user community until people can buy one, have it installed, and run their code on it.

Re:What about SGI?

[RageEX]

> Does this mean that IBM leapfrogged SGI

Yup. Though I don't think anything is 'official' till the new top500 list comes out.

> or does this mean that the SGI machine (to be
> built for NASA) wasn't all that exciting?

Columbia is definitely exciting. Even if it doesn't post the largest LINPAK numbers it is an exceptional machine on several different metrics. And don't use the future tense, the system was finished in October.

Re:What about SGI?

[RageEX]

> In this case, SGI has a computer reportedly
> faster than Blue Gene/L,

All IBM has is a report too. Nothing is 'official' until the next top500 list is released.

> but it is neither 1) in production by the
> deadline

Amazing what people believe and then pass on without checking out the facts. The SGI system was fully installed and running in *October*. This was on Slashdot last week. In fact one of the interesting things about the story was that the whole system was built in only 4 months. In fact it is the IBM system which is not running in production. This is such a glaring error I wonder if you typed SGI instead of IBM...

>nor 2) independently verified results.

Neither are IBMs. Wait for the top500 list. 3 numbers have come from SGI, 42-ish, 51-ish, and 60-ish. But ultimately who cares? It's merely interesting trivia. Besides the SGI system is interesting for several reasons and even if it is not at the top of the top500 list it still leads in other metrics.

How 'bout

[iminplaya]

I wonder what great things they can calculate in just seconds now...

How 'bout this? 1,000,000! It tatkes pretty long on my P3.

Re:How 'bout

[acidblood]

You just need the right software. With PARI-GP my 2.6 GHz P4 takes a little more than 13 seconds.

Re:How 'bout

[Eric604]

for something like that, asm is the right tool

Chaos Theory...

[oneiron]

They really need to get these things crackin on chaos theory... How many inhabited planets equals one amino acid chain? What are our odds of hitting the protein jackpot? You know?

Re:Chaos Theory...

[nwbvt]

"What are our odds of hitting the protein jackpot?"

100%. Given that we exist on this planet (which is of course a necessary fact in order for there to be a 'we' in the equation), proteins must exist on our planet. The probability of any given planet having proteins evolve is irrelevant as we do not live on just 'any given planet', we must (as a condition of our existence) live on a planet on which life did evolve.

Re:Chaos Theory...

[Coryoth]

They really need to get these things crackin on chaos theory... How many inhabited planets equals one amino acid chain? What are our odds of hitting the protein jackpot? You know?

Ah, Chaos Theory possibly the most widely popularised, yet least widely understood areas of mathematics ever. Exactly how is Chaos Theory going to help in counting extrasolar planets, or calculating probabilities? You need to actually have some understanding of the system before you can hope to actually apply any dynamical system theory to it at all. Presently, I don't think we do understand exactly how random chemicals manage to form proteins, and self replicating chemicals. I don't see how a fast computer and a fueld of math largely irrelevant to the subject at hand is going to help.

Jedidiah.

Re:Chaos Theory...

[oneiron]

Do we exist on this planet or do we exist in this universe? I'm not sure if you grasped my point. I was hinting at a wild idea/hypothesis/theory that popped into my frontal lobe, recently:

-Our universe as part of a pool of goop on a planet which exists on a size-scale millions of times greater than our own.

-Humans as the worker-bees trying to create the puzzle pieces of the amino-acids/protiens that are developing in this pool of goop.

-Or maybe a pool of goop inside of some alien uterus or maybe that's all existence is....an infinite loop of birth and death and alien uteruses... (new idea just now)

Ok, I could go on forever...more and more metaphysical with each passing sentence. You get the point. Is your scientific frame of mind bored yet? What about your spiritual one? Are they both just annoyed?

Re:Chaos Theory...

[oneiron]

Think in broader terms, please. I know what Chaos Theory is. It doesn't answer my questions directly, but it points us in the right direction.

Re:Chaos Theory...

[__aaijsn7246]

You do awesome drugs...
But seriously, I have had similar thoughts :)
Maybe our universe is just a logic gate?! All that we do just goes down to 1 output..!@#@

Re:Chaos Theory...

[div_B]

we must (as a condition of our existence) live on a planet on which life did evolve.

Or maybe life evolved somewhere else and arrived here sometime before now. You want to consider that? Or does that just complicate things to much for you, and you'd rather not make pesky distinctions like that between necessary and sufficient?

Re:Chaos Theory...

[Photar]

My wife hits the "proten jackpot" quite often. Oh wait, that protein jackpot. Nevermind.

Re:Chaos Theory...

[nwbvt]

Highly unlikely and completely irrelevant even if it did happen. It doesn't really matter if the "protein jackpot" evolved here from lose amino acid bases or came here from an extraterrestrial source. Their abundancy is still a necessary condition of our existence.

And it is certainly not a sufficient condition. Anyone with half a brain knows that just because amino acids were able to come into existence in no way means that intelligent life, much less us in particular, would also come into existence on the same planet.

Re:Chaos Theory...

[nwbvt]

Those are some powerful drugs you are taking there. Doesn't mean what you wrote is in any way meaningful.

Re:Chaos Theory...

[nwbvt]

"I know what Chaos Theory is."

No you don't. Put down the joint before you hurt someone.

Re:Chaos Theory...

[oneiron]

Meaning is a relative term.

I know my ideas are nothing more than a few of the billions of infinite possibilities of existence. Can you say the same of yours?

Is it dangerous to aspire to such great heights? I think it could be, but I'm still learning (you've helped). There is a balance. I know this much. I intend to position myself just to the left of center because I've found the surface in the middle is a tad slippery.

Re:Chaos Theory...

[div_B]

Highly unlikely and completely irrelevant even if it did happen. It doesn't really matter if the "protein jackpot" evolved here from lose amino acid bases or came here from an extraterrestrial source. Their abundancy is still a necessary condition of our existence.

Of course having amino acids in abundance is a necessary condition of our existence. However, you originally stated something quite different:

we must (as a condition of our existence) live on a planet on which life did evolve.
That's just plain wrong. Just because there are humans here, it doesn't immediately follow that life evolved here. It's a good bet, sure, but you were taking it to be a logical truth, which it aint. That's why it it's very relevant, because the possibility life evolved somewhere else, and arrived here, denies it the status of logical truth.

And it is certainly not a sufficient condition. Anyone with half a brain knows that just because amino acids were able to come into existence in no way means that intelligent life, much less us in particular, would also come into existence on the same planet.

Granted, and I should have spelled it out more carefully:

We see that human life exists here. That this life could be traced back through it's evolutionary ancestry to a few amino acids forming in a rock pool, on this planet, under god knows what conditions, would be sufficient to explain our observation. It wouldn't be necessary though, as there are other ways we could have arrived at the state we observe today. That's the distinction I was talking about.

Re:Chaos Theory...

[nwbvt]

"That's just plain wrong. Just because there are humans here, it doesn't immediately follow that life evolved here. It's a good bet, sure, but you were taking it to be a logical truth, which it aint. That's why it it's very relevant, because the possibility life evolved somewhere else, and arrived here, denies it the status of logical truth."

There are truths other than logical truths. Logical truths can only exist in an abstract representation of the world. I was clearly discussing a scientific truth. All evidence suggests life evolved on Earth, and no respected scientist will argue otherwise. Regardless, your nitpick is still irrelevant.

"That this life could be traced back through it's evolutionary ancestry to a few amino acids forming in a rock pool, on this planet, under god knows what conditions, would be sufficient to explain our observation."

Your egregious misuse of the terms "necessary" and "sufficient" make your nitpick concerning my use look downright silly.

Re:Chaos Theory...

[div_B]

Unless your hypothesis is as consistent or more consistent with our observations of reality as other hypotheses, and at least as simple it's not particularly useful.

[Sigh] Of course, that would be true if I were hypothesizing, however I was not, and nor was the GP. He was supposedly deducing that life must have evolved here, given that we are here today. I agree that it's very likely that life did evolve here, but it's certainly not so bloody self-evident that it has probability 1. And if he were hypothesizing, then being so ridiculously dogmatic about one's position wouldn't exactly make for good science, IMHO.

Re:Chaos Theory...

[oneiron]

I know.

Re:Chaos Theory...

[oneiron]

Maybe our universe is just a logic gate?! All that we do just goes down to 1 output..!@#@ In some sense...

Re:Chaos Theory...

[ratsnapple tea]

Dude, lay off the poor guy. You don't have a lot of friends, do you?

Yeah.

[Anonymous Coward]

But can it cook me dinner yet? Seriously how much f***ing computer power do we need to bake brownies? I can't wait to throw out my girlfriend 1.0 once they finally come up with one that doesn't put up a inpenetrable firewall in bed.

Re:Yeah.

[Anonymous Coward]

AMD makes the only chip hot enough to bake anything.

Re:Yeah.

[konstantinlevin]

/sbin/bake -Cc cookie
rpm frontdoor
order -pmsgo --timelimit=30 pizza

Re:Yeah.

[prockcore]

I can't wait to throw out my girlfriend 1.0 once they finally come up with one that doesn't put up a inpenetrable firewall in bed.

That reminds me of that old joke:

"The only thing that a hacker can't get past is a pair of panties"

Re:Yeah.

[YetAnotherGeekGuy]

But can it cook me dinner yet?

At 1.5 MW it can cook dinner for a whole city. Reminds me of all the Mainframes that got donated to universities since the 60's. They're really neat to have as a museum piece (even if they do consume the entire basement's space), but no University runs them because they can't afford the power to turn them on.

Re:Yeah.

[NanoGator]

" I can't wait to throw out my girlfriend 1.0...that doesn't put up a inpenetrable firewall in bed."

Impenetrable firewall, or are her available sessions in use?

Re:Yeah.

[IdleTime]

Ahem....

A few reasons do do an emerge -C girlfriend

This begs the question:

[Anonymous Coward]

Will this computer be able to beat humans at chess?

Re:This begs the question:

[Clueless Moron]

My ancient palmpilot can beat pretty well anybody on the street at chess.

Get over chess fetish already. It's not what humans are best for.

Re:This begs the question:

[Mathiasdm]

You don't even NEED a 70 teraflops computer. This one will do: http://www.research.ibm.com/deepblue/meet/html/d.3 .html

Re:This begs the question:

[sxtxixtxcxh]

... says our new supercomputer overlords.

Re:This begs the question:

[whataboutMike]

As mentioned in this article http://games.slashdot.org/article.pl?sid=04/01/06/ 0511220&tid=127&tid=208&tid=10 computers have no problem beating humans at chess. What would really make news is if computers would start beating humans at Go. Then again, Go is much less about brute force and deep searching; and more about pattern recognition. Something that humans seen to have a monopoly on.

Re:This begs the question:

[jackinluv]

Chess is a game of predictable logics. Only thing is that chances are 50/50 to win or loose. But even the worlds most fastest computer ever designed will led to just three options WIN LOSE DRAW. So not a big deal

Appliccations

[RealProgrammer]

They meant hydrodynamics, financial modeling, etc. But no mention at all of how to combat spurious lawsuits.

In an apparent first for /. today, mo mention of robots, either.

This is OT, but I never noticed it before - the following HTML works here:

link to slashdot:
<a href="/.">/.</a>

Re:Appliccations

[Anonymous Coward]

That html works anywhere, its an absolute path on the current server (slashdot.org) the path is /. which expands to http://slashdot.org/. and the . is either removed by your browser or redirected by the webserver to http://slashdot.org/ or http://slashdot.org/./ (which is of course, the same as http://slashdot.org/)

Re:Appliccations

[sik0fewl]

No, that's complate and udder (yes, like the cow) nonsense. This is magic and clearly only works when used on the real slashdot

Re:Appliccations

[Jeff DeMaagd]

I've tried it outside of slashdot and it doesn't work, not even in Firefox, which surprises me. IIRC, if it's a few words or at least not a valid URL, Firefox tends to look up the item in Google's "I'm feeling lucky" search and go there. It doesn't seem to work even with putting "/." in the address bar, it tries to do "I'm feeling lucky" but it doesn't work.

Yes, again. But...

[Thu25245]

...this time, it's from NASA. http://www.space.com/businesstechnology/nasa_super computer_040809.html

There's been a lot of turnover recently. For those of you keeping track at home, it's now:

IBM BlueGene/L (70.7 teraflops, up from36 in your article)

(?) NEC SX-8 (Not yet installed anywhere; estimated 58.5)

NASA/SGI Columbia (42.7)

NEC Earth Simulator (35.9)

Re:Yes, again. But...

[Jeremy Erwin]

Often, cluster builders test their systems with a submaximal number of nodes (so as to detect nasty bugs, and so forth). "Project Columbia" actually benched 51870 GFlops. And NEC's SX-8 is still close to vaporware.

Notice how the BlueGene/L folks released a benchmark just barely enough to beat the reigning Earth Simulator. Then, when NASA released its monster, they added nodes in earnest, retaking the "title".

Just remember

[Raul654]

Don't mess with people who measure their server power in acres. :p

Re:Just remember

[Orgazmus]

That is one of the finest lines of text i have read in a while. And the .sig is great too ;)

Re:Just remember

[darkmeridian]

ENIAC was a roomful. My TI is stronger than that.

Re:Just remember

[cuteseal]

But don't forget - smaller is not always better :)

Re:Just remember

[mog007]

Can your TI read punch-cards? I THINK NOT!

Re:Just remember

[roman_mir]

Yeah, well, you are certainly taking the parent's joke out of context.

Re:Just remember

[spitefulcrow]

No, it only takes up a room. I've seen the earlier model at IBM's TJ Watson lab in Yorktown, NY (dad works there). They have a few racks with processors and then a few more racks for storage (tape robots!). But each of the processor racks holds 2048 CPUs.

Re:Oh, this is so great

[HBI]

The parent poster is referring to this book, which was from about three years ago.

I have read it. It's fundamentally a hatchet job. IBM was the prime supplier of Hollerith punched card machines worldwide, whether they were sorters or keypunch machines or whatever. The fact that they supplied them to the Nazis was used to create a conspiracy whereby IBM favored the extermination of Jews.

The book appeared to be angling to tarnish Thomas J. Watson, the founder of IBM, primarily, rather than the modern company.

The future of patent law has just struck me

[zaxios]

This brings up an enticing possibility. What if Microsoft just patented "being first"? Wouldn't that get rid of all the prior art rubbish they have to cope with with their other patents? I mean, if someone showed prior art for "feline flatulence" or whatever else is developing in Bill Gates' unfortunately windowless office, they would be infringing Microsoft's "being first" patent. This is it folks! The future!

Re:The future of patent law has just struck me

[name773]

Bill Gates' unfortunately windowless office
wait, so he's using office on a mac?
the world i once knew is crashing down before my eyes...

No key cracking

[acidblood]

I recall reading on the RealWorldTech forums that these are highly specialized machines and particularly geared to floating point computation. As integer factorization, index calculus computation for discrete logarithm cracking, Pollard rho attacks for computing elliptic curve discrete logarithms, etc. are integer algorithms, crypto should be safe from this particular beast.

And before anyone asks about symmetric/secret-key cryptosystems and hash functions, recall that these are also based on integer operations, so they're safe from the BlueGene as well.

Re:No key cracking

[fatphil]

Anything that does arithmetic with integer maths can be done in FP too. My PIES project, like GIMPS and all the others, does integer maths almost entirely in the FPU units.

Logical operations, yup, they're out of scope, but addition and multiplication, which are the heart of all the arithmetic algorithms you mention, can all be hived off to the FPU.

Phil

Re:No key cracking

[acidblood]

Hello Phil, how are you doing? Sorry I didn't answer your email back when I received it.

However, I don't think we can easily do sieving, for instance, using the FPU. That would rule out the most powerful factoring algorithms like NFS and MPQS.

Also, GIMPS is completely different from your ordinary key cracking workload. They're doing huge FFTs on really huge numbers. You can't really compare it to a factoring program working on hundred-digits integers.

Re:No key cracking

[acidblood]

Yes, quantum computing can factor integers in polynomial take. However, it takes exponential time to simulate a quantum Turing machine. So in the end your algorithm is still exponential on a deterministic Turing machine, and probably with much worse asymptotics to beat. I bet my lowly Pentium 4, running an implementation of the Number Field Sieve, would be faster the Blue Gene running a simulation of Shor's quantum factoring algorithm.

Re:No key cracking

[Jeff DeMaagd]

I am guessing the chips do have integer units though. I suppose they could drop the intmultiply and intdivide units from the circuit masks but it doesn't make sense to do that. The strong point of this kind of supercomputer is the data interconnect, whether that data is int or FP is secondary, although it should definitely be fast at FP.

Actually, I think it makes sense to keep a intmultiply unit in there for calculating table addresses, it is silly to drop that, hardware is likely to do this faster than software.

Re:No key cracking

[fatphil]

Hi Decio, things are fine, the ECM job is pulling factors out by the dozen daily, but it's an enormous task. You're invited to join in for a bit if you like.

Inasmuch as the 'sieves' are memory bound, then yes, the task don't benefit vastly from use of the FPU. However, everything else can. The FPU concretely helps me in my sieving tasks on <64 bit numbers (50% int, 50% FPU, twice the throughput), and also for medium-sized numbers too. Look at DJB's zmodexp, for example. How do you do your modular square roots in MPQS? Most of the time it's a modular exponentiation, in which case, I'd definitely be using the FPU a la zmodexp.

Phil

1000000! in hex

[3770]

I did that in hex on a 486DX266 back in the day. It took approximately a month.

I did it in hex because it was easier to write an efficient algorithm.

And then I decided to write a program which would convert that huge resulting hex number to decimal.

Only, that is when I realized that it would take more computational power to convert that number to decimal from hex, than to start from scratch and do it in decimal "natively".

Re:1000000! in hex

[iminplaya]

I had to read that twice to realize you didn't have a 266mhz 486.

Re:1000000! in hex

[andreyw]

No, he had an i486-DX2 clocked at 66Mhz.

http://www.cpu-collector.com/menu/searchresults/ re cord/127.htm

Anyways I posted this just in case you thought he was jokeing. I

Re:1000000! in hex

[Scott Carnahan]

A month sounds awfully long for a 486, and it suggests that you were using a rather non-optimal multiplication algorithm.

Judicious use of an FFT-based algorithm should reduce the computation time by some orders of magnitude. There are also fast base conversion algorithms for large integers. These can be found in e.g. Knuth, TAOCP.

Oh sure

[Clueless Moron]

But can it play Doom3?

Penguin Power.

[Anonymous Coward]

"Big news for IBM! I wonder what great things they can calculate in just seconds now..."

Their sales of Linux.

Re:Penguin Power.

[NanoGator]

"Their sales of Linux."

Error: Division by 0.

Re:Penguin Power.

[Thomas Miconi]

"Their sales of Linux."

Error: Division by 0.

You mean they tried to calculate the ratio of people switching to linux over people switiching back to MS ? :)

Thomas-

Re:Hm

[dtfinch]

Oblicatory beowulf cluster comment

Re:Hm

[dtfinch]

Typed a little too fast there

If you were to RFTA before submitting...

[nwbvt]

"I wonder what great things they can calculate in just seconds now... maybe I should get a stronger PGP key."

It is quite clear what these computers are doing. They are designed to compute the folding patterns of protein molecules, a task which requires immense computational power.

Re:If you were to RFTA before submitting...

[Dept. Head Rawlings]

Yes, nwbvt, the ab-initio calculations are much harder. N-body quantum mechanics is not easy. The naive classical approximation has O(N^2) operations per timestep. You then need to integrate over a lot of timesteps to see the protein fold, so there's an enormous prefactor there. Does this prefactor depend on N? Of course not, that would require the existence of N(thantos) based equations. I don't see any reason the timestep depends on N, and recall N^2, but I may be wrong... From what I recall from work by Marcus Snir 4 years ago, current calculations didn't even come within orders of magnitude of the timescales on which proteins are known to fold. PGP is easy compared to this. Of course blue gene changes all this. Jim Rheingold would be happy, we all know the sacrafices he made to make this happen. One step closer, yes.

Re:If you were to RFTA before submitting...

[tho 1234]

The DOE does not compute folding patterns or anything else that will benefit humanity. The sole purpose of this computer is for the development/simulation of new generations of nuclear weapons.

Re:If you were to RFTA before submitting...

[nwbvt]

Actually they do. But feel free to keep on your tin foild hat if you like, it looks good on you.

Re:If you were to RFTA before submitting...

[bhima]

If you'd spend a little time dredging around IBM's Blue Gene/L website you'd find that it is not specifically designed to any one thing and computing folding patterns of protein molecules is sort of their demo task (I suppose because it does take a lot of computational power and you wind up with pretty pictures)

I read a power point sales presentation and the plan seemed to be secure the top spot on the super computer list and then sell little versions to all sorts of folks.

Anyway I read several interesting applications that are already in the works like brain mapping and large scale radio telescope interferometry. So a lot of cool stuff is going to going on in the Blue Gene world.

Re:If you were to RFTA before submitting...

[nwbvt]

Yeah, working out "safety, security and reliability requirements" is different from "developing nuclear weapons".

Confused? Think about what the negative effects of radiation (a byproduct of nuclear weapons) on the human body are.

Guessing vs Gut feeling

[Taco Cowboy]

In this business, more than others, Gut feeling plays a leading role.

Financial Data Modelling is a fine idea, but the whole thing boils down to human psyche - and unless someone comes up with a perfect AI - one that is one step ahead in psycho term than human, - be it GT or OR or whatever else, market trend is very much based on butterfly effect + herd instinct + stochastic resonance with a whole lot of chaos effects thrown in.

That is why it's so dynamic !

Re:Guessing vs Gut feeling

[Anonymous Coward]

butterfly effect + herd instinct + stochastic resonance with a whole lot of chaos effects thrown in.

Wow, that's basically meaningless. Nice work!

Re:Guessing vs Gut feeling

[metlin]

Absolutely.

That's why a lot of these systems use such things as socio-cultural influences, press and media data and the like.

Unfortunately, the stock market is an area that is an ecosystem of its own -- preys and victims -- and therefore, predicting that is almost as hard as predicting human behaviour.

And ofcourse, the only reason the economies _thrive_ is because of the chaos - everyone would like to believe that they can leverage it to make a profit for themselves. :-)

And the best part is, everyone can, if they played their cards carefully enough.

Re:Guessing vs Gut feeling

[name773]

an ecosystem of its own -- preys and victims...And the best part is, everyone can, if they played their cards carefully enough.
how can everyone be making money? you already said there are preys & victims, and people have to be making that money from somewhere

Re:Guessing vs Gut feeling

[DAldredge]

Everyone does not refer to the investor. It refers to all the brokerage, exchanges and clearing houses involved in the trade. They all will make money under an 'ideal' system, the trader isn't a factor.

Re:Guessing vs Gut feeling

[NanoGator]

" market trend is very much based on butterfly effect + herd instinct + stochastic resonance with a whole lot of chaos effects thrown in."

Suddenly, a bunch of monkeys find work...

Re:Guessing vs Gut feeling

[jrexilius]

Wealth is usually created not stolen. There isnt an absolute limit on the amount of money or products in an economy and so it doesn't have to be based on "victims" and "predators".

Not to say that there aren't a lot of thieves on wall street but they generally are not the ones creating the wealth, just pooling money for others to create wealth with.

Re:Guessing vs Gut feeling

[name773]

if it were truly an ideal system, the investors would be a factor...

Speed of the computer?

[Avuton Olrich]

Well, I'm confused, how can they figure out the speed so easily, when it's so hard to test the difference in speed between x86 AMDs & Intels? The other computers aren't faster at some things? Is it some special bench?

Re:Speed of the computer?

[vspazv]

The test is called linpack.

http://www.top500.org/lists/linpack.php

Re:Speed of the computer?

[a3217055]

Actually the speed is a function of a linpack problem size with communication functionality with the different cpus. Each linpack for each supert computer is highly highly optimized. They have mathematicians hacking FORTRAN libraries trying to get the last bit of performance out of the system.

Re:Speed of the computer?

[raxx7]

You are right.

The Top500.org list rank is based on one particular benchmark: Linpack. But Linpack is a very particular benchmark and a poor measure of system performance. In fact, it's regarded more as a pissing contest than a meaningful benchmark.

In defense of the authors of Top500.org, they didn't intended to use Linpack for ranking, they just wanted a census of the biggest machines.

Re:Speed of the computer?

[NanoGator]

"Well, I'm confused, how can they figure out the speed so easily, when it's so hard to test the difference in speed between x86 AMDs & Intels? The other computers aren't faster at some things? Is it some special bench?"

I think you answered your own question. Notice your use of the phrase 'benchmark' is singular? These aren't general purpose machines. I'll put it another way: If you bought a PC for the sole purpose of playing Quake 3, then the Quake 3 demo would be the definitive and ONLY benchmark you'd need.

Of course, I could be talking out of my rear. I don't know much about these types of computers. So if I'm wrong about the 'single purpose' implication I made, I'm happy to recieve corrective info.

Re:Speed of the computer?

[forkazoo]

They basically use three benchmarks:

-Raw peak performance

-Linpack

-SPEC

None of these is automatically going to indicate how the machine compares to another machine at a certain task. But, the primary difference between guys who work with Big Iron, and the gamers is that the Big Iron guys know that the benchmarks in isolation don't mean anything. By and large, there will be a lot of special custom code that takes up the bulk of compute time on a big system, so your "special purpose" idea holds some water. If this is the case, they just check to see how their custom code runs on the system, what the bottlenecks are, etc.

SPEC is generally considered the best benchmark for general purpose stuff, but it takes a fair amount of digging to understand exactly whether or not it means anything to you. Since it's broken up into many parts, you may find that your own application tracks a certain subset of the benchmarks very closely due to similar code. Other types of code are small, and tight, and don't care about bandwidth, and will track the raw FLOPS number very closely.

Just Imagine...

[mr_don't]

"A Beowulf Cluster of these!!!"

What? No one has posted that already?

Re:Just Imagine...

[perhj]

You know, I had hoped that the days when you could score some easy karma by throwing in spurious Beowulf references were over. Obviously, I was mistaken.

Re:Just Imagine...

[Wesley Felter]

No no, BlueGene/L is a cluster, so you have to imagine a single node of it.

Re:Just Imagine...

[sp0rk173]

That's SOOOOOOoooOOooOOOOOOooOO 2002.

I defeated your girlfriend's firewall...

[Anonymous Coward]

"...I can't wait to throw out my girlfriend 1.0 once they finally come up with one that doesn't put up a inpenetrable firewall in bed...."

I defeated your girlfriend's firewall: I used her built-in back door.

That new computer is so fast

[Anonymous Coward]

it can process an infinite loop in under 3.8 microseconds.

that we know of......

[Anonymous Coward]

I would rather get a list of top secret fatest computers in the world.

Re:that we know of......

[NanoGator]

"I would rather get a list of top secret fatest computers in the world."

I'm surprised somebody didn't make a stale joke about any machine running Windows being the 'fatest' computer. Come to think of it, where are the Beowulf, Longhorn, or Doom 3 jokes?

Re:that we know of......

[TheLink]

Yeah, I wonder how fast google's machines really are.

I believe recently some guy at Google mentioned running something that took weeks of total CPU time but only a few minutes real time. Or something like that. Anyone got the reference?

They can probably make money just by figuring out what the "dumb sheep" "investors" are going to buy next (most of them are dumb sheep).

BlueGene/L get 70.72T/s

[myisaacxu]

The IBM BlueGene/L now get 70.72 T/S,and this is beta machine,it is only 1/4 finale.The more information ,please visitor the site:http://www.llnl.gov/llnl/06news/NewsReleases/ 2004/NR-04-11-01.html

PGP

[Lord Kano]

maybe I should get a stronger PGP key.

I'm using a 4096 bit key. Simply because it's not perceptable slower to use it for encryption but it takes more work to brute force it for decryption. Well, when compared to a 1024 bit key.

LK

Yes but does it run Linux like the SGI?

[msevior]

That really increases the SGI usefulness. Develop/Debug/setup on your laptop/desktop and ship to the Supercomputer for production.

This is a pretty useful thing I think.

Re:Yes but does it run Linux like the SGI?

[nchip]

Yes and no.

The Linux-based host nodes manage user interaction functions, while the Linux-based service nodes provide control and monitoring capabilities.

Linux is also used in I/O nodes, which provide a gigabit Ethernet connection to the outside world for each group of 64 compute nodes, or every 128 processors. Thus, the full BlueGene/L system will have 1024 I/O nodes, which essentially form a Linux cluster.

The actual compute nodes -- the 128,000 processors -- do not run Linux, but instead run a very simple operating system written from scratch by the Project's scientists.

Bluegen at over 70TF

[a3217055]

Bluegen hitting over 70TF is a tremendous effort. There is nothing that comes near. And this is only 16 racks ( 25% of the total system) of the 64 rack system.
Hats off to IBM for doing an outstanding job. And to the others in the race better luck next year.

Also this runs ppc chips what else do you want an Itanium/Opteron what you want radiation burns.....

PS I posted this on thursday night but the moronic slashdot editor threw it out. This is old news... Anyway... C'est la vie.

Re:Bluegen at over 70TF

[a3217055]

yeah do you know how many processors the NASA machine needs and also the ammount of power thost Itanium chips suck up. Also the Nasa machine has a scalability problem. It doesn't like doing certain configurations. This is what you get... so what if NASA's SGI system goes at 80TF. Bluegen will beat it by size, cooling, and cpu too.. Also the BlueGene machine has 2 FPU. take care

Re:ob

[a3217055]

actually it can play doom, it might not have the opengl but it can play doom3. It is based on a ppc architecture with a good linux ppc virtual machine you can paralellise Doom3 for BlueGene.

modeling cortical circuits

[grounded_roamer]

The Brain-Mind Institute http://bmi.epfl.ch/ at EPFL (Lausanne, Swiss) is about to aquire one for modeling cortical circuits (or brain modules). This requires simulation of 100,000 neurons, each one is composed of a couple of differential equations. You need a lot of computer power (both speed and memory) just to get results for 1 second of real time. What will be learn from this kind of project is still a open question ...

Is it needed?

[Ceriel Nosforit]

The final machine will help scientists work out the safety, security and reliability requirements for the US's nuclear weapons stockpile, without the need for underground nuclear testing.

Could someone explain to me why this task requires such a monster of a machine? And how can one address (as in write code for) the numerous unknowable factors that seems to be included in the problem that is to be solved? The definition just seems to be too abstract to be an actual solvable problem, and if it is solvable it would require an immense human resource contribution for the code it is to run. Wouldn't it be simpler to just stick those people into a room and not let them out until they've solved the problem?

I've long wondered who comes up with the code they run on these 'pooters. Anyone who can offer some insight on the usual complexity of the code that is run/problems that are solved?

Re:Is it needed?

[Anonymous Coward]

Could someone explain to me why this task requires such a monster of a machine?

They're simulating nuclear explosions. They need to solve monstrously complicated coupled nonlinear differential equations for hydrodynamics, at high mesh densities. They also need to simulate aging of stockpiled weapons, a complex three-dimensional problem in fault dynamics, diffusion, etc. These are among the most difficult numerical simulations computers have been asked to run.

And how can one address (as in write code for) the numerous unknowable factors that seems to be included in the problem that is to be solved?

Which "unknowable" factors are those? Remember, we already have the benefit of detailed studies from real tests, a solid knowledge of how the warheads were constructed, and we have all kinds of instruments that can inspect their current condition.

Wouldn't it be simpler to just stick those people into a room and not let them out until they've solved the problem?

How are they going to solve it without either real-life tests, or computer simulations?? Pencil and paper? Pure thought??

Quantum cracking algorithms?

[mikelang]

And what if they build a little quantum computer?

Re:Quantum cracking algorithms?

[EsbenMoseHansen]

Indeed. Not so long ago I heard that a team had succesfully factorized 12 (into 4*3) using a quantum computer. (It was a 7 qubit 'puter) :)

Quantum computer has a way to go, even by paranoid standards.

Re:Quantum cracking algorithms?

[fatphil]

15 = 5*3

FP.

Too late

[vile7707]

*snip*
" maybe I should get a stronger PGP key."

We've already calculated your next 250 pgp keys, and divined your future. Hint: avoid badgers.

More Power to the People

[YetAnotherGeekGuy]

Just imagine a whole Beowulf cluster of these babies!

When they talk about 5 nines in big iron, who knew it would be the percentage of time spend in the System Idle Process?

Re:More Power to the People

[bhima]

OK, I know you are just amusing yourself with a tired joke but you should have a look at IBM's Blue Gene site and read up on their networking stratagies (all three of them). It's really interesting and makes Beowulf clustering seem sort of... tired.

That and it's really low power... somwhere in the powerpoint presentations they have a graphic showing the Blue Gene/L uses a little less power than the same volume of IBM thinkpads.

Oh.. and given the costs involved I think idle time will wind up being sort of low.

applications

[photonic]

Machine number one will go to Livermore, probably for doing some nuclear stuf. Number two will go to the Netherlands for the Lofar project. This is a 300 kilometer diameter radio telescope that observes at low frequencies (up to 250 MHz). It constists of thousands of small antennas spread across half the country. Their signals will be interferometrically combined to form the images (compare e.g. to the VLA). Blue Gene will be used to combine all the signals in real time, I believe the total bandwidth from the antennas is some terabyte/sec.

Climate modeling

[LarsWestergren]

Ok, so of the three fastest computers in the world, one is almost exclusively dedicated to environmental climate models, and the other two have it as part of their tasks.

Perhaps this could bury the arguments on Slashdot that there is no hard data or serious research about global warming.

Re:Climate modeling

[HeghmoH]

Playing devil's advocate:

Just because people are running a bunch of simulations on climate change doesn't mean the results are useful. If people were running a bunch of simulations on the existence of dragons and fairies, I would hardly expect reasoning people to use that as evidence that they're real.

Re:Climate modeling

[hawkeye]

While I'm certainly an advocate of playing it safe with the environment, I do, however, understand the point(s) of those that don't subscribe to the global warming theory.

In truth, we don't have enough data, from our past, to understand whether our climatic changes are just brief glitches or undeniable trends.

Obligatory Redundancy

[Mystilleef]

No, John, I can't imagine a Beowulf cluster of those.

Top Supercomputer?

[dutt]

I really wonder if the list of the worlds fastest supercomputers is fully updated. I wouldn't be surprized if in the future, it is revealed that NSA had the fastest computer all along. I don't think it's too far fetched to say that the worlds fastest computers are used to crack encryptions. ... so yeah, I guess it's time to upgrade those PGP keys! :)

Re:Top Supercomputer?

[cbrocious]

There are sooo many better ways than brute-force cracking to figure out keys. The NSA specializes in finding these methods, not throwing hardware at the problem like many people do. If they want to crack a key, they'll do so.

Does Moore principle apply to quantum computing?

[mikelang]

The question is:

• how fast do they make progress in the number of qubits "per chip"?
• how expensive the "chip" is?

Re:Does Moore principle apply to quantum computing

[EsbenMoseHansen]

Currently, it isn't even a chip (or at least, last I heard). It was (a lot of) molekyles with 7 "mutable" spots (I think it was rotation). The state was read using NMR spectroscopy).

It is about as close to a chip as a printing press to a photocopier ;-)

It's only going to get faster

[pyite]

Coincidentally, as I was hiking through the woods yesterday, I was talking about Blue Gene with one of the guys who works on the project at IBM. Blue Gene is intended to go to 400 TFlops, and the 70 TFlops number is due to the second piece of the machine being put together. For a while Earth Simulator was running LINPACK fastest at like 36 TFlops. Then IBM came out with piece one of Blue Gene and ran it at 40 TFlops. Then NEC announced something running at like 46 TFlops. Looks like IBM is back on top. It should be a good race.

Fast vs. Powerful

[SimonShine]

But really, is this not a general misconception? Are computers' capacity measured by how fast they are and not by how powerful they are? After all, the unit Hz already has "per second" defined, and the system isn't running anywhere. :)

Re:Again?

[NanoGator]

"Slashdot's Previous story"

I agree. On another note, it really bugs me when they hold the Olympics every four years when the record was set a long time ago. Old news.

Re:Does Moore principle apply to quantum computing

[Haydn Fenton]

IIRC, for each qubit you add, the computing power doubles. So while we're not into the really powerful stuff yet, progress should be pretty rapid. I've dug up a little info from google and this seems to be a great place to learn as to how it all works. Because of superposition (read the link), "the number of computations that a quantum computer could undertake is 2^n, where n is the number of qubits used. A quantum computer comprised of 500 qubits would have a potential to do 2^500 calculations in a single step. This is an awesome number - 2^500 is infinitely more atoms than there are in the known universe (this is true parallel processing - classical computers today, even so called parallel processors, still only truly do one thing at a time: there are just two or more of them doing it)". 7 doesn't seem like much, but a few times more and you have an extremely powerful computer. As for the price, I haven't got a clue, Google it.

Re:Hm

[irc.goatse.cx troll]

Not fast enough

Re:The Sims?

[oneiron]

Haha! I like that one.

OR, maybe this is all just kind of really stupid talk and we should focus on really really cool and very fast super computers.

You might be on to something... Personally, I think the focus should be somewhere between the two. That's where I like to keep it, anyway.

slash who?

[chyllaxyn]

wow, old news.
I think Drudge actually broke this and I submited it to slashdot on Friday.
You guys aren't still sore about that comment I made about Slashdot being run by a bunch of liberal lozers...are you? .)
awwww, I was just gloating
,
,
,

Is it powerful enough?

[jbohanon]

...to tell us that the meaning of life is 42?

the virginia tech's apple must drop out of top ten

[cyfer2000]

Ya know, it is November and apples drop.

Re:Hm

[dtfinch]

According to the comment id's and times it appears I was.

Yey

[Ambient_Developer]

Yey for IBM. Now I can stop hearing about the earth simulator and how great it is

Re:good use?

[sp0rk173]

Personally, and this is just me, solving scientific problems involving proteins and whatnot is a bit more important than the inconvenience of spam. Again, that's just me.

IBM PGP key retrieval service -- free trial

[SysKoll]

Hello dshaw858,

Thank you for trying the IBM PGP key recovery service. Your are entitled to a free 6-month trial period.

As per your request, we analyzed your PGP-signed posts and deduced your passphare. Your current PGP passphrase is:
My b3st computer for a g1rlfriend!
(Strength: 38%)

We suggest you pick a less obvious PGP passphrase in the future.

Thank you,

-- The IBM BlueGene team.