Japan computer smashes speed record

Nov 04, 2011
This picture, released by Riken, shows the "K computer" at Riken's laboratory in Kobe, Hyogo prefecture, western Japan. The supercomputer has broken its own record as the world's fastest machine by performing 10 quadrillion calculations per second, its developers announced.

A Japanese supercomputer has broken its own record as the world's fastest machine by performing 10 quadrillion calculations per second, its developers announced.

The "K Computer", which has more than 88,000 central -- the computer's "brain" -- compared with the fewer than four in the average desktop, smashed its own record of just over 8 quadrillion calculations during an experiment in October. A quadrillion is a thousand trillion.

Supercomputers operate roughly 10,000 times faster than ordinary personal computers.

Among other things, supercomputers are used by scientists investigating the effects of seismic waves and tsunamis to predict what effect they may have on buildings.

Ryoji Noyori, president of Riken, the company that jointly developed the machine alongside Fujitsu, said it was a landmark achievement in computing.

"The K Computer is a key national technology that will help lay the foundation for Japan's further progress," he said on Wednesday.

"I am delighted that it has achieved its major objective, demonstrating our strong technical power."

Explore further: World's largest 4K video board

add to favorites email to friend print save as pdf

Related Stories

Fujitsu develops world's fastest processor

May 15, 2009

(AP) -- Japanese computer maker Fujitsu Ltd. said Wednesday that it has successfully developed the world's fastest supercomputer processing unit with more than twice the speed of the current leader.

ORNL Jaguar supercomputer surpasses 50 teraflops

Aug 25, 2006

An upgrade to the Cray XT3 supercomputer at Oak Ridge National Laboratory, the most powerful supercomputer available for general scientific research in the United States, has increased the system's computing ...

Recommended for you

Review: 'Hearthstone' card game is the real deal

1 hour ago

Video game publishers don't take many risks with their most popular franchises. You know exactly what you are going to get from a new "Call of Duty" or "Madden NFL" game—it will probably be pretty good, ...

Switch on sunlight for a brighter future

2 hours ago

Imagine sitting in a windowless room yet having the feeling of the sun shining on your face. This unique experience is now possible thanks to the COELUX EU-funded project which recreates the physical and ...

Tackling urban problems with Big Data

2 hours ago

Paul Waddell, a city planning professor at the University of California, Berkeley, with a penchant for conducting research with what he calls his "big urban data," is putting his work to a real-world test ...

User comments : 44

Adjust slider to filter visible comments by rank

Display comments: newest first

Silan
4.2 / 5 (10) Nov 04, 2011
Sounds like this article was written for children, not a science website.
Torny
1 / 5 (1) Nov 04, 2011
how many hertz are 10 quadrillion calculations per second.

quadrillion is 10^24 or 10^15 ??
LVT
1 / 5 (2) Nov 04, 2011
90 Billion (Giga?)Flops per core? A bit less than an unclocked Intel 980X.
antialias_physorg
4 / 5 (6) Nov 04, 2011
how many hertz are 10 quadrillion calculations per second.

Since it's massively parallel that question doesn't make much sense.
Koen
1.1 / 5 (15) Nov 04, 2011
Japan should close down all its nuclear powerplants, large parts of Japan are radioactive now (just another record breaking nuclear disaster). We know what devastating effect a tsunami had on a nuclear powerplant, and NO supercomputer (no matter how many zillion flops/sec) would have been able to "calculate" this event.

Use this "K" computer for solar cell efficiency optimization and cost reduction of solar cell production.
LVT
2.6 / 5 (20) Nov 04, 2011
Japan should close down all its nuclear powerplants, large parts of Japan are radioactive now (just another record breaking nuclear disaster). We know what devastating effect a tsunami had on a nuclear powerplant, and NO supercomputer (no matter how many zillion flops/sec) would have been able to "calculate" this event.

Use this "K" computer for solar cell efficiency optimization and cost reduction of solar cell production.


Killed a lot less people than German organic food did. I think we should ban organic food first, then look at the lower risks from nuclear.

http://www.washin...agebreak
Eikka
4.3 / 5 (3) Nov 04, 2011

Use this "K" computer for solar cell efficiency optimization and cost reduction of solar cell production.


Better yet, use it to come up with a cheap, scalable, efficient battery in the megawatt-hour range to make existing solar panels actually useful.

Without batteries, there's no amount of solar power that will power your home when the sun don't shine.

To give you a hit of the scale of the problem at hand, the average Japanese will use roughly 8500 kWh each year, which is equivalent to one full charge of a Nissan Leaf, per day, per person.

And you only have about four hours each day when the solar panels work their best. The rest of the time the power has to come from somewhere else.
zz6549
5 / 5 (1) Nov 04, 2011
You know, it's just a really powerful calculator. It's not going to solve Japan's energy problem by itself. If you wanted to use it for solar research, you'd need a program that could find an optimized solar cell structure. However, there's so many parameters involved with this that the computing part is a small portion of the work.

These computers are best at simulating physical processes for which the parameters are already somewhat known. If there was a solar cell design that needed to be tested, it could help with that. However, given that solar cells are relatively quite simple, those kinds of simulations could be done a desktop. Nuclear research would be a much better candidate for simulation. If Japan invested in a new line of safe, efficient reactors (TWR's), they could solve their energy problems for the next century.
flashgordon
not rated yet Nov 04, 2011
where's blue waters? They said they'd have their at least part photonic supercomputer up and running by last summer!

http://en.wikiped...e_Waters

Well, that wiki does mention some kind of problem of funding.
Parsec
not rated yet Nov 04, 2011
90 Billion (Giga?)Flops per core? A bit less than an unclocked Intel 980X.

1000 teraflops is a LOT more than 90 gigaflops. What the devil is an unclocked computer? I know there were experimental asynchronous machines built in the 1960's, which failed, but all modern computers are clocked. Did you mean overclocked by chance?
CHollman82
1.3 / 5 (9) Nov 04, 2011
large parts of Japan are radioactive now


Wow...

We know what devastating effect a tsunami had on a nuclear powerplant, and NO supercomputer (no matter how many zillion flops/sec) would have been able to "calculate" this event.


What kind of a point do you think you are making here? I'm 90% certain it is either non-existent or ridiculous.

quadrillion is 10^24 or 10^15 ??


10^15... 10^24 is also known as quadrillion but less frequently used and would be too large a jump to be believable.
kaasinees
2.3 / 5 (3) Nov 04, 2011
We know what devastating effect a tsunami had on a nuclear powerplant,

Most damage was done by preceding earthquakes, the tsunami might have accelerated the problem by only a tiny bit, the building was build for warding against tsunamis, the tsunami alone would have done no damage at all. In fact the tsunami might have helped the fuel rods from melting down in the first place.

----------

Killed a lot less people than German organic food did

The article was propaganda, that food did not come from Germany, it came from Spain and the seeds from the feed were imported from Greece. It was a european problem, not just Germany. The seeds from Greece were infected with bacteria.

----------

We need more comparable data from this article, How much flops per watt? How much flops per surface area?

I can put two supercomputers together into one project and claim its the fastest "supercomputer" in the world.
CHollman82
1.9 / 5 (9) Nov 04, 2011
The article sounds like it was written for simple.wikipedia.org
Nerdyguy
4.5 / 5 (8) Nov 04, 2011
AFP wrote this. A French version of AP/UPI. How would I know something this trivial? Because these are the same yahoos who wrote that "Elvis in the Testicles" article the other day. It was so bad, I actually looked up the source. hehe
Vendicar_Decarian
0.8 / 5 (42) Nov 04, 2011
"Killed a lot less people than German organic food did." - LVT

Poor LVT, he is reduced to posting blithering nonsense from the Washignton Times. The money losing newspaper run by the Loonie Moonies.

When it comes to news and opinion given a choice between the Moonie times and the National Enquiror, it would be wise to select the National Enquiror as it has a better track record.

But LVT is a Consrvative Libertarian, and he will post any lie as long as it advances his sick political ideology.
Vendicar_Decarian
0.9 / 5 (43) Nov 04, 2011
How do we know that LVT is lying by referring to a source - the Washington (Moonie) Times - that is itself lying?

The Moonies claim that the European E-coli outbreak was a result of bean sprouts grown at an organic farm.

But from a reputable source - CBC Canada - we read...

"German officials announced Monday that initial tests show sprouts from an organic farm in the country's north are not the cause of the E. coli outbreak as first suspected." - CBC

http://www.cbc.ca...any.html

Now the CBC story appeared on June 6, 2011.
The Moonies false claim was printed on June 8, 2011.

So the Moonies persisted on blaming organic bean sprouts from a German farm days after those sprouts were tested and found to be the source of the e-coli.

I have NEVER encountered a Libertarian/Randite who wasn't a perpetual and congenital liar.
Vendicar_Decarian
0.9 / 5 (44) Nov 04, 2011
"Without batteries, there's no amount of solar power that will power your home when the sun don't shine." - Ekka

Just pump it into the grid.
Vendicar_Decarian
0.9 / 5 (44) Nov 04, 2011
Re: "Unclocked"

Like most Libertarian/Randites LVT isn't very bright, and he intended to say "unlocked" rather than "unclocked".

But even if he had used the word "unlocked", his comment about 90 billion gigaflops per core shows that he lives in the city of TardPrime on the imaginary planet of Libertaria.
Vendicar_Decarian
1.1 / 5 (43) Nov 04, 2011
A quadrillion is 10 ** 15 flops or 10 ** 6 gigaflops.

This makes the Japanese computer at 20 * 10 ** 6 gigaflops about 8 times faster then the next on the top 50 supercomputer list, which comes in at 2.5 * 10 ** 6 gigaflops and is located in Tianjin China.

Vendicar_Decarian
1 / 5 (44) Nov 04, 2011
I don't tolerate chronic Liars.

Why do you?
Kafpauzo
not rated yet Nov 05, 2011
10^15... 10^24 is also known as quadrillion but less frequently used and would be too large a jump to be believable.


To avoid confusion we should make the words distinct. One system would use the words million, abillion, atrillion, aquadrillion etc., and the other system would use million, milliard, ebillion, billiard, etrillion, trilliard etc.

To make the forms easy to remember, the a indicates the American usage and the e indicates the European usage. Not that the usage is necessarily uniform on each continent, but it's distinct enough to make the letters easy to remember. I'm sorry if other continents feel left out, but all continents except Europe begin with an A, and, really, the letter isn't supposed to mean anything, it's just chosen as a memory help.

For consistency we might say emilliard rather than milliard, etrilliard rather than trilliard, etc. Here it's not needed for disambiguation, but maybe it sound better when it's consistent.
Vendicar_Decarian
0.8 / 5 (40) Nov 05, 2011
Why reinvent what science has already decided upon?

SI prefix list.

http://www.unitar...prefixes

I'm Yotta here.
CHollman82
1 / 5 (6) Nov 05, 2011
Why reinvent what science has already decided upon?

SI prefix list.

http://www.unitar...prefixes

I'm Yotta here.


You do notice that there is both a "long scale" and a "short scale" defined there right? That is what we are talking about...
Torny
not rated yet Nov 05, 2011
how many hertz are 10 quadrillion calculations per second.

Since it's massively parallel that question doesn't make much sense.


I meant, what is quadrillion :))
I know what is hertz :P
TheGhostofOtto1923
1 / 5 (1) Nov 05, 2011
Most damage was done by preceding earthquakes, the tsunami might have accelerated the problem by only a tiny bit, the building was build for warding against tsunamis, the tsunami alone would have done no damage at all. In fact the tsunami might have helped the fuel rods from melting down in the first place.
?? The tsunami flooded the bldgs and took out the backup generators which would have prevented overheating. However the cores and spent fuel pools were not accessible by floodwater.

You like to make stuff up just so you have something to post?
CHollman82
1 / 5 (6) Nov 05, 2011
how many hertz are 10 quadrillion calculations per second.

Since it's massively parallel that question doesn't make much sense.


I meant, what is quadrillion :))
I know what is hertz :P


Yeah, but he meant that clock frequency is an almost irrelevant measure of performance when you're talking about SMP systems. The frequency of each processor core is only a small determining factor in overall performance of the machine... the most significant factor is efficiency and organization of the data bus network that facilitates communication between each processor and each bank of memory.
Kafpauzo
not rated yet Nov 05, 2011
Hertz is a measure of oscillations per second. Oscillations per second and calculations per second are quite different things, you can't make any direct conversion. The processor clock inputs oscillate at one frequency, the internal clocks have other frequencies, each core makes several calculations simultaneously during each internal clock cycle, each calculation takes several clock cycles, there are more than 88,000 cores working in parallel... Disjoint concepts.

As for quadrillion:

1,000,000 = one million
1,000,000,000 = one abillion
1,000,000,000,000 = one atrillion
1,000,000,000,000,000 = one aquadrillion
1,000,000,000,000,000,000 = one aquintillion
...etc...

1,000,000 = one million
1,000,000,000 = one emilliard
1,000,000,000,000 = one ebillion
1,000,000,000,000,000 = one ebilliard
1,000,000,000,000,000,000 = one etrillion
...etc...

But the a- and e- prefixes (or prefices if you like your words nerdy) are my own invention and shouldn't be considered standard. Not yet. :-)
Vendicar_Decarian
0.8 / 5 (40) Nov 05, 2011
"That is what we are talking about..." - Choll

That isn't what the post I was responding to was talking about.

Please try and keep up.

As to the discussion you are referring to.

I refer you to the same link which makes any "discussion" pointless.

http://www.unitar...prefixes
antialias_physorg
not rated yet Nov 05, 2011

I meant, what is quadrillion :))
I know what is hertz :P


A quadrillion is 10 to the 24th. This shoul however not be construed to mean that there is a chip (or system) capable of making 1 calculation in 1 over 10 to the 24th seconds.

The individual chips are much slower than that (by a factor of more than 88000 times the number of cores in each unit).

Hertz is a measure of frequency (as you know). But the computer is not running at 1 over 10 to the 24th Hertz because of the massive paralleism.

What I meant was: If two people tap on a table at 1hz each then the 'tapping frequency of the two-people system' is not 2Hz.
CHollman82
1 / 5 (6) Nov 06, 2011
"That is what we are talking about..." - Choll

That isn't what the post I was responding to was talking about.

Please try and keep up.

As to the discussion you are referring to.

I refer you to the same link which makes any "discussion" pointless.

http://www.unitar...prefixes


/scratches head...

You're link shows the difference between the long and the short scales... that is what we all were talking about... I think you missed the boat somewhere along the line. Yes, SI prefixes are what they are, but the long scale is what it is as well, as is the short scale... you're just suggesting we all agree to use one of the three... well great, big deal, I'd like it if all humans spoke the same language and had kilometer/hour speedometers in their car as well, what's you're point?
bluehigh
3.4 / 5 (10) Nov 06, 2011
Back to your picture book CHollman, this is all a bit too hard for you to cope with until you grow up. Try not to use the term 'we' when you are speaking for yourself. I'd like it if mentally deficient people like yourself, were culled out at birth as a priority over a common language but hey .. maybe this new super fast computer will find a way to solve these problems.

Hopefully this supercomputer is used for reasons other than guessing answers (or questions) for TV game shows or playing chess.

Buyck
not rated yet Nov 06, 2011
Ok 10 Petaflops achieved... by a couyple of months we will hear 20 to 50 Petaflops. By that i mean "Sequoia" from IBM and "Titan" from Cray with the XE 6.
CHollman82
1 / 5 (6) Nov 06, 2011
Back to your picture book CHollman, this is all a bit too hard for you to cope with until you grow up. Try not to use the term 'we' when you are speaking for yourself. I'd like it if mentally deficient people like yourself, were culled out at birth as a priority over a common language but hey .. maybe this new super fast computer will find a way to solve these problems.

Hopefully this supercomputer is used for reasons other than guessing answers (or questions) for TV game shows or playing chess.



I don't know what the fuck you're talking about other than that you're using an opportunity to insult me as you always do. I have a masters in computer science from Carnegie Mellon University and I work as firmware engineer with a company that makes fiber optic test and measurement equipment for public and private organizations worldwide...
bluehigh
3.7 / 5 (9) Nov 06, 2011
I don't know what the fuck you're talking about other ..


I understand. Its a bit hard for you to grasp anything beyond simple concepts. Perhaps you forget that you drew first blood and as I repeat .. I can hold a grudge indefinitely.

You belittle and insult others and seem to think that its okay and yet when the tables are turned you don't much like it.

Even in the past few hours you have told people that they are ignorant, that they have no understanding of the subject matter, that they should educate themselves etc etc and as far as I can tell it is you that have these characteristics. About that mirror ..

In a pissing contest over academic qualifications I would win by a light year but I'll skip that because its you that need to massage your ego, not me.

Regardless, the construction of this new supercomputer is a good technical achievement and as you say 'the most significant factor is efficiency and organization of the data ...'.

CHollman82
1 / 5 (6) Nov 06, 2011
I can hold a grudge indefinitely.


That sounds like a personal problem.

I am done indulging your childish antics.
bluehigh
4 / 5 (8) Nov 06, 2011
Certainly your mental problems outweigh any personal problems I might have. It is you that choose to continue this nonsense.

Try to stop insulting others (and I will avoid striking at your sockpuppets and bum chums). Back up your contradictions with references. Stop suggesting that you have a monopoly on the truth and generally have tolerance of differing points of view. Perhaps get a sense of humour too.

Then we will get along just fine.
CHollman82
1 / 5 (6) Nov 06, 2011
and I will avoid striking at your sockpuppets


I have no AE's on this site...
CHollman82
1 / 5 (6) Nov 07, 2011
Stop suggesting that you have a monopoly on the truth


When I say 1 plus 1 equals 2 I will claim that to be the truth.

I am sorry you can't see that what I call truth is equally truthful as that is.
CHollman82
1 / 5 (6) Nov 07, 2011
Or do you not believe in formal logic? If I were to tell you that:

If A then B, A therefore B

would you argue that I was claiming a monopoly on truth?

Truth exists.
Truth can be known in given contexts.
Truth may or may not be knowable in the universal context.

Personally, I don't believe that universal truth is knowable, but contextual truth is ABSOLUTELY knowable, to claim otherwise is to deny logic.
CHollman82
1 / 5 (6) Nov 07, 2011
In my personal experiential context the cup I am using is red insomuch as my perception of that cup is correlated to my conceptual understanding of the word "red".

"My cup is red" is a true statement given the above context.

Go ahead and object to my monopolistic claim on truth.
CHollman82
3 / 5 (2) Nov 09, 2011
Some loser with 5 or 6 different accounts has nothing better to do than down vote all of my posts with all of his accounts. He has been doing this for a while now, it probably gives him reason enough to go another day without killing himself. I am glad I can help some depressed miserable person find a reason to continue living.
Eikka
not rated yet Nov 09, 2011
"Without batteries, there's no amount of solar power that will power your home when the sun don't shine." - Ekka

Just pump it into the grid.


And pray tell how would that help?
Vendicar_Decarian
0 / 5 (34) Nov 13, 2011
@Eikka

"How would it help?" - Eikka

Foolish. First define the problem.

The problem in the context being discussed here is energy availability in a carbon fuel reduced world.

Every erg of energy produced from non-fossil sources - no matter the source - can be used to offset the consumption of fossil fuels.

The fact is, that Solar need not power the the current requirements of a home as the home requests it. The energy production can be deferred by using the grid to offset consumption of coal, oil, and natural gas for fuel in conventional generating plants while PV electric power is produced.

So the home may produce 5 KWh of electric power during the day but only need those 5 KWh at night. So you pump the power generated during the day into the grid - offsetting other methods of production - and suck it back out again at night from whatever generating sources there are operating at that time.

The perpetual failure is to believe that any one method of generating power cont..
Vendicar_Decarian
0 / 5 (34) Nov 13, 2011
be the only method used. This is a foolish assumption.

There will be many methods of generation employed, including fossil for a time - while it lasts. So the other sources will provide power at night and solar during the daylight hours.

It isn't rocket science. But some people just refuse to comprehend because reality doesn't mesh with their preferred and typically deluded world view.

More news stories

New breast cancer imaging method promising

The new PAMmography method for imaging breast cancer developed by the University of Twente's MIRA research institute and the Medisch Spectrum Twente hospital appears to be a promising new method that could ...

Research proves nanobubbles are superstable

The intense research interest in surface nanobubbles arises from their potential applications in microfluidics and the scientific challenge for controlling their fundamental physical properties. One of the ...

Using antineutrinos to monitor nuclear reactors

When monitoring nuclear reactors, the International Atomic Energy Agency has to rely on input given by the operators. In the future, antineutrino detectors may provide an additional option for monitoring. ...