US Regains Supercomputing Crown, Besting China and Japan 132
dcblogs writes "The U.S., once again, is home to the world's most powerful supercomputer after being knocked off the list by China two years ago and Japan last year. The top computer, an IBM system at the Department of Energy's Lawrence Livermore National Laboratory, is capable of 16.32 sustained petaflops, according to the Top 500 list, a global, twice a year ranking, released Monday. Despite the continuing strength of U.S. vendors globally, when China's supercomputer took the top position in June, 2010, it seemed to hit a national nerve. President Barack Obama mentioned China's top ranked supercomputer in two separate speeches, including his State of the Union address last year."
Conspiracy Nut (Score:5, Interesting)
Maybe, maybe not... (Score:5, Informative)
At least in the *specific* performance characteristic of 64bit precision linear algebra, it's perfectly likely that the biggest player is reported.
In the cases where secrecy is probably preventing you from knowing about it, it probably is optimized for 32-bit precision floating point and/or large storage throughput to fuel data mining.
Of course, then there are collections of systems that could probably easily place in the list that are at least moderately well-known but not submitted, if it wouldn't be a financial catastrophe to take it down for a few days to dedicate to an xhpl run. An EC2 datacenter comes to mind.
Re:Maybe, maybe not... (Score:5, Informative)
Or (in the case of NSA) decryption. There isn't a computer large enough to solve really difficult decryption problems, but whatever there is probably lives somewhere in the NSA, and is very likely very, very large.
Maybe not as large as Google's farm, though. Or even Amazon's.
rgb
Re:Maybe, maybe not... (Score:4, Informative)
Well, there's reports that NSA is building a massive spy center in Utah utilizing quantum computers [alexanderhiggins.com] but I'm not sure if I believe it...it reeks of tin-foil hats to me.
But, then again, it wouldn't really surprise me if it were true...I just didn't think quantum computers were anywhere near practical application yet, and wouldn't be for some time, in my extremely limited knowledge on the subject.
Re: (Score:3)
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Or rather, the correct application of brute force is to the individual holding the key -- waterboarding or threatening to remove vital body parts one at a time, that sort of thing
I do not like the use of the word "correct" in this context. Once torture becomes an accepted practice, you might as well go back to Medieval trial by torture instead of bothering with juries, lawyers and all that time-wasting shit.
This is true whether you are talking about terrorists, witches or children stealing a loaf of bread.
Re: (Score:2)
The point still stands, however. It's a lot easier and cheaper to hold a gun to somebody's head (figuratively or literally) than it is to crack any of the halfway decent encryptions. The legal and court system has recourse to direct coercion in the form of contempt of court and jail -- they can compel you to reveal keys or throw you in ja
Re: (Score:2)
Re: (Score:2, Insightful)
Each time I read a story like this I can't help think there are a bunch of faster machines that they don't tell us about.
Each time I read a story like this, I can't help but wonder who gives a fuck. For quite some time now, computers like this have been budget constrained more than anything, so it's a pointless dick measuring contest.
Re: (Score:2, Funny)
I don't see how that would matter. I'm pretty sure some guy from Uganda would win with or without including the point.
Re: (Score:2)
Dr. Frederick Frankenstein: For the experiment to be a success, all of the body parts must be enlarged.
Inga: In other vords: his veins, his feet, his hands, his organs vould all have to be increased in size.
Dr. Frederick Frankenstein: Exactly.
Inga: He vould have an enormous schwanzstucker.
Dr. Frederick Frankenstein: That goes without saying.
Inga: Voof.
Igor: He's going to be very popular with the ladies.
Re: (Score:2)
Each time I read a story like this, I can't help but wonder why anyone who doesn't give a fuck even reads slashdot, let alone clicks on the story.
Today's building-sized supercomputer is tomorrow's smart phone. You'll have one of these babies on your desk in less than twenty years. Reason enough to gove a fuck?
Re: (Score:2)
Today's building-sized supercomputer is tomorrow's smart phone. You'll have one of these babies on your desk in less than twenty years. Reason enough to gove a fuck?
Not really. You are assuming that Moore's law will keep applying in future, which I understand is unlikely.
Re: (Score:2)
You are assuming that Moore's law will keep applying in future, which I understand is unlikely
Ten years ago I would have agreed with you, as they were getting close to the physical limits, but they found new ways of getting around those limits. I have no reason to believe they won't find ways around whatever limits they're coming across now.
Re: (Score:3)
The NSA probably gave them this little gem after it had passed its useful life.
Re:Conspiracy Nut (Score:5, Insightful)
Here is a story outside of IT. There was a Chinese Pianist, he was very good, he always won Piano competitions. He went to America to work with some of the masters. They offered him a choice, does he want to become a better Pianist, or just keep winning awards. Being told these options really humbled him, in his mind, winning meant that you were the best. However it was a case in these competitions there were only a subset of skills that are measured, so if you just focused on what was graded then you can win the competition. However if you want to master your art, it is about working on other things as well, things that are not always part of the grading.
Re: (Score:1)
Your comment is insightful, but reminds me of an old joke.
A guy walks into a bar with a box, which when opened was shown to contain a very tiny man playing a very tiny piano. The bartender says "that's pretty cool, how did you get that?"
"Well," says the patron, "I was walking down the beach and found a bottle, and when I opened it a genie came out and said in appreciation for freeing him, he would grant me a wish. So I wished for a twelve inch penis and got this guy.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Finally you have the megawatt power demand...
http://www.wired.com/threatlevel/2012/03/ff_nsadatacenter/all/1 [wired.com]
If you where really smart you could map out the voice, telex, fax, data bandwidth per landing/sat per year/decade and what kind of data the US sorted in real time.
Voice to text, text to dictionary, match and keep or dump.
Re: (Score:1)
Re: (Score:3)
Wrong prize (Score:5, Insightful)
I'd rather have a big fraction of our workforce be highly competent in mathematics, than have a computer that's marginally faster than any other.
One wins a pointless pissing match, the other provides a much more solid basis for real strength and prosperity.
Besides, all this really shows is that China will lend us enough money for us to buy computer components built an assembled throughout the world.
Re:Wrong prize (Score:5, Insightful)
I'd rather have a big fraction of our workforce be highly competent in mathematics
You can build a more powerful supercomputer, but you can't educate your workforce beyond their intelligence. Especially, when they have no desire to be educated.
You can overclock a processor. I haven't seen this successfully tested on humans yet. The results of own experiments with my Tesla Tower and the neighborhood kids were rather unpleasant.
Re: (Score:3)
I'll bet you forgot to first bathe them in liquid nitrogen. Don't feel bad - common newbie mistake.
Re: (Score:2)
The top 25% do all the math anyhow. The USA's top 25% matches anybodies, but that is never reported. What is always reported is average or bottom 25% stats as they are scary numbers.
Re: (Score:2)
Overclocking humans is fairly easy. How many programmers do you see that don't have a mountain dew on their desk?
Re: (Score:2)
you can't educate your workforce beyond their intelligence. Especially, when they have no desire to be educated.
You need to make up your mind whether the plebs are stupid or just lazy. Such equivocation doesn't become a true fascist.
Re: (Score:2)
I think the supercomputer race is a very big deal. Even in a land whose people are marginally better at mathematics, they still just want to throw their problems at a faster computer. People who use math more will require quadratically or exponentially more computer speed because they want to solve more difficult problems not now but right now. They're not like, Gee I know math really well so I'm taking the day off, while people who don't know math need to use their electronic brains to do the thinking.
A co
Re: (Score:2)
Re: (Score:1)
I don't normally respond to AC"s, but I'm intrigued by your response.
In what way did my post demonstrate a staggeringly poor understanding of economics?
Also, I disagree about the "marginal" claim. At least according to this source [singularity.com], the computing power of top-end supercomputers doubles about every 1.2 years, and hardly a year goes by without a new super-computer coming online. So a 60% increase in performance is likely to keep the crown for no more than a year or two.
Re: (Score:2)
Re: (Score:2)
No, I think you're misunderstanding my point.
I call a 60% increase in computing power marginal because of the amount of time it is likely to hold the crown. I apologize for not being clearer about that.
Regarding the China funding, here you go [treasury.gov], and also here [wikipedia.org]. Also, this is a year in which about 1/3 [wikipedia.org] of the Federal government's funding came from deficit spending (i.e., bond sales).
I was not claiming the China and only China paid fro Sequoia. I was pointing out that it a valid symbol of neither or technical
Re: (Score:3)
China more or less stopped buying US bonds a couple of years ago.
Now the US federal reserve buys any 'leftover' bonds at all the auctions, same as Europe's central bank does.
There is no 'market rate' for government interest. Only the central banks 'target rate' which are politically set.
We're all fucked. Buy hard assets.
Re: (Score:2)
In what way did my post demonstrate a staggeringly poor understanding of economics?
GP is probably an extreme right winger who thinks there is no point in educating the workforce in the US/West as you can always hire them cheaper in China. As Oscar Wilde said, a cynic is someone who knows the price of everything and the value of nothing.
... WITH 100% CHINESE-SOURCED COMPONENTS !! (Score:5, Funny)
Way to go !!
Re: (Score:2)
Re: (Score:2)
With Chinese sourced money!
Re:... WITH 100% CHINESE-SOURCED COMPONENTS !! (Score:5, Insightful)
Re: (Score:2)
But how far back to we go for giving design credit? To the Arabs, Persians, Greeks, or Egyptians for the fundamental mathematics used? My point is just that even if Intel pioneered this technology, it doesn't strongly imply much about the U.S.'s current prowess.
Also, I know Intel has chip design facilities at least in Israel. Not sure where else.
Re: (Score:2)
Re: (Score:2)
It's the football fallacy. "My towns team won a football match so my entire town must be good, and me too."
It's more like "my country's team won a football match so I can enjoy a few hours of drunken pleasure before the next round of disappointment and defeat begin."
This is English football (soccer) I'm talking about. We're playing tonight.
Re: (Score:3)
Intel has chip design facilities in the US and Israel. They only have two of them, at least for desktop processors and they alternate, so one designed whatever the current product is, one is working on the next. Part of how they keep their tick-tock process going.
Fabrication is mostly in the US. They do have one fab in Ireland, one in Israel (which is the latest 22nm tech) and a new one in China (which is much older 65nm tech) but the other 6 are in the US, with a 7th being built for 14nm in Chandler, AZ.
Re: (Score:2)
Also, this is a Blue Gene, meaning IBM, not Intel fabs.
Re: (Score:2)
This uses IBM BlueGene/Q - PowerPC A2 chips ... not made by or designed by Intel ....
Intel or POWER? (Score:2)
Was this supercomputer based on either Xeons or Itaniums, or was it based on POWER7? Does IBM fab POWER7 in China? I thought that it was all in fabs in the US, since they don't make too many of those.
On a separate note, since Intel hardly sells many Itaniums, it could take all the chips it could get out of 1 lot, and make a supercomputer based on that. Something like SGI's Altix.
Re: (Score:1)
So you'd prefer the components to be made here?
Feel like paying $3,000 for that Dell?
Because that's how much it would cost to make a PC if the factories that make the parts were in the US, dealing with US Environmental Regulations, US Taxes and having to pay US Workers who expect to be able to afford a McMansion and Porsche Cayenne for working in any aspect in the Tech Sector and would likely Unionize overnight.
The Days of Made in the USA are long over -- it's all about Globalization. The day Manufacturing
Re: (Score:2)
I think many of them would settle for a living wage (where I live that would be about fifteen bucks an hour) plus health care benefits. As far environmental regulations go, Chinese e
Re: (Score:2)
Funny, I seem to recall Dell doing much better before they moved their manufacturing to China.
Re: (Score:2)
Dell doesn't make the parts -- and it's the parts I was talking about. HDDs are made in Thailand and Indonesia, CPUs are made in Ireland and China, Mainboards and RAM are made abroad as well. These components contain heavy metals and require the use of toxic material to produce and as such, the environmental compliance costs, OSHA compliance, insurance fees for facility and workers and the inevitable EPA fines make it cost prohibitive.
Re: (Score:2)
Dell doesn't make the parts -- and it's the parts I was talking about.
Not a very good example then.
Oh and if you think EPA fines are bad, can you imagine how bad it would be if a company poisoned an entire town? EPA fines are nothing compared to a serious clean-up.
China is polluting like mad, they will have to start dealing with clean-ups very soon and it will severely effect their economy.
Actually... (Score:2)
This was a BlueGene, meaning in all likelihood the processors and circuit boards were manufactured in New York. There were probably memory,storage, and other components sourced from southeast asia, but the most expensive bits probably were actually fabricated domestic.
Re: (Score:1)
Re: (Score:2)
No, it is not an Intel based supercomputer, it is POWER based. The processors were manufactured in Fishkill, NY, and the system was assembled in Rochester, MN.
Moore's Law (Score:2)
Re:Moore's Law (Score:5, Funny)
Moore wasn't an IBM employee, so Turek hasn't really heard that much about him.
Re: (Score:2)
Well, it used to be that the increased transistor density predicted by Moore's prediction (not a law in any sense) lead to faster processors. That has not been true for quite a while. So the guy is 100% right when he says you can't just wait for Moore's law to lead to faster processors. Today, the increased density is used to make more, not faster, processors. Using all those processors effectively is the challenge.
And yes, the limits of Moore's law will be hit in the relatively near future, possibly wi
They already have a practical use for the champ (Score:1)
Processing petabytes of clickstreams yielded from third party cookies to better identify hidden cross-selling and upselling opportunities.
Just curious: practical applications? (Score:2)
Re: (Score:2)
Re: (Score:3)
42
Re: (Score:2)
Well, this one is intended for simulating nuclear weaponry. Basically testing nukes in a way that doesn't involve nuking ourselves.
Re: (Score:2)
A Short List of Applications (Score:2)
Currently material scientists are building
MFLOD? (Score:2)
Here is one example our chair is working on: simulation of dendritic growth. Ever heard of that? Doesn't sound particularly relevant to your everyday life? Well, it is. Material scientist are interested in understanding how crystalline structures form in cooling metal alloys, The crytal structure is ke to building stronger, lighter metal work pieces. Ultimately a solid understanding of this will lead to e.g. higher fuel efficiency in jet planes and lighter cars.
Currently material scientists are building computational models of these processes. To check whether a new models works out we need to simulate it. This takes Terabytes of RAM and Petabytes of disk space. That's what such a machine is good for.
Here is a list of other flagship applications. Most of them are simulation codes that replace experiments which would either be too costly, happen under too extreme physical stress/forces, or simply could not be carried out at all in practice because of scale.
Yeah, I worry about dendritic growth all the time, and tin-finger growth. That's a big concern too; much more so now that there's no lead in most of my electronics and they've changed my fluxes and cleaning processes. (Effing ROHS!) But dendritic and tin-finger growth are easily tested with real world models as opposed to simulations and I always trust real world models more than simulations. It's too easy for a tiny error in a mathematical model to cause big errors in results. Numerical simulations ar
Re: (Score:2)
The MegaFLOpD metric you propose is interesting, but also limited. A $700 dual-GPU gaming rig will always be more cost efficient than an HPC system with similar performance, and cost efficiency gets worse the larger the machine gets (the K computer was $1.2 Billion IIRC). And yet these machines are sometimes the only ones that can run such simulations.
In essence, I don't think there is a single sufficient metric. PetaFLOPS are one thing, FLOPS/$ are the other. Both have to be kept in mind today already.
Too bad... (Score:3)
The fastest computer in the world is useless if it's not being used for a productive purpose.
Re: (Score:1)
Re: (Score:2)
Think of it - massively parallel virtual Bonnaroo.
Re: (Score:3)
Re: (Score:2)
Exactly. Do you have any idea how many World of Warcraft bots I could be running on one of those farming me gold?
Re: (Score:2)
The point is that the little guy is not going to "invest" in a faster computer until government shows that the faster computer is actually a game changer. The government is the economic guinea pig.
Re: (Score:1)
Re: (Score:1)
LLNL is run by the University of California system? Is that why the URL is http://www.llnl.gov/ [llnl.gov] and the front page clearly states that it is run by the DOE for the National Nuclear Safety Administration?
It does not say that it is "run by the DOE." You made that up, or are illiterate. It says it is run for the DOE:
Operated by Lawrence Livermore National Security, LLC, for the Department of Energy's National Nuclear Security Administration
And it says that because the DOE owns the lab, which is operated by contract:
https://www.llnl.gov/about/mgtsponsors.html [llnl.gov]
The LLNS management team includes Bechtel National, University of California, Babcock and Wilcox, Washington Division of URS Corporation, and Battelle.
Government Owned, Contractor Operated. The Laboratory is a government-owned, contractor-operated (GOCO) facility managed through a contract between the LLNS Board of Governors and the DOE’s National Nuclear Security Administration (NNSA).
(Emphasis theirs.)
In case you're wondering, here's what LLNS, LLC (http://www.llnsllc.com/ [llnsllc.com]) has to say for itself:
Our team includes Bechtel National, University of California, Babcock and Wilcox, the Washington Division of URS Corporation, and Battelle. Our cutting-edge science is enhanced through the expertise of the University of California and its ten campuses and our team's affiliation with The Texas A&M University System.
(Emphasis mine.)
You may also find the History section [llnl.gov] enlightening.
PETAflops (Score:1)
and to meet peta meta-requirements, no cute kittens or bunnies were flopped or otherwise harmed in the building of this superb computer.
Interesting thing (Score:3)
Just a little interesting thing I noticed while reading up on this.
Most American media refers to it as an "American" supercomputer first and foremost, and an "IBM" supercomputer second.
Most non-American media refers to it as an "IBM" supercomputer first, and an "American" supercomputer second.
Not really wrong either way - it's a big win for both the US and for IBM - but it's interesting to see the little differences.
What happens to the "obsolete" supercomputers? (Score:3)
Seriously, if they are migrating to new supercomputers, what, each year or two, what are they doing with the "scraps"? Do they refurbish them to be used elsewhere? do they sell them as "used goods" to other companies?
I can imagine taking quite a bit of time to install/uninstall such installations, so if its going to stay 2 years and it takes 3 months to setup/remove, are the installation/uninstallation crews constantly on site? or maybe it takes much less time to do so?
does anyone has any idea on this? I mean, that's a lot of stuff to "just" throw away when upgrading. Something useful must be done with the "obsolete" equipment, no?
Re:What happens to the "obsolete" supercomputers? (Score:5, Interesting)
Re: (Score:2)
Great! an even more specific response. This is why I love /., at somepoint *someone* has been there, or at least close enough, to give a realistic answer. If I had mod points (and hadn't posted yet) I'd give you some :)
Re: (Score:2)
Most of the obsolete supercomputers are broken up an reused, or in a few cases sent to museums ...
Re: (Score:1)
At least when my university (unam.mx) buys a computer and their lifetime ends, technicians come and scrap all the electronics leaving an empty shell. As in the case of software it seems you don't own the computer, you're only leasing the technology, this may vary with technology developed with money from the US goverment.
Re: (Score:2)
Usually they keep the Newest-1 machine around for a little while so that users can slowly migrate over to the new machine. Even with subsequent generations of supercomputers from the same manufacturer, there are often little difference that make migration tricky. Once most users have migrated, they generally decommission the machine and scrap it. Used machines just aren't worth running. They are big and heavy to move, require so much power that you need lots of expensive electrical work to power, and HVAC i
Re: (Score:2)
Wow, so they keep adding space to their facilities to add more and more supercomputers? it figures that they use it at least that long, and of course i would guess that lower priority projects get moved to the "obsolete" machines, as the big guns are moved to the newest and latest, but at some point you have to discard them. SO are they absolutely of no use after 10 years or so?
Thanks for the answer AC, I really had no idea about the life span of one of these monsters.
Re: (Score:2)
I did the math a while ago - a 20-year-old supercomputer is about as powerful as a modern 1U server, or maybe a really high-end workstation (at least going by Dhrystone benchmarks). But still using the supercomputer-sized room and supercomputer-sized power consumption.
I can't be assed to do the precise research, but I would estimate that a 10-year-old supercomputer would be about as powerful as a small cluster. But, again, a small cluster that sucks down several megawatts and takes up a small building. Ther
Holy shit - BOTH #1 and #2 are RISC-based! (Score:2)
I just now learned that Sequoia is based on POWER CPUs. And #2 I already knew is based on SPARC.
This doesn't mean much for RISC, perhaps, but at least some bragging rights.
Re: (Score:3)
RISC and CISC are no longer really meaningful distinctions. The new POWER systems have a lot of non-RISCy SIMD extensions and such, as do the SPARCs used in the K supercomputer. Likewise, the modern CISC processors (read: x86) have become more RISC-like, at least internally.
I actually think the modern x86 way is a really good compromise (although the x86 implementation of the idea ranges from terrible to tolerable). Internally, use a simple, efficient RISC-like micro-op processor, while presenting a CISC-li
Re: (Score:2)
The problem with that translator is, it's always working. Many parts of the CPU will be idle at any time or the other, but the translator, as small as it is (not too small), it'll always consume some power. That is a very unambiguous disadvantage of CISC-outerly archs, while we might argue about the rest.
Re: (Score:3)
You commit several severe logical errors during this (to say nothing of your grammar). Let's look at it in depth.
Your first fallacy is assuming all cores are equal. Notably, that PowerPC A2 is running at half, maybe even a third, the clock speed of the i7. That gives it much better efficiency, yes - but it also gives much higher latency. In a supercomputer processor, that's no problem - throughput is the only important factor. But for a desktop? Latency matters.
Had you read more, you would also know that on
so wat (Score:2)
something else (Score:2)
Oh the irony.. (Score:1)
Re: (Score:2)
The National Nuclear Security Administration has little to do with energy efficiency, at least in the sense you're referring to. However, from the Top500 press release: "Sequoia is also one of the most energy efficient systems on the list".
Re: (Score:1)
Re: (Score:2)
It gets interesting down the list too... SuperMUC is a Sandy Bridge install without GPUs, gets more performance than Tianhe also at lower power. One of the big big things about GPGPU was drastically better performance per watt, but SuperMUC beating Tianhe is really interesting. I really expected a stronger GPGPU showing, but it just isn't happening...
China and/or India will regain it within a year. (Score:2)