Slashdot Mirror


Japan's 8-petaflop K Computer Is Fastest On Earth

Stoobalou writes "An eight-petaflop Japanese supercomputer has grabbed the title of fastest computer on earth in the new Top 500 Supercomputing List to be officially unveiled at the International Supercomputing Conference in Hamburg today. The K Computer is based at the RIKEN Advanced Institute for Computational Science in Kobe, Japan, and smashes the previous supercomputing records with a processing power of more than 8 petaflop/s (quadrillion calculations per second) — three times that of its nearest rival."

179 comments

  1. The fastest rival isn't the show computers. by sethstorm · · Score: 1

    So it's faster than the Crays on the list, the nearest competitors?

    --
    Twitter supports and protects racists - by smearing their critics with the "Hate Speech" label.
    1. Re:The fastest rival isn't the show computers. by ElizabethGreene · · Score: 2, Funny

      In other news, company announces the system will be used initially for Bitcoin mining. The Payback period is expected to range between three weeks an two hundred years, depending on market conditions....

    2. Re:The fastest rival isn't the show computers. by rubycodez · · Score: 1

      PRNewWires 13 Jun e2011

      Today a spokeswoman for MEXT has revealed that the bitcoin mining system for the K Computer was written using j2ee, "We have estimated that it will only take four and a half years to get the JVM up and running, but after that it will be 'faster than greased soba noodles' ".

    3. Re:The fastest rival isn't the show computers. by Stormtrooper42 · · Score: 1

      Meh. The current processing power of the Bitcoin network is reportedly equivalent to 103 PetaFLOPS (according to http://bitcoinwatch.com/)

      Of course, it is technically incorrect to refer to FLOPS (Floating Point Operations Per Second) in the case of Bitcoin, as the hashing algorithm doesn't use any floating point operations, but I digress.

  2. fastest known by decora · · Score: 1

    the NSA has it's own chip fab

    1. Re:fastest known by fuzzyfuzzyfungus · · Score: 2

      I imagine that their in-house fab churns out some very interesting niche designs; but it would be a real surprise if it has any terribly impressive capabilities in general-purpose compute applications. Staying on the bleeding edge of fabrication requires serious money, while just quietly gobbling up commodity stuff from Intel or Nvidia or whoever won't raise any eyebrows.

      They probably have some cool specialized crypto-crunchers based on cryptoanalysis that hasn't officially been done yet, and I suspect that they are the chaps to talk to when you need a chip that absolutely hasn't been backdoored in china; but I suspect that their process density and clockspeed capabilities are middling at best.

    2. Re:fastest known by durrr · · Score: 1

      But NSA only use theirs for spying on their own and other nations citizens and corporations so it's not very relevant in the big picture. Think of it like building the worlds tallest skyscraper, in a pit deeper than the height of the building, then you put a lid on it, fill the pit with water and house a family of goldfishes in it. That's more or less what a NSA "supercomputer" is.

    3. Re:fastest known by fuzzyfuzzyfungus · · Score: 1

      Except that all the goldfish are wearing dark glasses and shoulder holsters, and are trained to use their unnaturally bulbous eyes to tap fiber optic lines...

    4. Re:fastest known by Anonymous Coward · · Score: 0

      Not anymore.

    5. Re:fastest known by MightyYar · · Score: 1

      But it's the world's best goldfish bowl.

      --
      W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
    6. Re:fastest known by L4t3r4lu5 · · Score: 1

      Rubbish. They want the highest Prime95 and SuperPi scores and to find the next Mersenne prime.

      --
      Finally had enough. Come see us over at https://soylentnews.org/
    7. Re:fastest known by rasmusbr · · Score: 1

      I imagine that their in-house fab churns out some very interesting niche designs; but it would be a real surprise if it has any terribly impressive capabilities in general-purpose compute applications. Staying on the bleeding edge of fabrication requires serious money, while just quietly gobbling up commodity stuff from Intel or Nvidia or whoever won't raise any eyebrows.

      They probably have some cool specialized crypto-crunchers based on cryptoanalysis that hasn't officially been done yet, and I suspect that they are the chaps to talk to when you need a chip that absolutely hasn't been backdoored in china; but I suspect that their process density and clockspeed capabilities are middling at best.

      That's what they want you to think.

      It's obviously a new world order government conspiracy to mine all the remaining Bitcoins.

    8. Re:fastest known by fuzzyfuzzyfungus · · Score: 1

      Reasonable enough. The Greys only accept payment in Bitcoins or cattle mutilation, and the latter causes unfortunate political pushback in some of the early primary states...

    9. Re:fastest known by AJH16 · · Score: 1

      And shoot lasers?

      --
      AJ Henderson
    10. Re:fastest known by djdanlib · · Score: 1

      I suspect that they are the chaps to talk to when you need a chip that absolutely hasn't been backdoored in china

      Right on! You get a chip that has been backdoored in the good old US of A instead.

    11. Re:fastest known by BancBoy · · Score: 1
      --
      [UID-HeinzIntel]
    12. Re:fastest known by Anonymous Coward · · Score: 0

      ----
      Staying on the bleeding edge of fabrication requires serious money, while just quietly gobbling up commodity stuff from Intel or Nvidia or whoever won't raise any eyebrows.
      ----

      I agree; I've been ordering a graphics card per week for some time now, and no one at work has noticed yet.

    13. Re:fastest known by fuzzyfuzzyfungus · · Score: 1

      I suspect that that bothers you a great deal less when you are the sinister G-men in dark glasses...

  3. oblig. by mevets · · Score: 4, Funny

    640K cores is enough for anyone.

    1. Re:oblig. by jellomizer · · Score: 1

      Is that SIMD or MIMD?

      --
      If something is so important that you feel the need to post it on the internet... It probably isn't that important.
  4. quadrillion? by CurryCamel · · Score: 1

    Anyone here who find that 'quadrillion' is more descriptive than peta? (or 1e15, for that matter?).

    1. Re:quadrillion? by Anonymous Coward · · Score: 0

      Perhaps so, but this thing is powered by cats, which is why peta got involved.

    2. Re:quadrillion? by Lord+Lode · · Score: 1

      I find peta more descriptive, it has only one meaning, while quadrillion has two.

    3. Re:quadrillion? by niklask · · Score: 1

      Hey now, I'm a proud member of People for the Eating of Tasty Animals.

    4. Re:quadrillion? by Splab · · Score: 2

      I agree, big numbers are a bit of a problem since the naming scheme isn't the same around the world.

      There can be 6 orders of magnitude difference between trillion and trillion depending on where you are.

      Danish for instance goes.
      million, milliard, billion, billiard, trillion (10^18)
      vs. US:
      million, billion, trillion (10^12)

      Peta on the other hand has a somewhat more unified meaning.

    5. Re:quadrillion? by Anonymous Coward · · Score: 0

      Nanometers, micrometers, millimeters, centimeters, decimeters, meters, decameters, (hecto... hm.), kilometers, megameters... hey wait. I guess meters and flops are just differently-prefixed. Sometimes. *sigh*

    6. Re:quadrillion? by Nadaka · · Score: 1

      Peta has two meanings. 1 is a very large number prefix, the other is a bunch of animal "loving" hippies.

    7. Re:quadrillion? by Anonymous Coward · · Score: 0

      So.... how much is it in units of bit coins?

    8. Re:quadrillion? by Teun · · Score: 1
      Not at all, quite the contrary.

      The world does not not agree on the name of any number beyond million, the Americans call it billion, the Europeans milliard and so forth.

      http://en.wikipedia.org/wiki/Names_of_large_numbers

      --
      "The likes of Facebook and WhatsApp are free to those whose privacy is of zero value."
    9. Re:quadrillion? by MobileTatsu-NJG · · Score: 2

      Anyone here who find that 'quadrillion' is more descriptive than peta? (or 1e15, for that matter?).

      I would lean towards "brazillian!"

      --

      "I like to lick butts!" by MobileTatsu-NJG (#32700246) (Score:5, Informative)

    10. Re:quadrillion? by ColdWetDog · · Score: 1

      I suppose that's why we have scientific notation, yes?

      --
      Faster! Faster! Faster would be better!
    11. Re:quadrillion? by Splab · · Score: 1

      Did you bother to rea he thread or just felt like random trolling?

    12. Re:quadrillion? by pszilard · · Score: 1

      Anyone here who find that 'quadrillion' is more descriptive than peta? (or 1e15, for that matter?).

      Firstly, while for some quadrillion might sound more descriptive, in fact it's simply confusing. In some parts of the world it means 10^24 while in others 10^15: http://www.thefreedictionary.com/quadrillion

      Secondly and most importantly, "peta-" is an SI prefix so it's inherently univocal: http://en.wikipedia.org/wiki/SI_prefix#List_of_SI_prefixes

    13. Re:quadrillion? by mswhippingboy · · Score: 1

      If God didn't want us to eat animals, he wouldn't have made them out of meat.

      --
      Sometimes the light at the end of the tunnel is the headlight of an oncoming train.
    14. Re:quadrillion? by TheLink · · Score: 1

      If PETA didn't want us to eat animals they should stop encouraging hot women to go nude and have the pictures published because we eat animals.

      Think about it... The more we eat meat the more sexy nude/semi-nude women we'll get to see? OK.... Meat buffet time!

      So either they are retards or they want us to eat more meat. Perhaps they're all having cognitive issues due to dietary deficiencies? ;)

      --
    15. Re:quadrillion? by tomhudson · · Score: 2

      The world does not not agree on the name of any number beyond million, the Americans call it billion, the Europeans milliard and so forth.

      Soon you'll be able to take out one of those "nots", because the Americans have figured out that if they change the definition of a billion to the European one, it chops 3 places off the national debt.

    16. Re:quadrillion? by ladoga · · Score: 1

      Nanometers, micrometers, millimeters, centimeters, decimeters, meters, decameters, (hecto... hm.), kilometers, megameters... hey wait. I guess meters and flops are just differently-prefixed. Sometimes. *sigh*

      What do you mean? Prefixes are always the same:
      ...Gm (10^9m), Tm (10^12m), Pm (10^15m).
      Same goes for flop/s. Prefix is only a multiplier.

      It's correct to say that Sun's average distance from the Earth is roughly 149.6 gigametres. (149,600,000 kilometres).

  5. Cray is way down the list - China has the nearest. by Anonymous Coward · · Score: 0

    The Chinese Tianhe-1A is nearest with 2.6 petaflops

  6. Imagine by guttentag · · Score: 1

    Imagine a beowulf cluster of... wait. Are we still doing that? OK, now I feel old. In Slashdot years.

    1. Re:Imagine by fuzzyfuzzyfungus · · Score: 3, Funny

      I'm pretty sure that we've moved all our beowulf clusters to a cloud that runs linux in soviet russia...

    2. Re:Imagine by Anonymous Coward · · Score: 0

      Which means all your petaflops are belong to us now?

    3. Re:Imagine by circletimessquare · · Score: 1

      how large is this beowulf cluster in Library of Congresses?

      --
      intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
    4. Re:Imagine by CSMoran · · Score: 1

      It doesn't run linux, it's being run linux -- it's in soviet Russia.

      --
      Every end has half a stick.
    5. Re:Imagine by Anonymous Coward · · Score: 0

      No, we aren't doing that anymore.

      Now it's, "8 petaflop/s! That's almost enough to run Vista!"

    6. Re:Imagine by GameboyRMH · · Score: 1

      It's OVER NINE THOUSAAAAAND!

      --
      "When information is power, privacy is freedom" - Jah-Wren Ryel
    7. Re:Imagine by Anonymous Coward · · Score: 0

      Protected by sharks with lasers?

    8. Re:Imagine by bryan1945 · · Score: 1

      You forgot Natalie Portman and grits. Possibly "All your base belongs to us."

      --
      Vote monkeys into Congress. They are cheaper and more trustworthy.
  7. Supercomputers seem to evolve faster than PCs by Lord+Lode · · Score: 1

    For supercomputers, it seems at least once a year something doubles. For desktop computers... Mine is 4 years old and still similar in specs to PCs that are being sold today.

    1. Re:Supercomputers seem to evolve faster than PCs by jpapon · · Score: 3, Informative

      Two things, your 4 year old desktop is nowhere near as fast as the new i7s. It just seems that way because you're not doing number crunching on it; for normal applications, you'd probably see a bigger boost by switching to a SSD then a new CPU. Second, these supercomputers are massively parallel, so while the procs themselves do get faster, the real increase in speed seen comes from adding lots more cores.

      --
      -- Let us endeavor so to live that when we pass even the undertaker shall be sorry. -- M. Twain
    2. Re:Supercomputers seem to evolve faster than PCs by Yvanhoe · · Score: 2

      How many cores ? Have you considered the GPU as well ? These years, it is the form factor and the price that are halfed every year or so. It will continue until we find a use for the ridiculous power we have today.

      smartphones are now dual-core, mainstream computers DO continue to improve.

      --
      The Wise adapts himself to the world. The Fool adapts the world to himself. Therefore, all progress depends on the Fool.
    3. Re:Supercomputers seem to evolve faster than PCs by fuzzyfuzzyfungus · · Score: 3, Insightful

      I'm not sure that you are being entirely fair to the desktops:

      On the one hand, since a vast percentage of desktops are sold to budget-conscious users with fairly defined needs, the bottom end of the desktop market moves fairly sluggishly(of course, the bottom end of 'supercomputers' also moves more sluggishly; but nobody bothers to talk about the "250,000th fastest supercomputer!!"); but the top end has been moving at a reasonably steady clip.

      Back in mid 2007, a Core2 quad was Pretty Serious Stuff, with maybe a Geforce 8800 or 9800 and 4-8 gigs of RAM if you were hardcore like that.

      That will still go head to head with a contemporary budget to midrange box; but if you spent the same money today that you would have had to spend on that, you could be talking a high-end i7, a markedly more powerful graphics card(or 3 of them), and two or three times the RAM. Plus, the now-reasonably-cost-effective-even-when-large-enough-to-be-useful SSD that will have driven your I/O numbers through the roof.

      Apathy and diminshing returns keep the desktop market boring; but if those are no object, you can still go nuts.

    4. Re:Supercomputers seem to evolve faster than PCs by IceNinjaNine · · Score: 1

      Apathy and diminshing returns keep the desktop market boring; but if those are no object, you can still go nuts.

      As a VMWare workstation and Virtualbox junkie, I can attest to this.

    5. Re:Supercomputers seem to evolve faster than PCs by Nadaka · · Score: 1

      The big advancements in personal computing these last few years have been mostly been in graphics cards. Though density has improved, the benefits have been going more towards power efficiency than towards raw speed.

    6. Re:Supercomputers seem to evolve faster than PCs by fuzzyfuzzyfungus · · Score: 1

      It is true that GPUs have seen a lot of the excitement(particularly with their transition toward being general-purpose); but part of that is arguably definitional:

      Because(contemporary, I'm sure SGI and Sun were doing cool stuff back when I was knee-high to a grasshopper...) multi-GPU tech started its life as a hardcore gamer feature, you can get "desktop" motherboards with support for 3 or 4 16x(physical, usually 8x electrical) PCIe graphics cards. The moment you add a second CPU socket, though, it becomes a "workstation" or "pedestal server" part. Especially now that the FSB is dead, removing one of the major bottlenecks that made older multi-socket systems scale like pure suck, multi-socket systems run like a bat out of hell, support huge amounts of RAM, and benefit as much as the single socket ones do from the number of cores per socket that are now available. However, all that is, by definition, "workstation" or "server".

      With supercomputers, on the other hand, since that is a more or less nebulous category that simply refers to whatever is huge and fast at the time, you can add a whole additional row of floor-to-ceiling racks and the category doesn't change.

    7. Re:Supercomputers seem to evolve faster than PCs by ustolemyname · · Score: 1

      What if your 4 year old desktop is a dual-quad core (8 core) machine? Sure, it benches slower than some i7's, but it's as good as an i5.

    8. Re:Supercomputers seem to evolve faster than PCs by Anonymous Coward · · Score: 0

      For desktop computers... Mine is 4 years old and still similar in specs to PCs that are being sold today.

      No, it is NOT even close! It is sad that people still view GHz as some metric of computing.

      http://www.cpubenchmark.net/

      There, go and find your old CPU and compare it to a modern one that sells at a similar price.

    9. Re:Supercomputers seem to evolve faster than PCs by Bengie · · Score: 1

      An i7 quad is about twice as fast as a Core2Quad. SandyBridge i7 is about 30% faster per core than the original i7 and has 50% more cores, the Ivy bridge coming out next year is about 20% faster than the SandyBridge per core and has another 50% more cores.

      Assuming you use a Core2Quad(about 4 years old), current CPUs are about 2*1.3*1.5 = 3.9 times faster, and that's not including the new AVX instructions that are about twice as fast as SSE. Add in AVX and you're talking about very large performance differences. Ivy bridge next year will be about 7 times faster.

      These are low estimates. I've seen server benchmarks where an Nahalem-i7-quad was about 4-8 times faster than a Core2Quad because of extra cache and better prefeching+memory-subsystem.

      Most applications/games don't make use of the extra threads yet, but there are several games coming out in the next year that will scale well.

    10. Re:Supercomputers seem to evolve faster than PCs by eqisow · · Score: 1

      If that's the case, then it was hardly a fair comparison to begin with.

    11. Re:Supercomputers seem to evolve faster than PCs by grim4593 · · Score: 1

      You Live! How have you been?

    12. Re:Supercomputers seem to evolve faster than PCs by djdanlib · · Score: 1

      When he's saying "the new i7s", he is talking about the 2000 series i7s, not the 800/900 series.

      There have been a lot of improvements since those four years ago. Check out AVX, for example: Sandy Bridge has 256 bit vector operations. Sure, if you can crank integer math along at 4.2 GHz overclocked, you can probably beat the pants off a 3.4 GHz machine, but... when that 3.4 GHz machine includes operations that take the same amount of time to execute but process twice as much data, or reduce a multiple-instruction process significantly, then you're talking about a totally different ballgame. SSE4.2 came out on the Nehalem i7s, so you missed those extensions that process int and float values, and move data around more efficiently. And if you missed out on SSE4.1, which was just starting to come out four years ago, then you're really missing a lot. Also, the memory controllers have been revamped quite considerably twice now. Then there's the electrical efficiency of the product, which has improved significantly.

      So really, even if your older system may be able to compete on one of the many benchmarks, it doesn't mean it's competitive on all fronts. But I suppose that you should consider what makes you happy instead - if you're okay with what you've got, then you're good! No sense in upgrading for the sake of numbers. I built a 3.0 GHz P4 back in 2003, and finally replaced it this year because I was working on a project that exceeded its processing capacity. Otherwise, it was fine.

    13. Re:Supercomputers seem to evolve faster than PCs by Anonymous Coward · · Score: 0

      Then you probably use like 25% of the power of of a 4 year old computer and would use 10% of a new computer. Nothing really shocking here.

  8. Cool, what are they using it for? by Sgs-Cruz · · Score: 0

    Bigger news than a new fastest supercomputer on the planet would be one that had as its primary mission peaceful uses. The Roadrunner (fastest computer in the USA) is for H-bomb simulations ("stockpile stewardship") and gives some of its time to climate change and magnetically-confined plasma (for fusion power) simulations. (Possibly just for PR.) Hopefully since Japan doesn't have a nuclear weapons program this machine will be used 100% for peaceful purposes!

    --

    Karma: pi (Mostly due to circular reasoning in posts).

    1. Re:Cool, what are they using it for? by circletimessquare · · Score: 5, Interesting

      peace is not a static state of being. peace is a balanced state of tension between armed foes. war is a disruption in this equilibrium that is then restored. the goal of maintaining peace is to not have any sudden shocks to the status quo

      you will never, ever, have a world where peace is simply a static state of being that requires no armed maintenance. why? human nature is why

      show me a place where everyone is unarmed and peaceful, and i'll show you a warlord's pillaging grounds

      sorry, but this is reality. stop asking for things that don't exist, and never will, as long as human beings are human beings

      --
      intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
    2. Re:Cool, what are they using it for? by cashman73 · · Score: 1
      The Roadrunner (fastest computer in the USA) is for H-bomb simulations ("stockpile stewardship") and gives some of its time to climate change and magnetically-confined plasma (for fusion power) simulations. (Possibly just for PR.)

      Might want to review the top 10 list again. The fastest supercomputer in the USA is Jaguar at ORNL. Much of it's CPU time is dedicated towards energy research -- biofuels, cellulosic ethanol, that sort of thing,. . .

    3. Re:Cool, what are they using it for? by Anonymous Coward · · Score: 0

      VERY well stated.

    4. Re:Cool, what are they using it for? by Nadaka · · Score: 1

      Japan does have a nuclear weapons program. It is just sleeping. If/when the US withdraws its naval forces from the defense of Japan, perhaps due to pressure from China, japan can be mass producing nuclear weapons in just a few months.

    5. Re:Cool, what are they using it for? by M0j0_j0j0 · · Score: 1

      Absolutely nothing , they are using it for the "lulz" and bitcoin farming, and troll feeding on mondays.

    6. Re:Cool, what are they using it for? by Princeofcups · · Score: 1

      It is also true that the US is spending about ten times more on the military than is needed for peace. It is also true that in the US we use capitol letters and periods to make things easier to reads.

      --
      The only thing worse than a Democrat is a Republican.
    7. Re:Cool, what are they using it for? by SnarfQuest · · Score: 1

      We spend that much to make absolutely sure that there won't be any war, and so far it's been working great. Except for radical islam, Afganistan, Lybia, Iran, Iraq, Syria, Egypt, Palestine, ...

      --
      Who would win this election: Andrew Weiner vs Andrew Weiner's weiner.
    8. Re:Cool, what are they using it for? by marnues · · Score: 1

      Costa Rica.

    9. Re:Cool, what are they using it for? by GameboyRMH · · Score: 1

      Peace can be a static state of being, most countries are not in a "balanced state of tension between armed foes." You're describing something like the relationship between Israel and Iran, would you say that's the same relationship shared between countries in the Americas or Europe?

      The whole world isn't in one giant Mexican standoff.

      --
      "When information is power, privacy is freedom" - Jah-Wren Ryel
    10. Re:Cool, what are they using it for? by That+Guy+From+Mrktng · · Score: 1

      show me a place where everyone is unarmed and peaceful, and i'll show you a warlord's pillaging grounds

      the kingdom of Bhutan

      http://en.wikipedia.org/wiki/List_of_countries_without_armed_forces

    11. Re:Cool, what are they using it for? by circletimessquare · · Score: 1

      yes, the whole world is in a giant mexican standoff

      where areas of the world are at peace, they are at standoff with other areas of the world. this isn't to say peace isn't impossible without land wars. conflict is inevitable, war is not. within a country, the police keep things from descending into physical violence, and the courts resolve conflicts verbally. this is as close as you get to peace on a global basis, or within and maybe even between a handful of nation blocs like ASEAN or the EU, should they congeal further and invest more common destiny in with their neighbors

      as i hope they do: in the case of europe, a continent famous for hundreds of bloody wars killing millions, there is only peace today: there is hope. but it's a fraught financial marriage, not a political one, which it must evolve into, without fragmenting

      if peace is to be achieved, it is with a rational reading of human nature, not a castle in the sky way of thinking. that is: a police force to keep those who always try violence to a simmering minimum, and courts used to resolve conflict. in this "perfect" world, the only war that will be left will be street level gang stuff, which is the nucleus for all larger wars, and therefore must always be maintained at a minimization, constant effort, or the street gangs will metastasize and we will have warlords across the countryside

      --
      intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
    12. Re:Cool, what are they using it for? by Wyatt+Earp · · Score: 1

      Japan is considered to be 6-12 months from a weaponized thermonuclear warhead if they decided they wanted on.

      They also have a Peacekeeper/SS-25 class rocket that could be weaponized quite easily.

      http://www.largeassociates.com/R3126-A1-%20final.pdf
      http://en.wikipedia.org/wiki/Japanese_nuclear_weapon_program#De_facto_nuclear_state

    13. Re:Cool, what are they using it for? by Wyatt+Earp · · Score: 1

      Cutting the US defense budget by 90% would strip the entire US military down to something just a little bigger than the current US Marine Corps or a little smaller than the French Armed Forces.

    14. Re:Cool, what are they using it for? by Wyatt+Earp · · Score: 1

      The United States does not spend all that to insure there won't be any way. The US spends all that and has that giant military established to "provide the military forces needed to deter war and to protect the security of our country."

      http://www.defense.gov/about/#mission

    15. Re:Cool, what are they using it for? by Wyatt+Earp · · Score: 1

      There wasn't peace in Europe three years ago when Georgia and Russia fought. Nor was there peace in Europe during the Madrid or the London bombings.

      Nor was there peace in Europe from 1991 to 1999 when the Balkans were at war.

      Ingushetia and Chechnya are still at war, both of which are on the European continent.

  9. Built by Fujitsu by Anonymous Coward · · Score: 1

    The article misses out who built it... Fujitsu did, and it's nice to be proud of the company you work for and that my colleagues in Japan achieved something rather cool :)

    Oh, and it's not finished yet :)

    1. Re:Built by Fujitsu by gman003 · · Score: 1

      Fujitsu is actually mentioned in the article. It's not big (they seem to give more credit to Japan as a whole than the individual company), but they're still credited.

    2. Re:Built by Fujitsu by afidel · · Score: 3, Interesting

      Huh?
      The K Computer was built by Fujitsu, and contains more than 80,000 2GHz SPARC64 VIIIfx CPUs, each with eight cores, to deliver a total of more than 640,000 processing cores.

      That said I'm fairly surprised that it managed to be the 4th highest efficiency system, the SPARC64 isn't really known for being a hugely efficient and the low density of FLOPS/chip would normally mean it needs more support infrastructure further lowering the efficiency. Obviously the guys at Fujitsu have managed to do some great system engineering since Rmax is so close to Rpeak, kuddo's to them for making an awesome system around an ok chip!

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
    3. Re:Built by Fujitsu by Talcyon · · Score: 1

      Apologies. I'm on a lot of drugs right now for an infection. I meant the slashdot entry doesn't mention Fujitsu, not the linked article.

    4. Re:Built by Fujitsu by afidel · · Score: 1

      Wow, after reading up on the VIIIfx I take back what I said about ok chip, the thing is an HPC beast. 128GFLOPS in a 58W envelope is crazy good. Increasing the dual precision FP registers from 32 to 256 is huge, this is obviously not a chip designed to go into Sun or Fujitsu business class system but tailor designed for HPC (heck going from the VII to the VIIIfx they actually lost L2 cache while increasing the core count, not something you would do for commercial workloads). I wonder if it is worth it for Fujitsu though, 80k parts is a big single order but unless they get a lot more HPC wins I can't see it paying off financially as modern designs like this cost hundreds of millions if not billions of dollars.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
    5. Re:Built by Fujitsu by the+linux+geek · · Score: 1

      The fact that it's SPARC is basically incidental - the VIIIfx is far closer to the traditional vector processors in design. The SPARC scalar cores are basically underclocked versions of the awful VII's currently in use in commercial systems, and comprise only a few percent of the VIIIfx's compute power, and probably only a few percent of the die. It doesn't primarily use standard SPARC instructions, but a custom vector instruction set called HPC-ACE.

    6. Re:Built by Fujitsu by Anonymous Coward · · Score: 0

      The article misses out who built it..

      What is truly unfortunate is that the article also glosses over the machine's cost: more than 6 times that of any other supercomputer in history.

    7. Re:Built by Fujitsu by antifoidulus · · Score: 2

      Keep in mind that one of the defining features of this CPU is 2 vector coprocessors per core, Intels offerings only have 1. For number crunching applications like matrix multiplication(which is what these benchmarks are based off of), the extra vector coprocessor makes a HUGE difference.

  10. Re:Remarkable stuff by Lord+Lode · · Score: 2

    Did you post that just to be able to use the word that is in your sig?

  11. But.. by Anonymous Coward · · Score: 0

    Can it run Crysis?

    1. Re:But.. by burisch_research · · Score: 1

      Could most certainly ray-trace Crysis in real-time. This is a beast of note.

      --
      char*f="char*f=%c%s%c;main(){printf(f,34,f,34);}";main(){printf(f,34,f,34);}
  12. Not the fastest by Anonymous Coward · · Score: 0

    A birdie says a supercomputer exists with a 2000-8000 petaFLOP performance rating.

    1. Re:Not the fastest by Anonymous Coward · · Score: 0

      How much power do you think that would draw?

    2. Re:Not the fastest by Anonymous Coward · · Score: 0

      1.21 gigawatts?

  13. Not fair by Anonymous Coward · · Score: 0

    Not fair. Theirs runs on radiation.

    1. Re:Not fair by sethstorm · · Score: 1

      Better than Chinese slave labor.

      --
      Twitter supports and protects racists - by smearing their critics with the "Hate Speech" label.
  14. SPARC PR? by Anonymous Coward · · Score: 0

    I wonder if Oracle (nee Sun) will use this as a PR tool given that it uses SPARC CPUs (though it does run Linux, and not Solaris).

    In general though, the CPUs are generally the least interesting parts of these types of machines. I'm personally often more interested in how the cooling and power distribution systems have been designed to run efficiently and have low PUE numbers than any of the raw compute stats. Things like water / glycol / non-air cooling and UPS / generator / power distribution architectures can become challenging when you start talking about megwatts of energy. Also data management and backups too.

    CPU and memory is less interesting than how to tie it all together IMHO.

    1. Re:SPARC PR? by Jeremy+Erwin · · Score: 1

      Seymour Cray would be rolling in his grave if he heard you talking about his beloved supercomputers as if they were glorified refrigerators.

      Oh.... wait.... Never Mind.

    2. Re:SPARC PR? by the+linux+geek · · Score: 1

      I doubt they will, since the SPARC's in this are totally different than the SPARC's in any commercial system. Almost all of the floating-point oomph is provided by massive vector units using an instruction set that isn't (and probably will never be) part of standard SPARC.

  15. How long by Anonymous Coward · · Score: 0

    until the software retards bring that thing to a crawl for reading a website? Or a 10 gigabyte "hello world"?

    1. Re:How long by Anonymous Coward · · Score: 1

      Are you joking? Firefox is "faster" every version!!! (Req: faster computer)

    2. Re:How long by Iamthecheese · · Score: 1

      Your brain, which runs on a few tens of watts and weighs as much as your laptop, has a 10 gigabyte "hello world". you chose the wrong measure for efficiency.

      --
      If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
    3. Re:How long by Anonymous Coward · · Score: 0

      If you think that's in any way comparable, you are more than just a software retard. No words exist yet for your condition.

    4. Re:How long by GameboyRMH · · Score: 1

      At least 3.6 --> 4 was faster (as in more efficient, regardless of computer speed).

      --
      "When information is power, privacy is freedom" - Jah-Wren Ryel
  16. Wow, a really fast compouter! by boyfaceddog · · Score: 1

    Sure the Singularity and Computer Sentience cannot be far-off now. Yay. A faster computer. I guess this is news for today but this is 'dog bites man' stuff. Tomorrow the sun will rise again and there will be a faster computer than this one. Yawn

    --
    Here will be an old abusing of God's patience and the king's English.
  17. Here's What It Looks Like by 1sockchuck · · Score: 2

    Here are some images of the system, which currently uses 672 cabinets and uses about 10 megawatts of power. The K system is more powerful than the next 5 systems combined. It's a big-ass system.

  18. BRAAIINNNNNNNS by Iamthecheese · · Score: 1

    Can anyone tell us the most recent accepted figure for human brain emulation in petaflops and terrabytes of memory?

    --
    If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
    1. Re:BRAAIINNNNNNNS by Anonymous Coward · · Score: 0

      Approximately 3.1415 Libraries of Congress per second.

    2. Re:BRAAIINNNNNNNS by RivenAleem · · Score: 1

      If you are interested, try giving this page a read.

      http://www.transhumanist.com/volume1/moravec.htm

      I googled it, so can't verify it's accuracy, but it looked reasonable.

    3. Re:BRAAIINNNNNNNS by Ecuador · · Score: 1

      Sorry, but we are so far from figuring how the brain works, your question has no meaning at all.
      If you are thinking something along the lines that a big "neural network" can emulate the brain, I would have to tell you that the artificial neuron is a very useful math construct that is only related to a biological neuron via a crude abstraction. Replacing biological neurons with artificial neural networks is similar to replacing a fisherman with a perfect sphere in a math problem : useful in some context, no relation beyond that.

      --
      Violence is the last refuge of the incompetent. Polar Scope Align for iOS
    4. Re:BRAAIINNNNNNNS by dogmatixpsych · · Score: 1

      I research brains and haven't ever seen good numbers. It's really hard to equate brain processing with computer processing. In some ways brains are far superior and in others, far inferior. There seems to be no upper limit on what our brains can learn so essentially we have unlimited hard drive storage. However, for active processes, we're quite limited in what we can do in parallel - sort of. We do many things in parallel but not many consciously (i.e., our sensory systems are always going and our motor and vestibular systems are always going but what we are actively think about is generally limited to one thing at a time, although we can rapidly switch between thoughts).

      The link below (http://tech.slashdot.org/comments.pl?sid=2252976&cid=36500020) is interesting but is still too simplistic. Again, people's brain don't really work like computers so it would be quite a stretch to come up with any sort of FLOPS value. RAM is also difficult to equate with brain processing power but our brains have infinite hard drive space (we just have difficulty accessing information some times). Even then, we don't really store memories like bits on a hard drive. We recreate memories whenever we access them, as far as we know.

    5. Re:BRAAIINNNNNNNS by RightSaidFred99 · · Score: 1

      Also, processing power doesn't matter. Even if we had computers 1,000,000 faster than this that used 50W of power we couldn't simulate a brain. We don't have the model to do so.

    6. Re:BRAAIINNNNNNNS by Anonymous Coward · · Score: 0

      He said brain emulation, so I assume he meant physical simulation of a human brain (taking whatever shortcuts we can).

    7. Re:BRAAIINNNNNNNS by Anonymous Coward · · Score: 0

      Well I know the accepted spelling is "terabyte". Terra is the name of our planet. Complicated, eh?

    8. Re:BRAAIINNNNNNNS by kyle5t · · Score: 2

      The EU's Human Brain Project has an estimate of 1000 times the current fastest supercomputer (probably written about a year ago), so maybe around an exaflop.

      "Today, simulating a single neuron requires the full power of a laptop computer. But the brain has billions of neurons and simulating all them simultaneously is a huge challenge. To get round this problem, the project will develop novel techniques of multi-level simulation in which only groups of neurons that are highly active are simulated in detail. But even in this way, simulating the complete human brain will require a computer a thousand times more powerful than the most powerful machine available today." (link)

      That's 10 doublings, so if Moore's Law holds up this level of capability should be roughly 20 years away. I think it's interesting to note that this also suggests the feature size will halve 5 times to right around 1 nm: atomic scale. My rough understanding is that no matter what you may have heard from semiconductor physicists we are currently pretty clueless as to what, if anything, is going to drive the progress of Moore's Law beyond about 10 nm.

  19. Re:Remarkable stuff by Anonymous Coward · · Score: 0

    Eh? We already know what causes disease. The body has these four 'humors', and when they get out of balance bad things happen. All of these bad things can be cured by magically tapping, twisting, and massaging the spine. Don't they teach kids anything these days?

  20. K Computer by chocapix · · Score: 1

    According to wikipedia, K Computer (or Kobe Computer) refers to cuts of computer from the Tajima-ushi breed of Wagyu semiconductors, raised according to strict tradition in Hyogo Prefecture, Japan. The output from such computers is generally considered to be a delicacy, renowned for its flavour, tenderness, and fatty, well-marbled texture.

    1. Re:K Computer by M0j0_j0j0 · · Score: 1

      This computer power comes from fine beer destillary and it gets massages daily by caretakers who talk to him and never leave him alone for more than two hours.

  21. all thanks to Sony by Gravis+Zero · · Score: 0

    what else was sony going to do with the PS3s they bricked for an "upgrade"?

    --
    Anons need not reply. Questions end with a question mark.
  22. Sparc based by LWATCDR · · Score: 1

    It is using Sparc CPUs and no GPUs. I wonder if Oracle is watching? It will be interesting to see since they now own the Zombie formally known as Sun.
    So when are we going to see nVidia get into this game with ARM+GPU based super computer?

    --
    See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
    1. Re:Sparc based by yarnosh · · Score: 1

      Since SPARC is an open spec, why should Oracle care?

    2. Re:Sparc based by LWATCDR · · Score: 1

      Because Sun developed SPARC and if for no other reason than PR.
      The worlds fastest computer is powered by SPARC makes a great lead in for selling SPARC based servers.

      --
      See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
    3. Re:Sparc based by joib · · Score: 1

      This one uses SPARC chips designed and fabbed (IIRC?) by Fujitsu. Sun/Oracle has nothing to do with it. AFAICT the politics behind this machine is that a few years ago NEC pulled out from the project to design a next generation vector chip for use in a Japanese Earth Simulator follow-up. Hence the project resorted to the Fujitsu SPARC chips, which are not really designed for HPC but are still a domestic design. I wouldn't expect this machine design to become popular outside Japan.

  23. Fastest computer on Earth? by rossdee · · Score: 1

    So there are faster computers on some other planet somewhere?

    I might think about ordering one of those instead, except shipping cost (and time ) would be a problem, and my credit card would expire before they got the order (the speed of light is a bitch)

  24. What's up with 'Earth'? by Quantum_Infinity · · Score: 0

    "fastest computer on earth".

    Lately, the phrase ' in the world' has given way to ' on earth'. Things that used to fastest in the world are now fastest on earth. Factually both are correct but their implications are slightly different. When you use the word 'world' you are mostly concerned with what's happening on earth and you are not at all concerned with other heavenly bodies, however the moment you use the word 'earth' it implies that you are talking in a broader context with encompasses at least our other solar system bodies if not the entire universe. 'Fastest on earth' seems to imply that there are other computers on other planets and that you have knowledge of their existence and that you know that this supercomputer is fastest on earth, though it may not be fastest in the solar system. Of course there can be computers on other planets too (there probably are and this one is indeed fastest on earth) but the incorrect part is that 'on earth' implies that you have knowledge of existence of super computers elsewhere outside the earth too. I think 'in the world' is more appropriate.

    1. Re:What's up with 'Earth'? by Anonymous Coward · · Score: 0

      Regardless of Earth vs World

      It is Earth with a capital e. Earth denotes the planet 3rd from the sun. earth is the dirt that plants grow in.

      So, fastest on earth would disregard the many supercomputers on boats, beaches or hard rock.

    2. Re:What's up with 'Earth'? by Anonymous Coward · · Score: 0

      Since I see stars at night, they are part of my world. We don't know if there are faster computers around them, so saying "fastest on Earth" is technically better.

  25. Performance ... by advance-software · · Score: 1

    so that's approx. 100 Gflop per sparc cpu. (double precision?)

    what can a high end xeon do ?

    1. Re:Performance ... by the+linux+geek · · Score: 1

      70-80, and a Power7 can do a quoted 256. That being said, all of these numbers (including the Xeon and P7 ones) are coming from using strange vector extensions - ones that, in the case of the SPARC VIIIfx, are not available in any commercial-workload system, and are therefore basically irrelevant except for supercomputing.

  26. Increasingly Irrelevant Benchmark by Lawrence_Bird · · Score: 1

    the top 500 is based upon the Linpack benchmark and it is not really a good reflection on 'how fast' a super computer really is. Newer benchmarks, such as graph500 and NAS parallel benchmarks try to make the benchmark more real world. But if all you plan to do is solve linear equations then I guess Linpack is your thing.

    1. Re:Increasingly Irrelevant Benchmark by advance-software · · Score: 1

      yeah - how many polys can this baby real-time raytrace @ 60 fps ?

    2. Re:Increasingly Irrelevant Benchmark by serviscope_minor · · Score: 1

      But if all you plan to do is solve linear equations then I guess Linpack is your thing.

      You are right but for the wrond reasons. An awful lot of HPC stuff spoves linear equations. It forms the inner loop of many PDE solvers, for instance. However, LINPACK is dense linear, whereas many problems where linear equations are the inner loop solve large, sparse systems. That said, there are many inner blocks which are solved as dense problems.

      The Graph500 note that the graph problems are ill-suited to machines which solve 3D pyhsics problems.

      In other words, liesd, damn lies and benchmarks.

      --
      SJW n. One who posts facts.
    3. Re:Increasingly Irrelevant Benchmark by InfiniteZero · · Score: 1

      In other words, liesd, damn lies and benchmarks.

      liesd? Some kind of demon that generates benchmarks?

    4. Re:Increasingly Irrelevant Benchmark by Lawrence_Bird · · Score: 1

      In other words, liesd, damn lies and benchmarks.

      liesd? Some kind of demon that generates benchmarks?

      no, its a daemon that generates politicians

  27. BitCoins Supreme! by PmanAce · · Score: 1

    I wonder how many bitcoins this monster can mine per hour lol?

    --
    Tired of my customary (Score:1)
  28. failure rate? by pigwiggle · · Score: 1

    What I don't quite get - and maybe someone can enlighten me - is how they keep 80K compute nodes going. Even with very reliable hardware, several of these nodes will fail each day. The massively parallel codes I work with (MD) can't deal with a compute node going out. Do other massively parallel codes have a way to deal with this sort of thing? This seems to be a big challenge for parallel computing. When you have a code and problem that can use several thousand nodes, hardware failure will be a daily occurrence. Incidentally, I've had the opportunity to use several thousand cpus in one go. Before New Mexico's Encanto was released for general use I was one of several people that had access to the machine. There really wasn't a problem running million or billion atom systems over several thousand cpus. But this was just brief benchmarking runs. Not data production.

    --
    46 & 2
    1. Re:failure rate? by Anonymous Coward · · Score: 0
    2. Re:failure rate? by Junta · · Score: 1

      Frequently, job restart. Long running jobs have checkpoint and restore. Generally, fault is isolated to a job, so yes, on 80,000 systems you'll have a failure, but if you were doing 800 large jobs, you only lose 1, and 799 jobs didn't even know something went wrong. Generally something like this runs a few benchmarks across the whole thing in the very very beginning, and never again does the whole work as one toward a single task.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    3. Re:failure rate? by Junta · · Score: 1

      Oh, and that one job 'lost' gets restarted shortly thereafter, with the user maybe realizing that it took longer than he thought it should.

      --
      XML is like violence. If it doesn't solve the problem, use more.
    4. Re:failure rate? by pigwiggle · · Score: 1

      Frequent checkpoints would be onerous for billion atom systems.

      --
      46 & 2
    5. Re:failure rate? by Anonymous Coward · · Score: 0

      Yes. There's a return-receipt of sorts (I don't fully understand it so I won't attempt to explain it). Job gets split. Farmed out. Accepted. Worked. Returned. If it never comes back, the compute node is checked for health. If it's down, that chunk gets re-farmed elsewhere. All very fast, of course.

    6. Re:failure rate? by evilbessie · · Score: 1

      Mostly systems fail at start because of manufacturing defects or after a few years due to mechanical failure. At least in one image I saw they were using water cooling on their boards, which should reduce the mechanical failure issue of fans, I would also assume they are using some fault tolerant storage which is the other major failure route. If you take a, possibly optimistic, failure rate of 1% that is about 1 machine failing a day over a 3 year life span. Being as I would make them as similar as possible to make the job of swapping out much easier I can't see it being that much of an issue.

    7. Re:failure rate? by Anonymous Coward · · Score: 0

      What I don't quite get - and maybe someone can enlighten me - is how they keep 80K compute nodes going. Even with very reliable hardware, several of these nodes will fail each day.

      I can't speak for RIKEN's machine, but yes, node failure is a big issue for Top500 runs. Their run would have taken slightly longer than one day to complete. Speaking from experience, having run a few of the top 100 runs myself (with lots of help, of course), you start the job, cross your fingers, and hope all the nodes make it to the end of the job. If they don't, you start the run over and try again. Some machines are nicer than others.

    8. Re:failure rate? by joib · · Score: 1

      Most codes can't deal with node failure. So far it seems the solution is to checkpoint frequently (say, once per hour). There's not much else that is sensible. E.g. running pairs of nodes in lockstep is more expensive than an IO subsystem capable of the checkpointing.

    9. Re:failure rate? by Anonymous Coward · · Score: 0

      Checkpoint restart.

      You partition the problem into workunits, which you then dispatch to the nodes of the computing cluster or some portion of it. If a work unit fails to complete due to a node failure, you redispatch it on another node. These large machines don't tend to be global shared-memory systems, they are cluster systems with very fast message passing systems (MPI/OpenMP). Because the workunits are deterministic, messages can simply be replayed when a workunit is rescheduled, and it picks up where it left off from the last checkpoint. What this does mean is the dependent data for each workunit must be retained, either on the source node or on the central working storage.

    10. Re:failure rate? by pigwiggle · · Score: 1

      This isn't a solution, it's a kludge. All the codes I've worked with have the ability to checkpoint, but in practice it's done infrequently because of the overhead. Increasing the checkpointing commensurate with number of nodes results in diminishing returns - which is antithetical to "massively parallel".

      --
      46 & 2
  29. Japan's Response by Anonymous Coward · · Score: 0

    "And the President responded to the incident by saying 'Screw those commie bastards, and screw their little wussy supercomputer."

  30. oblig. oblig. by Kamiza+Ikioi · · Score: 1

    But does it run BSD?

    --
    I8-D
  31. I believe you want the show computer by sethstorm · · Score: 1

    That is, you want the unrankable Tianhe-1 GPU "show computer".

    --
    Twitter supports and protects racists - by smearing their critics with the "Hate Speech" label.
  32. Re:oblig. oblig. oblig. by captainpanic · · Score: 1

    But will it blend?

  33. It's Linux, like the entire top 10... by Anonymous Coward · · Score: 0

    Of course it's running Linux. Apparently Linux can scale from the smallest devices (even much more than the cellphone market well Linux also reigns king) to the biggest computers on the planet.

    All computer from the Top 500's top 10 are running Linux.

  34. Tagged by bluefoxlucid · · Score: 1

    penismeasuringcontest

  35. Using One Right Now by friedmud · · Score: 1

    Hah - I just started a ~10000 proc job on the machine sitting in position 99....

    I also regularly run jobs on Jaguar (#3)....

    The advances in supercomputing in the last year have been simply astounding. GPUs are changing the game of course.... but the density of CPUs is getting insane. Being able to plug 4x12 core processors into a 1U mobo is getting crazy. Can't wait to see where it goes in the next year!

  36. "fastest on earth".... by mark-t · · Score: 1

    .. as opposed to the fastest in space?

    If they mean the fastest computer we've ever made to date, why don't they just say that instead of using a qualifier like "on earth" which implicates a specifically limited scope on account of an awareness of something faster elsewhere?

    ... or maybe I'm just too damned literal.

  37. Why is GPU a dirty word? by Macman408 · · Score: 1

    Even the story seems to think that you're cheating if you use a GPU - when in fact, depending on the problems you're trying to solve, use of GPUs can make your computer more power-efficient, less expensive, and faster. OK, sure, if you want to argue that not every algorithm will map efficiently to a GPU, I'll accept that argument. But then you have to grant me the reverse argument; not every algorithm maps efficiently to CPUs. The problem in thinking here is just that the CPU is the "correct" way to do things, and sometimes you can "cheat" and use a GPU to do it faster. In reality, some problems just need parallel horsepower, and that's what the GPU excels at. Don't believe me? Well, here are some examples: Graphics (duh - why do you think they make GPUs?), n-body simulations (just need lots of number crunching here), and LINPACK (if you believe the people that complain that the GPU gets an unfair advantage - when really, the CPU is just at a disadvantage because of its architecture).

    There can be a hurdle in learning how to rewrite your code to take advantage of the GPU, some of which might still be hanging around from the 1970s. But if you're going to spend millions of dollars on a new computer, you probably also have millions of dollars to spend on the research that you'll be doing on it. And if you've saved some of those millions of dollars by choosing GPUs over, say, 10x as many CPUs, then you can spend some of what you've saved (and will continue to save in lower power bills) on a programmer or two.

    I don't intend this to attack the K Computer itself - hopefully, its architects listened closely to what the researchers wanted when designing it, and chose CPUs because they were the right choice for the programs that will be running on the computer. But the article (and some observers) seem rather biased, rather than recognizing that both CPUs and GPUs have strengths, and a supercomputer that includes both will likely be better at many problems than a computer that doesn't.

    Of course, there's also the argument that *everything* on the Top500 list is biased, and not just the systems with GPUs, because not everybody's problem looks anything like LINPACK. So sometimes, something that's neither a CPU nor a GPU is the best choice (and whatever that something is might not run LINPACK well) - which is a very valid observation as well.

    1. Re:Why is GPU a dirty word? by JanneM · · Score: 1

      The problem with GPU's for large-scale simulations is in the communications latency, and to some extent the numerical accuracy. As long as your problem needs only single-precision floats and can fit in a single GPU (or two connected with a dedicated bus) it can be insanely fast. Go to doubles and the speed drops quite a bit but is still very fast. Future hardware will likely be much better at high-precision as well.

      But when your problem is too large to fit in a single system, the communications will eat up most of the gain. Remember, you have a hop out to the host system, then from the destination host to the GPU over there; two extra steps over and above the normal node communications. It's still faster, but for things like large-scale neural simulations the speedup may be ~2x or so, no more.

      Now, doubling the speed may sound like a good deal. But remember that it takes time to write, debug and verify that extra code, working off the communications bottlenecks, load distribution and so on. For a given project you really need to decide if:

      * will you save more simulation time over the course of your project than you spend on the extra development time? If you need five extra weeks to use GPUs, you need to have some very long-running simulations in order to save five weeks of calculation time.

      * When you consider hardware and development time and cost, is it more or less efficient for your project to use GPUs or to just add more computing nodes instead?

      For some projects it will turn out to be worth it. For many others it will not. But it's not a slam dunk by any means.

      --
      Trust the Computer. The Computer is your friend.
  38. China? by SnarfQuest · · Score: 2

    Wasn't there an article recently about China's world's fastest supercomputer, and no competition in site for the next several years? Or did I fall asleep at my seak for several years?

    --
    Who would win this election: Andrew Weiner vs Andrew Weiner's weiner.
  39. ah yes this myth by circletimessquare · · Score: 1

    costa rica's famous nonmilitary has to with the unpopularity of a costa rican military dictator who was overthrown in the 1940s. the military was hated as a source of the dictator's support. so the military was "disbanded" to much fanfare (and little substance): you see, of course, costa rica DOES have a military: law enforcement, foreign peacekeeping troops, etc. the famous propaganda that you believe is just an entertaining narrative of the reaction of the people and politicians of costa rica to the military in the 1940s, due the hated deposed dictator. the myth has no reflection in reality

    and you forgot another great myth: the amish. it's easy to preach nonviolence when one in embedded in the middle of a militaristic country: your peace is provided for, so you don't have to provide for it. your peaceful philosophy is therefore a parasitical arrangement with your host country, not a whole, genuine philosophy that can stand on its own. for example, if you were to transport the amish to say, the caucasus mountains, one or two things would happen:

    1. the extinction of the amish
    2. the all-new improved well-armed amish

    likewise, costa rica benefits from the usa's umbrella interest in keeping the status quo in latin america. it's easy to have a lackadaisical military if your peace is guaranteed by someone else. for example, if nicaragua or panama went nuts and started making inroads into costa rica, costa rica would beg the usa for military aid, and/ or bulk up it's existing military

    --
    intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
  40. Big disappointment by Anonymous Coward · · Score: 0

    The supercomputer I'm using is no longer in the Top 100 ;-(

  41. only 18x larger problems in 25 years by peter303 · · Score: 1

    The fastest computers are 100,000x faster in 25 years. But the typical problem in my discipline (seismic) is a 4D grid. If you increase the side of a grid 18x to the fourth power, then you reach 100,000. I was at a conference earlier month where people are still talking about tricks like custom hardware (GPUs) and data compression to squeeze ever larger problems into supercomputers. This aspect has not changed since I went to grad school a quarter century ago.

  42. It still won't be able to predict earthquakes by KeithH · · Score: 1

    although it should give SkyNet a run for its money.

  43. What makes the K-computer fast? by Aater+Suleman · · Score: 2

    The K computer is able to do work more efficiently than GPUs because it uses a very power-efficient core, the Sun VIIIfx. If you peel the onion, it seems like the real reason for energy efficiency is special purpose units and the HPC-ACE instructions. I did a quick investigation of what this core has (and what it doesn't) to make it so energy efficient. It may be an interesting read for some of you guys so leaving a link here: http://bit.ly/kTvvDE

  44. see the Seymour Cray comment above by decora · · Score: 1

    someone above is talking about a Cray being the fastest

    A few decades back, Cray was kept in business solely because the NSA funded him. He had no other customer to support his stuff.

    These 'private companies' with their 'general purpose computers' are often linked to secret government projects in ways we will not understand or know about for decades.

  45. militaries do not advertise their fast computers by decora · · Score: 1

    the NSA has been on the forefront of computer technology for it's whole existence practically.

    almost nobody had ever heard of it before the mid 1970s.

    there was no book about it until circa 1980, and only a handful of books since then

    and yet, it singlehandedly is responsible for a large amount of supercomputer development in the US.

    and it is part of the Department of Defense

  46. yeah. like the war between virginia and new york by decora · · Score: 1

    i mean, people keep saying there could be 'peace' between those states, but it was only a short time ago, 150 years, that those to 'peaceful' states were killing each other.

  47. and yet. by decora · · Score: 1

    every nation practically has a peace movement sooner or later.

    it's endemic to the human species to have anti-war movements.

    so which is really the 'steady state'... the desire for war or the desire for peace, which is behind just about every major religion?

    1. Re:and yet. by circletimessquare · · Score: 1

      everyone wants peace, yes. everybody wants to be rich too. everybody wants to be famous. so what? empty vapid platitudes without any understanding of what it takes to achieve what you want has no value

      well, of course, those who want peace without force of arms do have a plan: all it takes is for human beings to just start acting like human beings never have. simple!

      the "desire for" X != X. dreamers don't produce anything. empty dreams also don't fill you with any wisdom or any rationale on which to pass judgment on others. its just building castles in the sky. a pleasant, empty exercise

      you want peace? good for you. something a toddler can smile and clap at, with no other meaning and value as to actual results in the world

      --
      intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
  48. best comment on earth by decora · · Score: 1

    i mean in the world. sorry.

  49. what the flop is a petaflop? by Anonymous Coward · · Score: 0

    These numbers are getting stupid. It's time to switch to calculations per nanosecond.

  50. Fellow Americans: tomhudson in a nutshell by Anonymous Coward · · Score: 0

    Where you were caught trolling others as ac replies and telling others to do so with you :

    "Wait until he starts on another kick, then reply to him as an AC. It's the new meme". - by tomhudson (43916) on Sunday May 09 2010, @08:29PM (#32150544) Homepage Journal

    QUOTED FROM -> http://slashdot.org/comments.pl?sid=1646272&cid=32150544

    And for what?

    Just because you were shown as technically weak in computing so many times here after you did things like the above to the wrong person:

    http://slashdot.org/comments.pl?sid=2230966&cid=36418796

    That all you have is your geek angst & tricks like ac stalker trolling proven above, and also downmodding unjustly after doing so, shown by your fellow troll friends here http://slashdot.org/comments.pl?sid=2245866&cid=36491652

    * Some "ethics" tomhudson... you don't know the meaning of the word!

  51. Come'on it's not even built yet by Anonymous Coward · · Score: 0

    Wtf? That machine is a year from operation and they're already claiming "fastest"?

    TFA reads: currently in the configuration stage

    News from other sources say it'll take them a year before the system is fully built.

    Anyone can claim they have something "100x faster" by simply saying they have some magical machine "in configuration stage".

  52. Possible because others with armies defend them by Quila · · Score: 1

    Generally, countries without armies have official defense treaties with other countries that do have armies. The US, would defend Costa Rica and other countries in the Americas and Pacific, France would defend Monaco, and Australia would defend many Pacific countries. Others are unofficial, such as everybody knows the Italian Army would defend the Vatican were it attacked.

    Bhutan, having a small standing army, is not one of them, although India would protect them should the Chinese decide to invade.

    1. Re:Possible because others with armies defend them by That+Guy+From+Mrktng · · Score: 1

      Indeed, Bhutan have a military assistance treaty with India. But as the original poster asked "..in which everyone lives happily.." Bhutan is one of such places. Also in developing nations you usually get a small isolated town or province that have near zero violence or crime. So having a community that behaves peacefully is not impossible, just that the high population density and diversity of the average city makes it impossible to achieve.

  53. Bitcoins? by Anonymous Coward · · Score: 0

    I wonder if you could have a machine like this pay for itself by mining bitcoins for the first year or so?

  54. if pacifists have no influence by decora · · Score: 1

    then why did hitler kill all of them?

    why waste the time?

  55. Programming is the thing by Goonie · · Score: 1

    I'm not so impressed with the "world's fastest computer" schtick. Take a CPU/GPU with more cores, add a couple of cabinets, and there you go - world's fastest computer (with all due respect to the hardware engineers who put these things together).

    The tricky part is figuring out how to divide your particular problem up so you can actually keep those 100,000-odd cores working, not waiting for intermediate results from other cores.

    And if you thin it's difficult now, in ten years from now the world's fastest supercomputer will probably have something in the order of a billion (thousand million) cores.

    But then again, your average PC (or smartphone) may well have hundreds of CPU cores on it, and today's average CS student won't have a clue how to take advantage of that architecture.

    --

    Any sufficiently advanced technology is indistinguishable from a rigged demo
    --Andy Finkel (J. Klass?)
  56. not so fast by Anonymous Coward · · Score: 0

    Theres an IBM Blue Gene thats 10 petaflops and of course Cray's 50 petaflop supercomputer

  57. Top spot with a SPARC architecture by Kyusaku+Natsume · · Score: 1

    I think that this is far more important than the comparison with the chinese machine, since this means that there is serious competition again in the high end market.

    --
    Mexico: 100% conservative's America now!
  58. Go figure by Anonymous Coward · · Score: 0

    With a tsunami cooling their data centers, one should expect their computers are the fastest in the world!

  59. In All Modesty by PingPongBoy · · Score: 1

    The aspirations of these supercomputer builders knows no bounds. They build an 8 petaflop system, and think the computer is just K. 80 petaflops will probably just be OK, and 800 might be A-OK.

    --
    Know your pads. One time pad: good for cryptography. Two timing pad: where to take your mistress.
  60. 411 on the K Computer by Anonymous Coward · · Score: 0

    I would like (and hope) to think that the "K" represents "Kunle" ...

    It would be nice (if not appropriate) if those in Nihon acknowledged and perhaps even kindly extended an invitation to the man who behind the corporations, is responsible for this. Having created the "Hydra" chip, sold to Sun and renamed SPARC - professor Kunle Olukotun. Still very much holding office at Stanford (Gates/room 408, which coincidentally or not happens to be the area code for San Jose, California - Silicon Valley) and often appearing at conferences, here is his page:

    http://www-hydra.stanford.edu/~kunle/

    That was the real work, this computer is an assembly of it. Just imagine the ideas he has floating right now. People should be acknowledged for their efforts and contributions, let's not forget and give credit where credit at the core is due.

    Re: the comment of no people in the images, perhaps a cardboard cut-out of Mr. K. Olukotun at minimum, I'm sure he wouldn't mind.