Slashdot Mirror


DARPA Wants a 19" Super-Efficient Supercomputer

coondoggie writes "If you can squish all the processing power of, say, an IBM Roadrunner supercomputer inside a 19-inch box and make it run on about 60 kilowatts of electricity, the government wants to talk to you. The extreme scientists at the Defense Advanced Research Projects Agency this week issued a call for research that might develop a super-small, super-efficient super beast of a computer. Specifically, DARPA's desires for Ubiquitous High Performance Computing (UHPC) will require a new system-wide technology approach including hardware and software co-design to minimize energy dissipation per operation and maximize energy efficiency, with a 50GFLOPS per watt goal."

200 comments

  1. extreme scientists by johncandale · · Score: 0, Offtopic

    Extreme Scientists will be on x TV next year

    1. Re:extreme scientists by Anonymous Coward · · Score: 1, Funny

      Back *waaaaay* off, man. I'm an *extreme* scientist!

    2. Re:extreme scientists by causality · · Score: 2, Funny

      Back *waaaaay* off, man. I'm an *extreme* scientist!

      That sounds like a nice bumper sticker. For the rear bumper.

      --
      It is a miracle that curiosity survives formal education. - Einstein
    3. Re:extreme scientists by Quothz · · Score: 4, Funny

      Back *waaaaay* off, man. I'm an *extreme* scientist!

      Oh, yeah? Where's your badge?

    4. Re:extreme scientists by Kjella · · Score: 1

      Lots of funny ones there, but I think Hannibal Lector did the captions:

      The "I've eaten what I study" badge.
      Recipients have prepared their object of study as a cuisine item for eating. Hopefully, the minority of MDs are ineligible for this one. (J)

      --
      Live today, because you never know what tomorrow brings
  2. oblig by Kardos · · Score: 0

    Now, imagine a Beowulf cluster of those....

    1. Re:oblig by Kratisto · · Score: 5, Funny

      Grendel wouldn't stand a chance!

      --
      Conscience is the inner voice which warns us that someone may be looking.
  3. Yeah sure by Anonymous Coward · · Score: 1, Interesting

    If you can squish all the processing power of, say, an IBM Roadrunner supercomputer inside a 19-inch box and make it run on about 60 kilowatts of electricity, the government wants to talk to you.

    And just as soon as they go back to loving and protecting freedom, then and only then will the government deserve my help with anything.

    1. Re:Yeah sure by rtyhurst · · Score: 3, Informative

      19 inch box?

      The IBM Roadrunner:

      "occupies approximately 6,000 square feet..."

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

      Good luck with that...

    2. Re:Yeah sure by Anonymous Coward · · Score: 0

      19 inch box?

      The IBM Roadrunner:

      "occupies approximately 6,000 square feet..."

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

      Good luck with that...

      Yeah, the physics/logistics sure did have a lot to do with GP's comment about whether it is moral to do any favors for a government that has abandoned the good principles it once stood for. Oh, wait, no they don't.

      Next time you want to be noticed by responding to a post that's near the top of the discussion, please choose one that is related in some way to what you want to say. I really am asking - if I wanted to rant and rave and give you a hard time I could do much worse than this. It's just that, for anyone who is not too oblivious to notice it, the rather blatant attempt to get attention distracts from the point you were trying to make.

    3. Re:Yeah sure by Luke+has+no+name · · Score: 3, Insightful

      17" tower? 3.8 GHz?

      I'm sure the thinkers of 1941 would be shocked to know what we can do now, given they were running 10 Hz on this:

      http://en.wikipedia.org/wiki/Z3_(computer)

    4. Re:Yeah sure by TCM · · Score: 5, Funny

      19 inch box?

      They didn't say how high.

      In other news, progress on a space elevator has been confirmed. Curiously, it's 19 inch wide.

      --
      Of course it runs NetBSD. BTC: 1NT7QvbetmANwaMzhpVL6
    5. Re:Yeah sure by Anonymous Coward · · Score: 0

      I find your ideas fascinating. Where, pray tell, may I subscribe to your newsletter?

    6. Re:Yeah sure by russotto · · Score: 2, Interesting

      And just as soon as they go back to loving and protecting freedom, then and only then will the government deserve my help with anything.

      Yeah, my first thought on this was whether perhaps those were the requirements to get the things inside every AT&T-style NSA listening room.

    7. Re:Yeah sure by Anonymous Coward · · Score: 0

      I suppose you can throw a Roadrunner into a black hole, to squeeze the space...just keep it on a leash so it stops before crossing the horizon...

    8. Re:Yeah sure by catdriver · · Score: 2, Interesting

      Exactly.

      It's totally unscientific, but I just ran Xbench on my 2009 Mac mini and got around 3 GigaFLOPS. It's not very accurate, but probably good to within an order of magnitude.

      According to Wikipedia's supercomputer article, that compares roughly with a 1985 Cray-2, which cost about $25 million at the time and was the size of a large closet.

      All we have to do is wait about 25 years.

    9. Re:Yeah sure by marafa · · Score: 0

      19 inch box
      they dint say if this was 19 inches wide or 19 inches high.
      so i presume this could be 19 inches high and 500 feet wide and still meet those requirements ;)

      --
      _ In Egypt Networks: Network Solutions with a Twist
    10. Re:Yeah sure by Duckie01 · · Score: 1

      Yeah and on top of that... 50 Gflops per watt ...

      The IBM Roadrunner:
      444.94 megaflops per watt

      So... to meet the DARPA goal... you'd only have to be more than 100 times as efficient.

      I don't think it's going to happen this year ;-)

    11. Re:Yeah sure by gtall · · Score: 1

      So, in your opinion, the entire government is devoted to unloving and unprotecting your freediom, eh? How about the NSTA, the fellows who investigate transportation accidents? They are taking away your right to freely die due to a repeat error. How about the Coast Guard? They'll be taking away your ability to completely die due to operator stupidity. Errr...NSF...taking away your rights to no money for science. NIH...clearly out to fund medical research allowing you to die without a remedy to stop it. FDA...stopping you from taking those 'cures' which Jim Bob's Drug and Beanery has been producing for your ailments.

    12. Re:Yeah sure by railmeat · · Score: 1

      And just as soon as they go back to loving and protecting freedom, then and only then will the government deserve my help with anything.

      Why would any government be interested in freedom?

    13. Re:Yeah sure by EbeneezerSquid · · Score: 2, Insightful

      If they thought it would happen this year, DARPA wouldn't be interested.

      Defense Advanced Research Programs Agency

      These are the guys who fund the crazy stuff, like robotic exoskeletons (Starship Troopers), Electronic Telepathy (via radio/net), and more.
      NOTHING they fund is expected to bear fruit quickly, but when it does bear fruit, that fruit is like gold (case in point, a little thing once called ARPANET.

    14. Re:Yeah sure by EbeneezerSquid · · Score: 1

      Teach me to post without previewing.
      d'oh! *facepalm*

    15. Re:Yeah sure by EbeneezerSquid · · Score: 2, Interesting

      Contrary to your apparent view, the vast majority of the individuals in the US Government do love freedom and wish to protect it.
      The defense department tends to be the place that these people are most concentrated: the same Defense Department that DARPA serves.

      I doubt that ANYONE in the government does not Love and wish to protect Freedom:
      There are those, however, who may undermine it unintentionally due to a lack of understanding of what their "payment to supports," will actually cause in the long run.
      This is called "Being naive".
      With luck, it can be corrected.
      With time, any damage can be reversed, as long as those who understand what has happened and do truly Love Freedom persevere.

      Complaining and Blaming others accomplishes little.

    16. Re:Yeah sure by selven · · Score: 1

      Square feet? Computers are three dimensional.

    17. Re:Yeah sure by ScrewMaster · · Score: 1

      I suppose you can throw a Roadrunner into a black hole, to squeeze the space...just keep it on a leash so it stops before crossing the horizon...

      Nah. Just build it inside a TARDIS. Problem solved.

      --
      The higher the technology, the sharper that two-edged sword.
    18. Re:Yeah sure by Anonymous Coward · · Score: 0

      No, you've underestimated how big a leap of technology this is...

    19. Re:Yeah sure by mdwh2 · · Score: 1

      Everyone loves freedom. The problem is that some have a different definition, e.g., "We have to lock these people up without charge, to protect our freedom!"

    20. Re:Yeah sure by Duckie01 · · Score: 1

      Yeah I knew all that, i still call it bullshit though. Nice you mention arpanet, because well it's nifty and all... but it didn't require science that wouldn't be available for at least the next decade.

      According to wikipedia they took a Honeywell DDP-516 and turned it into a router.  It had 24kb of memory... nice for the time and all but not exactly mind blowing science.  They got existing hardware to do what they wanted.

      It was all about ideas... creating protocols and implementing them.  Twelve companies submitted bids and BBN won the contract in 1968.  WITHIN A YEAR, arpanet was created, consisting of 4 IMPs!  Funny, isn't it!?

      A very nice accomplishment.  And look how it's growing!  It looks like everybody is using the internet nowadays.  But that won't mean I'll grant DARPA a hero status and believe they can wave their funds like a magic wand and defeat Moore's law like that.

      I call it bullshit, vapor, publicity stunt, or a joke at best.

  4. These aren't normal scientists... by Karganeth · · Score: 4, Funny

    ...They're EXTREME scientists!

  5. 57KW air-cooled 19" Rack? by lobiusmoop · · Score: 1

    You must be joking. That's like packing in 30 2KW electric fan heaters into a rack, obstructing the airflow with a ton of other junk and praying it won't melt. Good luck with that.

    --
    "I bless every day that I continue to live, for every day is pure profit."
    1. Re:57KW air-cooled 19" Rack? by CarpetShark · · Score: 1

      A KW of heating power is a lot different from a KW of processing power I imagine. Especially if some of that power is for refridgeration.

    2. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      There are 15 KW rated routers that are rack sized, and they only suck air in at the bottom through a small hole.

      Google CRS-1

    3. Re:57KW air-cooled 19" Rack? by Brian+Gordon · · Score: 1

      It's not so far fetched I think. Look at those tiny tiny little cores on your processor die. No, that's the cache, look harder. Wayyy down there.

    4. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 1, Funny

      If it's Cisco, they suck a lot more than that.

    5. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      A KW of power ends up as heat, no matter what you do with it. That's why opening the fridge door doesn't cool the room. Read up on something we like to call the laws of thermodynamics.

    6. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      Thats why its called Research.

    7. Re:57KW air-cooled 19" Rack? by CarpetShark · · Score: 1

      Yes, but we're not talking about a room, we're talking about the confined space of a rack. Nothing was said about not being able to produce heat, as long as it gets out of the case before the hardware fries.

    8. Re:57KW air-cooled 19" Rack? by CarpetShark · · Score: 1

      IANAP, but... how much of that 15kw is used locally in processing though, versus transmitted as energy/signals to another network node?

    9. Re:57KW air-cooled 19" Rack? by CarpetShark · · Score: 1

      Never mind actually, someone in another thread has pointed out the obvious: that this is a rack, not 1u of a rack :) 57kw in a rack is probably quite doable.

    10. Re:57KW air-cooled 19" Rack? by kenh · · Score: 1

      Sure it is, just shove 42 1,500 Watt 1U power supplies in a standard cabinet.

      --
      Ken
    11. Re:57KW air-cooled 19" Rack? by John+Hasler · · Score: 1

      99.99...%

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
    12. Re:57KW air-cooled 19" Rack? by John+Hasler · · Score: 1

      I think it is doable. The whole machine must be air-cooled but nothing in the RFP says that liquid cooling could not be used internally.

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
    13. Re:57KW air-cooled 19" Rack? by Repossessed · · Score: 1

      Expended energy always becomes waste heat. There's not much difference between drawing 500 watts in a space heater and drawing it with some other appliance from a heating perspective unless the appliance absorbs energy in some way (such as boiling water).

      Even with a fridge.

      --
      Liberte, Egalite, Fraternite (TM)
    14. Re:57KW air-cooled 19" Rack? by Dutch+Gun · · Score: 1

      You know... all they have to do is wait 10 years and they can just pick up a few Playstation 5's or Xbox 4's from Wal-Mart to do the job. In case they didn't realize, all those nifty bullet-points are highly desirable things for computers in both the consumer and commercial markets as well:

      * Highly efficient energy usage
      * New systems and programming models to develop for massively concurrent processing
      * Highly fault-tolerant

      Advanced computer technology is something that the market is pushing ahead at a phenomenal rate anyhow, and specifically, computer gaming is driving it even faster. I don't think the defense department really has to push development of this to advance the state of the art - it will happen with or without their incentives.

      --
      Irony: Agile development has too much intertia to be abandoned now.
    15. Re:57KW air-cooled 19" Rack? by GaratNW · · Score: 2, Insightful

      Exactly. There are always incredulous responses to this kind of challenge. Everything is impossible. Until it's not anymore. That's research, that's progress. There's no better way to get people to innovate on crazy shit then to tell them it's almost impossible.

    16. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      Heh, you'd be amazed at how many people leave the fridge open to cool down a room.

      Well, I guess somebody has to vote democrat, right ?

    17. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      actually if you're talking about the register file, instruction decode and fugs vs the memory controller, dram, high speed serdes, and cache its probably closer to 10

    18. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 1, Insightful

      That can't be true, unless you're taking the extreme long view. Let's say that you raise a 1 ton block 10 feet using a block and tackle and your own muscles. You'll certainly generate heat, but, when you're done, you'll have something like 30,000 joules stored as potential energy. Until it's released, it's not going to turn to heat.

    19. Re:57KW air-cooled 19" Rack? by inKubus · · Score: 1

      Exactly, and you could use something with a high density like molten lead in titanium pipes to move the heat off the chips. And by chips I mean a single piece of silicon 19" wide and 24" deep. That's going to be the only way to do it with today's tech.. I mean, according to this, the PowerXCell 8i die size is 212mm^2 and they do about 100GFLOPS each. So to hit 1PFLOP, which is 1000000GFLOPS, you'd need 10000x212mm or 212000cm^2 or 2120m^2 of silicon. The aforementioned 19x24" chips would be 2207cm^2 each so you'd need about 100 chips. You could probably do 100 in a rack. Each of these chips would be dissipating 9.6KW for a total of 960KW/PETAFLOP.

      Starting with that as a basic design, you could build on it to increase the FLOPS per mm. And it'll have to start using signals in the EHF to FIR range, which means new, faster switches are needed. Carbon nanotubes will probably be present on top of the silicon. The silicon will probably be a support structure for the optical processors, probably. You could probably get to their specs in 5-10 years with this plan.

      --
      Cool! Amazing Toys.
    20. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      A KW of heating power is a lot different from a KW of processing power I imagine. Especially if some of that power is for refridgeration.

      LOL, ultimate thermodynamics fail.

      A hairy guido named Joules is going to break your kneecaps now.

    21. Re:57KW air-cooled 19" Rack? by inKubus · · Score: 1

      Incidentally, 960KW = about 909 btu's per second. 1 ton of refrigeration can move 3.333 btu per second meaning you would need 273 tons of cooling for the rack just to keep up.

      --
      Cool! Amazing Toys.
    22. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      Not really, if the heat output from the refrigeration unit is not in the rack itself where the rest of the heat is. The point is to dissipate the heat, because the same amount of energy over a larger area means a lower temperature, so refrigeration energy will go into moving the heat to another area and will only be neutral if you are looking at a much larger system.

    23. Re:57KW air-cooled 19" Rack? by John+Hasler · · Score: 1

      > Incidentally, 960KW = about 909 btu's per second

      Where did you get 960kW?

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
    24. Re:57KW air-cooled 19" Rack? by John+Hasler · · Score: 1

      No need for such extreme measures.

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
    25. Re:57KW air-cooled 19" Rack? by Anonymous Coward · · Score: 0

      The point is those dies are only a few hundred microns thick. And for its very tiny volume, it consumes a huge volume of cooling resources (heat sinks, fans, airflow space). The problem is improving the ratio.

    26. Re:57KW air-cooled 19" Rack? by EbeneezerSquid · · Score: 1

      Well, I guess a form of compact cooling would also be part of the project.
      Or, more likely, part of another DARPA project that has been underway for awhile.
      After all, don't you think that soldiers in full Chem-Warefare gear in the desert (be it Nevada, Iraq, or any other) would perform better if they weren't just about to fall over from the heat?
      I do. And if it can cool a Soldier's Uniform, many of them could no doubt cool a server.

    27. Re:57KW air-cooled 19" Rack? by EbeneezerSquid · · Score: 1

      At high frequency ranges, the separation between wires begins to act as a capacitor, resulting in signals jumping between contacts. Unless you are planning on using waveguides.
      As Intel discovered in the P4 generations, upping the clock frequencies is an impractical, low FLOPS/W solution.
      And, using your math, the yield is only 10MFLOPS/W, versus the stated goal of 50GFLOPS/W.
      I think the researchers would be better off going the Centrino/Core Duo Route.

    28. Re:57KW air-cooled 19" Rack? by Odemia · · Score: 1

      You are right to point out that the cache is a major problem.

      BUT, you try writing a compiler for a multi core system withOUT all those huge caches. And then try to tune a random selection of non trivial programs to maximize their throughput to get 50GFLOPS. And if you manage that you are going to have some serious heat/density/power issues as your tiny little cores are placed side by side all using dynamic logic to maximize their throughput

      If you think the answer is a larger core just ask Intel, IBM or AMD why they have moved to multi-core systems in such a strong way. The simple answer is performance to cost, in other words more complex cores are to difficult to design and per core performance diminishes with complexity (ie too much power/heat for not enough performance).

      The goal set by DARPA is meant to be manageable but not easily. Certainly not as easy as "look at those tiny tiny little cores".

    29. Re:57KW air-cooled 19" Rack? by Brian+Gordon · · Score: 1

      I wasn't saying cache is a problem; I was merely emphasizing the small size of the actual logic

  6. Could at least editors have a look at TFA? by Gruturo · · Score: 4, Informative

    It's a 19" _rack_, not _box_. As in, the standard (non-telco) datacenter rack size, accomodating up to 42U, 19" wide.

    --

    Vacuum cleaners suck. Kings rule.
    1. Re:Could at least editors have a look at TFA? by eliphalet · · Score: 2, Funny

      How tall does this rack have to be?

    2. Re:Could at least editors have a look at TFA? by Anonymous Coward · · Score: 0

      42AU? That's a tall order.

    3. Re:Could at least editors have a look at TFA? by Gruturo · · Score: 2, Informative

      How tall does this rack have to be?
      Typically 42 Rack Units, as I said in the original post. A rack unit is 1.75" (and we use them even here in Europe so at least servers fit in racks, fortunately :-) ) so this makes the standard rack able to contain a little over 6 feet worth of hardware, or 185ish cm. Of course the rack itself is usually a bit taller since it has a base and some fans on top (let me stress: usually).

      --

      Vacuum cleaners suck. Kings rule.
    4. Re:Could at least editors have a look at TFA? by Kratisto · · Score: 1

      Are you trying to say that Extreme Scientists have nice Racks?

      --
      Conscience is the inner voice which warns us that someone may be looking.
    5. Re:Could at least editors have a look at TFA? by Anonymous Coward · · Score: 0

      Some 19" boxes are 1U high, 2U high, 6U high. They're just asking for one that's 42U high. Maximum. I'm sure if you could give them a 6U version, they'd be happy with that also.

    6. Re:Could at least editors have a look at TFA? by Anonymous Coward · · Score: 0

      Are you trying to say that Extreme Scientists have nice Racks?

      Of course extreme scientists have nice racks! Because they're EXTREEEEEEME!

      Oh, sorry. Having a Buzzed Bunny flashback. Will go dunk my head in water before I start to get graphic. EXTREEEEEEMELY graphic.

    7. Re:Could at least editors have a look at TFA? by Smallpond · · Score: 2, Insightful

      hmmm... fits in one rack and has enough processing to do word recognition on all of the calls coming in to one telephone central office simultaneously. I wonder what they want a whole bunch of these for?

    8. Re:Could at least editors have a look at TFA? by kenh · · Score: 1

      Naw, how deep can I make the cabinet ;^)

      --
      Ken
    9. Re:Could at least editors have a look at TFA? by Anonymous Coward · · Score: 0

      The editors simply copied the first line of TFA.
      Maybe you should write the author and complain.

    10. Re:Could at least editors have a look at TFA? by AHuxley · · Score: 1

      In house Narus for the rest of the US gov?

      --
      Domestic spying is now "Benign Information Gathering"
  7. No problem by Anonymous Coward · · Score: 5, Funny

    Just stick a human brain in a bucket. It's small, quiet, cool and just feed it a Cheeto every once in a while to keep it running.

    1. Re:No problem by Jeremy+Erwin · · Score: 2, Funny

      Well, since Darpa also wants the operating system to be self aware, that's a start.

    2. Re:No problem by oneirophrenos · · Score: 5, Interesting

      Just stick a human brain in a bucket. It's small, quiet, cool and just feed it a Cheeto every once in a while to keep it running.

      And since the human brain has a computational power of 100 petaflops at 20 watts, it'd well exceed DARPA's requirements.

    3. Re:No problem by Kjella · · Score: 4, Funny

      And since the human brain has a computational power of 100 petaflops at 20 watts, it'd well exceed DARPA's requirements.

      Only before you subtract the power used to think about pr0n. Or you can use a female brain, it'll no nuclear on you only once a month.

      --
      Live today, because you never know what tomorrow brings
    4. Re:No problem by dkleinsc · · Score: 2, Funny

      What happens if we use an Ab Normal brain?

      --
      I am officially gone from /. Long live http://www.soylentnews.com/
    5. Re:No problem by Runaway1956 · · Score: 1

      Uhhhmm, removing the female brain from the female body precludes the nooclear monthly. Or, is that the monthly nooclear? Whatever - no hormones, no nookie.

      --
      "Windows is like the faint smell of piss in a subway: it's there, and there's nothing you can do about it." - Charlie Br
    6. Re:No problem by Anonymous Coward · · Score: 0

      Unfortunately, it would waste too much processing time watching re-runs of "The Facts of Life" on TV Land.

    7. Re:No problem by Anonymous Coward · · Score: 2, Insightful

      It's interesting how the "estimated computational power of the brain" seems to increase every few years to keep it very far ahead of the fastest supercomputer of the time.

    8. Re:No problem by weicco · · Score: 1

      Unless we are talking about these who an earth would like to stand in 19" rack for the whole business day!

      --
      You don't know what you don't know.
    9. Re:No problem by Anonymous Coward · · Score: 0

      You just have to choose a retarded brain then.

  8. Going against Resistance is futile. Sidestep it. by jack2000 · · Score: 3, Interesting

    Supercool that fucker! That might help a lot!

  9. 60,000 watts? by retro128 · · Score: 1

    Is that all they're allowing? Power nazis.

    --
    -R
  10. Simple solution by Anonymous Coward · · Score: 0

    1. Buy EEE PC
    2. Soak it in fluid herbal viagra sold by Mr. Hammand Zo-Moki Gabbaballa on the internet
    3. ???
    4. Profit !

  11. Where will be that cabinet? by gmuslera · · Score: 1

    If could be put at near 0 K (and the power to maintain that temperature is not counted) maybe a superconducing supercomputer could get that speed in that size.

    1. Re:Where will be that cabinet? by Anonymous Coward · · Score: 0

      No. "Ubiquituous" and supercomputer in the same line means only one thing: onsite cryptanalysis in datacenters, trying to read your pgp-encrypted mail. The NSA doesn't care about extreme temperature superconductivity; it cares about putting these in racks across the world to read data like crazy.

    2. Re:Where will be that cabinet? by Anonymous Coward · · Score: 0

      Along with a supercolliding superbutton for emergency power off?

    3. Re:Where will be that cabinet? by EbeneezerSquid · · Score: 1

      NSA is not part of the Defense Department.
      Doesn't mean they won't reap the benefits of this DARPA project, but it's unlikely that they are making it for them.

      I can think of a heck of a lot of things that the DoD could use this tech for: reading your PgP emails isn't one of them.
      Reading Ahmadinejad's, Kim's, or BinLaden's on the other hand would be very useful.

      But just off the top of my head, what could we use such a powerful, compact computer for? Hmmmm. . .
      *Server consolidation for field command-control centers
      *Tactical and targeting computers for various vehicles (Tanks, aircraft, etc)
      *Control systems for lighter, more mobile Ballistic Missile Defense systems (The present ones are mounted on converted Oil Derricks - Be nice to mount one on, say, a truck. Or a Destroyer.)

      I could probably think of half-a-dozen more if I wasn't pulling a 36-hour-plus day right now in an attempt to get my sleep schedule back on track after two weeks of midshift.
      But since I can't, I'll leave that as an exercise for you.

      Just remember: The US Military does not normally operate domestically, and is forbidden to take action against US Citizens unless actually under combat conditions without Presidential Authorization, which is only given on a case-by-case basis.
      If the boys in the Pentagon Basement are asking Obama if they can read your email, You are doing some seriously BAD shit.

  12. 19" tower and space elevator in one by davidwr · · Score: 5, Funny

    Two! Two! Two projects in one!

    --
    Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
    1. Re:19" tower and space elevator in one by CarpetShark · · Score: 1

      Two! Two! Two projects in one!

      So seven projects then?

    2. Re:19" tower and space elevator in one by Anonymous Coward · · Score: 0

      <the count>Two! Two projects, ah-ah-ah!</the count>

  13. Bio-computing perhaps? by davidwr · · Score: 1

    Perhaps with bio-computing or near-atom-scale computing, with the equivalent of transistors being not much bigger than a baker's dozen standard-sized buckyballs.

    But with conventional technology? I think Moore's law will break down before we reach this goal.

    The brain-in-a-bucket comment earlier is probably more insightful than funny in this context.

    --
    Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
    1. Re:Bio-computing perhaps? by EbeneezerSquid · · Score: 1

      Since when has DARPA been conventional?

      Course, I seem to remember reading about an artificial diamond company which was planning on scaling up to make diamond semiconductor wafers. No idea what how different diamond lithography would be to Silicon, but it would at least have a much larger thermal envelope, allowing higher processor loading, all else being equal. . .

      Semi-conventional perhaps?

  14. If i could do that by nurb432 · · Score: 1

    Do you think id admit it and have the Feds take it from me for nothing and classify it? No thanks.

    And for the record, it wouldn't be that hard to do, as long as you wanted a semi-dedicated supercomputer and not a general purpose box.. But no, i wont tell you how, even if i was authorized.

    --
    ---- Booth was a patriot ----
    1. Re:If i could do that by Anonymous Coward · · Score: 0

      They don't take it from you. It is very well paid for.

  15. Thermionics and stuff by DynaSoar · · Score: 1

    Pack thermionic converters between the components. They'll help cool and recover some power from heat back to power. They can be on the board, or placed on a cover over it in such a way as to fit between the board components. http://web.mit.edu/newsoffice/2001/electricity-1205.html

    Build in parallel processing with 16 processors, 4 on each side of a 4D-cube, as in the Connection Machine http://en.wikipedia.org/wiki/Connection_Machine

    Three boards, stacked. Top, thermionics on the underside fitting between the main components, top side of the board is keyboard. Main in the middle, components top side. Bottom board, cram full of memory, below main board to keep it away from the heat. Vents underneath and through memory and main boards, so convection can feed heat to the top board.

    --
    "I may be synthetic, but I'm not stupid." -- Bishop 341-B
    1. Re:Thermionics and stuff by jank1887 · · Score: 1

      thermionics or thermoelectrics: either cool or recover power. you won't do both. And you'll add more net heat in the box if used for cooling. If you can keep that heat away from critical components, there may be a net benefit. If you use it to recover power, it'll have to be in the thermal path and will increase the thermal bottleneck in getting heat out. So you're recovered power comes at the cost of hotter heat generating components.

    2. Re:Thermionics and stuff by EbeneezerSquid · · Score: 1

      Also, I think they want a total power of 50GFLOPS/W, including the power used for cooling.
      If I was planning to put this rack in a tank or a ship, that's how I would write the spec.

  16. DUH PC by OverZealous.com · · Score: 1

    I think they should include DARPA in the acronym, so you get to spend all this money on a DUH PC.

    Maybe it could even run WinDUHs!

    It would be an induhspensible part of our computing future, duhtermining the ability of our government's uhbility to duhrive new induhstries!

  17. I'm In: +1 , Interesting by Anonymous Coward · · Score: 0

    "If you can squish all the processing power of, say, an IBM Roadrunner supercomputer inside a 19-inch box and make it run on about 60 kilowatts of electricity, the government wants to talk to you."

    I can; however, BEFORE we talk, I'll need to have Euro
    100,000,000 deposited in my bank account and the same
    amount AFTER the discussion so they can proceed with their new WEAPONS DESIGN PROGRAM.

    Yours In Condensed Matter,
    Kilgore Trout, Physicist (Mad)

  18. Re:wtf by poopdeville · · Score: 1

    Yes, please tag this story "wheresmypony".

    --
    After all, I am strangely colored.
  19. NVIDIA by DaMattster · · Score: 1

    What about the NVIDIA Cuda architecture? They claim it is a super computer for under 10 grand and doesn't require special power requirements. But, I wonder if it will only perform as a super computer for graphics ....

    1. Re:NVIDIA by stevelinton · · Score: 2, Interesting

      I imagine they will build something along those lines. Lots of highly specialised cores that can do Floating Point really well if it carefully compiled for them; some switches for some fast short-range network protocol probably and a few general purpose cores to manage things. Maybe some field-programmable components so that you can customise the hardware for new applications. The current nVidia Tesla series achieves around 1GFLOP per Watt, and you can get 1 TFLOP, consuming 1 KW per U, (ignoring host processors and many other things), so they're looking at roughly a 50 fold improvement by designing for HPC from the ground up, rather than graphics first and HPC as a side-show. That, plus a couple of generations of Moore's law doesn't sound too improbable.

    2. Re:NVIDIA by Jeremy+Erwin · · Score: 1

      You're missing the point-- Darpa feels that it is ill served by current commodity supercomputers, and wants something revolutionary. The deadline for delivery is in 2017, so it's unlikely that today's tech comes close.

    3. Re:NVIDIA by Jeremy+Erwin · · Score: 1

      Careful Compilation?

      â Develop new technologies and execution models that do not require application programmers to explicitly manage system complexity, in terms of architectural attributes with respect to data locality and concurrency, to achieve their performance and time to solution goals - programmability.

      Billion way parallelism is also mentioned. So. It has to be easy to program

    4. Re:NVIDIA by Anonymous Coward · · Score: 0

      Yehaa, Stanford's Merrimac has a change of becoming a reality in a modern form!

    5. Re:NVIDIA by EbeneezerSquid · · Score: 2, Interesting

      Not necessarily easy to program.
      It just cannot require explicit knowledge of the system architecture in order to program, like the old mainframes did.
      Of course, there is something to be said for explicitly managed systems. A mainframe with 512kbytes of memory ran the air defense of the United States from the 1970s until 2004 (well, three of them). Why wasn't it replaced earlier? Because they tried to, four times, with general purpose computers but, until 2004 (and a dozen-or-so Opterons), they couldn't handle the load.

      But the military no longer trains many programmers. And hiring Contractors (or G.S.'s) to program for explicitly managed systems is very, very expensive.

  20. Comment removed by account_deleted · · Score: 2, Funny

    Comment removed based on user account deletion

  21. Should it... by juanergie · · Score: 1

    ... run Linux?

    --
    Aeroespacio.org
    1. Re:Should it... by John+Hasler · · Score: 1

      > ... run Linux?

      It probably will.

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
    2. Re:Should it... by EbeneezerSquid · · Score: 1

      No, It probably won't.
      For some reason, DoD always goes for a proprietary flavor of Unix.
      And not the same flavor either. A wonderful mishmash of Solaris, HP-UX, and others.
      What's really nice is that in tech school they give you a 3-day lecture on basic Unix commands, and tell you "You'll never use this".
      Then you'll get on-station, and have to install one from scratch.

  22. Will fit inside your Car Analogy by sabre86 · · Score: 4, Insightful

    This combination of power required and volume would allow essentially for current day supercomputer in every single military vehicle, assuming the weight and heat exhaust constraints aren't too onerous. 60 kW is about 80 horsepower and even a 19 in x 19 in x 19 in cube is only about 4 cubic feet*, which is less than than the trunk space on a Mazda Miata (5.1 cubic ft for a 2006 model), so it's within the space-power envelope of a small sports car, albeit the engine would need to be uprated some to account for the power drain.

    Having such great computational power available to every single vehicle would open up a huge realm of possibilities: Combine it with sensors you could detect damage and minimize its effects by comparing the vehicle's response to a detailed finite element model. You could do on the fly aerodynamic analysis, allowing a fighter to keep performing to it's best even after damage has significantly altered it's shape. You could manage the control of thousands of actuators, allowing you to create a shapeshifting walker out of programmable matter, and you could definitely do learning/optimization algorithms that would allow for an AI capable of a significant amount of learning. Combine this with the amount of image processing it could do, and you're very near a completely autonomous, smart enough combat vehicle.

    While it's a too big for a man portable system, with work, you could fit such a device (and a power source) into something as small as a motorcycle or a somewhat scaled up iRobot Warrior. That's not much more than man sized. It may not be a T-800, that much computation in that small size and power envelope is enough build a near-man sized autonomous fighting vehicle that can see, learn and adapt with an endurance on gas of several hours. It's a bit frightening to consider.

    --sabre86

    1. Re:Will fit inside your Car Analogy by JobyOne · · Score: 1

      But where, exactly, would the batteries that can push 60 kilowatts go? I don't think they would fit in the trunk of a Mazda Miata with this magical imaginary computer.

      --
      Porquoi?
    2. Re:Will fit inside your Car Analogy by sabre86 · · Score: 2, Informative

      But where, exactly, would the batteries that can push 60 kilowatts go? I don't think they would fit in the trunk of a Mazda Miata with this magical imaginary computer.

      Or more importantly, batteries that can push 60 kW for any period of time. I think that with enough cells, which you can make about as small as you want, you might get the power, but you definitely won't have the energy to run it for anytime whatsoever. The energy density is nowhere near good enough. But, Sticking with the Miata example, there's easily enough power under the hood to drive both the car and the computer, particularly with a high output option like the BPT. You just need a generator, like the 53 kW version in the Volt. For an automotive sized and powered vehicle, using year 2000 level power and materials technology, you could easily add such a computer and all it's benefits. That means effective, mobile, car sized autonomous fighting vehicles (since this is a DARPA project, I'm considering the military applications first) are extremely easy if you have this kind of computer, and motorcycle/Terminator sized units are probably possible, just using gas burning engines -- no advanced technology except the computer.

      I apologize that wasn't clear from the original post.

      --sabre86

    3. Re:Will fit inside your Car Analogy by russotto · · Score: 1

      But where, exactly, would the batteries that can push 60 kilowatts go? I don't think they would fit in the trunk of a Mazda Miata with this magical imaginary computer.

      I have a battery rated for 62.5A @11.1V. I could easily fit 100 of them within the trunk of my Mazda Miata, with lots of room to spare, which gets you about 70kW. Granted, for only two minutes, but in fact a few hundred could fit in the Miata...
      (the battery is 16mmx44mmx141mm and weighs about 200g. )

    4. Re:Will fit inside your Car Analogy by Anonymous Coward · · Score: 0

      It's not necessarily an optimal technology, but for reference-
            various sources claim that an NVIDIA 8800 GTX can peak at 550 GFLOPS using about 150 watts of power.
          That would be about 15 watts for 50 GFLOPS - 15x more power consumption than the goal.
          I also figure that you might be able to mount up to 16 of these (plus connecting logic) on a 19" square board. If you stack the boards in 2" intervals (to allow for cooling) for a total computing power of 1.9 petaflops, it would fit in a rack 32 feet high - 5x too big to be a single 19" rack, 20x too big to fit in a 19" cube.

            The IBM Cell has 8 SPEs with 32 GFLOPS each, requiring 135 watts (1.9 GFLOPS/watt); this is 26x more power consumption than the goal. If you put 16 on each board, 2" apart, your rack would be 77 feet high. It looks like IBM contemplated putting 32 SPEs on each chip at one point; if they pulled that off, your rack would only have to be 20 feet high.

      And that's assuming that interconnects don't matter (that everyone just sits and works on their own little bit of data).

    5. Re:Will fit inside your Car Analogy by jeffliott · · Score: 1

      I'd prefer to do without Skynet if you don't mind!

    6. Re:Will fit inside your Car Analogy by EbeneezerSquid · · Score: 1

      I was going to say DARPA already has a project working on this, but I was wrong.
      They have three (or more).

      DARPA's ongoing research is listed on their site. Have a look.
      http://www.darpa.mil/off_programs.html

    7. Re:Will fit inside your Car Analogy by temojen · · Score: 1

      60 kW is about 80 horsepower

      And if you had the most efficient type of diesel engine and generator, that's about 11L of diesel fuel per hour per vehicle. All the time, not just when accelerating of climbing a grade. And between the generator and computer, a 100kW heat signature (computers do minimal "work", almost all of their consumed energy ends up as heat).

      19 in x 19 in x 19 in cube is only about 4 cubic feet*

      They're talking about 19" rack, not cube. That's about 150 cubic feet.

      Combine it with sensors you could detect damage and minimize its effects by comparing the vehicle's response to a detailed finite element model. You could do on the fly aerodynamic analysis, allowing a fighter to keep performing to it's best even after damage has significantly altered it's shape.

      CF-18's already do that, with no super computer.

      I don't think this project is intended for vehicle mounted use.

    8. Re:Will fit inside your Car Analogy by sabre86 · · Score: 1

      Good post.

      I misunderstood about the size of the box. Thanks for the correction.

      I think your points about the fuel use, thermal signature and -- most importantly -- volume, probably push the system requirements out of the small car range. 11 L an hour is still enough for maybe 3 to 5 hours for a car sized fuel tank, so you'd have to have a good support system, but I think we're definitely pushed out of the motorcycle range and the volume puts at a vehicle not any smaller than a minivan, but I think a minivan is big enough -- apparently the interior volume of a Dodge Caravan is about 160 cubic feet. So we're talking more like an APC sized Terminator at the smallest, but it's definitely doable in the attack helicopter, main battle tank, fighter jet range. I'd say a 150 cubic foot volume and 100 kW power requirement/thermal signature increase is easy for them to work in.

      I'm talking about something beyond a Hornet's flight control system. This would be recomputing the plant matrix for any arbitrary aircraft shape rather than just adjusting the gains of an adaptive feedback controller. Also, you'd be able to search the entire control space on the fly (pun not really intended) for the maximum amount of performance given any arbitrary damage. If the Hornet can do that, please send me a link to how it does it. Because that is awesome.

      Anyway, thanks for well done and interesting post.

      --sabre86

  23. Alternate way to get one by gmuslera · · Score: 2, Funny

    Just stay around girls called Sarah Connor. A supercomputer of around that size will appear eventually, and you will take as bonus a portable nuclear reactor, and a somewhat aggressive AI. Be sure to erase memory because it surely will contain a nasty trojan horse.

  24. Heat by JobyOne · · Score: 1

    I can't imagine pushing 60 kilowatts through a 19" rack mount ANYTHING without EVERYTHING catching on fire.

    Seriously, that's a lot of electricity.

    --
    Porquoi?
    1. Re:Heat by InfiniteWisdom · · Score: 1

      That's 50A at 120V. A very common and standard config is to have two 20A PDUs in a rack, so 50A isn't that much of an increase. You do need an efficient cooling setup, but it's nothing that most commercial datacenters couldn't handle.

    2. Re:Heat by InfiniteWisdom · · Score: 1

      Actually, never mind. I was off by a zero there. That is quite a bit of cooling!

    3. Re:Heat by Pyrus.mg · · Score: 2, Insightful

      The downwash from the rotors on the black helicopter its mounted on should help.

    4. Re:Heat by John+Hasler · · Score: 2, Informative

      By my calculations 1 m^3/sec of air can carry away 65kW at a 50 degK temperature rise. That's doable, though you don't want it exhausting into your office.

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
    5. Re:Heat by Anonymous Coward · · Score: 0

      You might want to allow for a factor of 2 to allow for high elevation where air is thin and an additional factor of 2 for Theta JA for the components you are using.
      Mil-spec calls for max junction temperature of 105C and sometime your CPU vendors specs only 85C to meet timing.

  25. playstation 4

    --
    intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
  26. What's it going to be used for? An idea.. by Anonymous Coward · · Score: 0

    Looking at the general specs, it basically needs to be small, have low power consumption, and be programmable by current means - no special programming ability. If it were to be used here in the States, why the size and power consumption limitations? Why not allow for specialized programming? It's almost as if they need something that will be able to be used and maintained by a typical military technical grunt out in the field. As if it would be put in a back of a truck and run on a gas powered or solar generator.

    Just putting that out there.

    1. Re:What's it going to be used for? An idea.. by kenh · · Score: 1

      It would have to be where a network connection of any usable speed is impossible.

      BTW, just how big is a steady, 60 KWh solar panel? It appears to require several thousand (around 4,000 100 Watt panels in the mid-atlantic region (and suitable battery infrastructure) to generate 60 KWh for 24 hours a day (1.44 MegaWatt/hours for every 24 hours).

      The above is from http://www.batterycountry.com/ShopSite/sec.htm

      --
      Ken
    2. Re:What's it going to be used for? An idea.. by cryptoluddite · · Score: 1

      It would have to be where a network connection of any usable speed is impossible.

      Clearly this supercomputer is intended for an autonomous killing machine... a terminator, if you will. As such it doesn't not need a massive network connection, just enough to control some servos really. At 19" this component would compromise the torso of the machine.

      BTW, just how big is a steady, 60 KWh solar panel? It appears to require several thousand

      Perhaps, but this machine will no doubt will contain some type of nuclear power core.

    3. Re:What's it going to be used for? An idea.. by SpinyNorman · · Score: 1

      It would have to be where a network connection of any usable speed is impossible.

      Having a network connection to a traditional room-filling air-conditioned supercomputer isn't an alternative unless they only want a couple of these, which I doubt is the case (on a side note I'd guess the input bandwidth they need isn't that great - probably just real-time video/voice).

      I think it's more a matter of wanting to have this sort of computing power available in the field on a potentially widespread (not one-off) scale. They probably have in mind things like AI and advanced image/pattern recognition for robo-warrior type applications.

  27. Zombie bait by PolygamousRanchKid+ · · Score: 1

    Just stick a human brain in a bucket.

    Yeah, we tried that already . . . but brains in buckets tend to attract too many zombies . . . you end up spending way to much money on ammunition for gun shots to the zombie heads . . . though, the sysops seem to love the action.

    --
    Schroedinger's Brexit: The UK is both in and out of the EU at the same time!
  28. And a pony by Daniel+Weis · · Score: 1

    If you can squish all the processing power of say an IBM Roadrunner supercomputer inside a 19-inch box and make it run on about 60 kilowatts of electricity, the government wants to talk to you.

    Well then. I'm sure people will go with the more traditional routes of terrorism, theft, and tax evasion to get a one on one session with the government. After all, it just seems easier.

    1. Re:And a pony by EbeneezerSquid · · Score: 1

      No, Tax evasion just gets you a job.
      Setting tax policy.

  29. I have one by BountyX · · Score: 1

    My 19" laptop has a super man logo on it...

    --
    Trying to install linux on my microwave, but keep getting a kernel panic...
  30. It's a sad state of affairs... by HotNeedleOfInquiry · · Score: 0

    When the most important specification of a supercomputer is it's power consumption.

    --
    "Eve of Destruction", it's not just for old hippies anymore...
    1. Re:It's a sad state of affairs... by symbolset · · Score: 1

      Supercomputers, and computers in general, are becoming a more significant percentage of our total power usage. By demanding more efficient power for their supercomputer, DARPA are insisting in more power efficiency in general because their technology does "filter down" to the rest of us. So, if they get their efficient supercomputer, we'll get PCs that operate at under 3 Watts and give us the same utility your 1KW gaming monster does today.

      And that's going to impact the building of windfarms, hydro dams, salmon habitat, nuclear energy, coal fired plants, and a good number of other things. Good on 'em.

      --
      Help stamp out iliturcy.
    2. Re:It's a sad state of affairs... by EbeneezerSquid · · Score: 1

      When you don't want to have to put a nuclear reactor in the tank that the supercomputer is going to be shoehorned into. (As a higher-up poster suggested).
      I'm sure the Army would like all the IT/InfoWarfare abilities that comes with a prepared Command post in their field HQ, and still be able to run off of the truck mounted generators they use now.
      Power is a BIG concern in field operations, as you can't count on being able to tap into the local grid, as it may be unstable due to fighting (like Iraq, though it wasn't stable to start with), or entirely nonexistant (like Afganastan).

    3. Re:It's a sad state of affairs... by coopex · · Score: 1

      Ah, so Trickle Down Economics is a crackpot idea when conservatives use it, but if you're using it to argue for "Green" causes, it's now en vogue? I moved and the hassles with changing your address in the system means I don't get the memos for what fads are Morally Superior.

      --
      The road to hell is paved with good intentions.
  31. Simpler solution. by tomhudson · · Score: 1, Insightful

    Redefine a Gigaflop. Say 1 billion floating point instructions per century.

    Hey - It worked for hard disk manufacturers for gigabytes.

    It works all the time for food companies when they say something now has "only" X number of calories per portion, by making the portions something like "2 potato chips."

    It works for ISPs for "unlimited Internet access".

    It works for Microsoft for "most secure [insert whatever] ever."

    It worked for George "Mission Accomplished" Bush. Kinda ...

    It'll work for Barack "No tax increase for anyone making under $250,000" Obama. (okay, I'll give you that it's really doubtful for that one)

    Now where's my grant money?

    1. Re:Simpler solution. by Jah-Wren+Ryel · · Score: 2, Informative

      Redefine a Gigaflop. Say 1 billion floating point instructions per century.

      Gigaflop doesn't even have a time dimension.

      --
      When information is power, privacy is freedom.
    2. Re:Simpler solution. by sleigher · · Score: 1, Troll

      Are you sure?

      --
      All points of time and space are connected.
    3. Re:Simpler solution. by tomhudson · · Score: 4, Informative

      Redefine a Gigaflop. Say 1 billion floating point instructions per century.

      Gigaflop doesn't even have a time dimension.

      Are you on drugs? Sure it does: FLoating point Operations Per Second.

      Hint - they're looking for a machine that can do 50 gigaFLOPs. Such performance is always measured per unit of time. Same as 1 horsepower is 550 foot-pounds per second.

      If you google for it yourself, you can keep your beginners-level trainee deck swab geek card :-)

    4. Re:Simpler solution. by blueg3 · · Score: 3, Informative

      That's right -- gigaflops has a time dimension.

      Gigaflop, on the other hand, doesn't. One gigaflop is a billion floating-point operations. One gigaflops is a billion floating-point operations per second. Contrary to "obvious" rules of grammar, the "s" isn't pluaralization, it's the unit "seconds".

    5. Re:Simpler solution. by tomhudson · · Score: 1

      Don't sweat it - I'm still pissed off about kib, mib, etc. kib - what's that - a unit of dog kibble? "My dog eats 3 kibs a day."

    6. Re:Simpler solution. by blueg3 · · Score: 1

      It doesn't get any better if you write out the full term "kibibits". Thankfully, I've never had to say "gibibits".

    7. Re:Simpler solution. by Jah-Wren+Ryel · · Score: 1

      Yes, I am sure. You should have noted the lack of an 's' at the end of the word "gigaflop."
      I can't believe two responses have been modded up for such blatant illiteracy.

      --
      When information is power, privacy is freedom.
    8. Re:Simpler solution. by Jah-Wren+Ryel · · Score: 1

      If you google for it yourself, you can keep your beginners-level trainee deck swab geek card :-)

      Wow. Two uber-confident and completely incorrect responses in a row now, both citing the exact same text that should have been made it completely clear that without an 'S' in GigaFLOP there is no seconds.

      --
      When information is power, privacy is freedom.
    9. Re:Simpler solution. by Anonymous Coward · · Score: 0

      shouldn't it be Gigaflo then? seeing as the P stands for per?

    10. Re:Simpler solution. by Retric · · Score: 1

      They want 50GFLOPS/w so you start talking about GFLOP people assumed it's a typo.

      PS: Hand them something that does 50GFLOP per year / w and they will say it's out of spec because they said 50GFLOPS/W.

    11. Re:Simpler solution. by Jah-Wren+Ryel · · Score: 1

      shouldn't it be Gigaflo then? seeing as the P stands for per?

      No.

      --
      When information is power, privacy is freedom.
    12. Re:Simpler solution. by EbeneezerSquid · · Score: 2, Interesting

      No, as it is a different abbreviation.
      FLoating point Operations Per Second
      FLoating point OPerations
      Neither are really good acronyms, (That would be FPOPS and FPO), but they are the accepted terms.

      HOWEVER: It is all nitpicky geek-out, I'm-better-than-you, You're-So-Dumb-You-forgot-the-"S". B.S.
      Please.
      This is Slashdot, not middle school.
      hard to tell some times, I know.

    13. Re:Simpler solution. by tomhudson · · Score: 1

      Wow. Two uber-confident and completely incorrect responses in a row now, both citing the exact same text that should have been made it completely clear that without an 'S' in GigaFLOP there is no seconds. Everything in context, grasshopper - the context being the article, which uses the term "supercomputer" multiple times. Supercomputer performance is always measured in FLOPs per second. Otherwise, an old 8086 would qualify if you just left it on for enough centuries - hence the joke. Please turn in your geek card until you can get a humour transplant.

    14. Re:Simpler solution. by Jah-Wren+Ryel · · Score: 2, Informative

      Supercomputer performance is always measured in FLOPs per second.

      Sorry, as a guy who has built supercomputers (not the piddly CoW type either) and now consults on them on a daily basis, I can tell you that's not true. Common, but not an "always" worthy of bolding - dhrystones and whetstones and their modern versions in SPECint, SPECfp and their _scale counterparts and mcalpin's STREAM benchmark which reports bytes/s in addition to flops. Not every FLOP is created equal which is why more sophisticated measurements exist.

      In fact, it is precisely because of my experience that I recognized your error of using "gigaflop" in the first place and made fun of you for doing so, its an extremely common error. One that anyone who does work in the high performance computing arena would recognize off the bat.

      While you don't need a humor transplant, you do need an augmentation.

      --
      When information is power, privacy is freedom.
    15. Re:Simpler solution. by Anonymous Coward · · Score: 0

      PS: Hand them something that does 50GFLOP per year / w and they will say it's out of spec because they said 50GFLOPS/W.

      Not necessarily. FLOP's per time and FLOP's per watt are two completely different metrics.

      50GFLOPS/watt =
      50GFLOPS/(joules/sec) =
      (50GLFOP/sec) / (joules/sec) =
      50GFLOP/joules

      Thus FLOPS/watt is an inherently timeless benchmark.

      I'm sure the government's specs are at least a little more detailed than that though.

    16. Re:Simpler solution. by tomhudson · · Score: 1

      This still doesn't change the fact that all those measurements of performance (and even your current reply) have time as a component. Measuring computer performance requires, even in your latest example a (operation)/s.

      The context of the article (supercomputer with specific performance) and my joke (changing the denominator to achieve the desired result) also make it clear that we're measuring performance with a time component.

      Also, even a single FLOP has a time component. Nothing happens instantaneously. All actions require time in which to execute. Here are several different ways to look at it, all with the same result.

      Let's invoke a relative of the paradox of Zeno. Without some time component, you can't have any way of differentiating one instructions' execution from another - so you can't measure or even posit the existence of more than one FLOP without a time component somewhere, even if only as a hidden or assumed variable. cpu executes instruction. then at some future time, it executes another instruction. Now, remove the "at some future time" - how is any observer to know that they're not observing the same instruction? You can't "prove" the existence of more than that one FLOP without invoking time and duration.

      If they're executing on less than (1 billion cpus) /(number of simultaneous floating point instructions each cpu can execute), you can't "have" a billion FLOPS. Any snapshot will show a certain number of FLOPS being executed - the others are gone or not yet there. So a gigaflop, unless it's being simultaneously executed on such a large machine, definitely has a time dimension that extends beyond the duration of the execution of one FLOP.

      Lets further simplify - again, since the execution of the instruction is not instantaneous, each FLOP definitely has a time component, just like every other event in this universe. It has a start and an end. Just like I have 5 cookies. Duration (time) is implicit in the act of my "having" 5 cookies. Remove the time component, and the cookies vanish - they lack duration. Same with the gigaFLOP. Contrary to your original nit-pick, my nit-pick shows that a gigaflop must have a time component :-)

      It's also by manipulating the time component (allowing for the passage back in time of information on state) that we can solve the halting problem, at least in the confines of a thought experiment - and this is a lot more interesting than any nit-picking if only because of the side-effects. (yes, the flow of state info back in time isn't allowed, but this is a thought experiment - it's allowed by the same rules that give us an infinitely long tape. You'd be surprised how people who allow the tape without blinking an eye don't want to allow information to go back in time - guess they never heard that information wants to have many degrees of freedom :-)

    17. Re:Simpler solution. by Jah-Wren+Ryel · · Score: 1

      Same with the gigaFLOP. Contrary to your original nit-pick, my nit-pick shows that a gigaflop must have a time component :-)

      No more so than saying that "cars" or "oranges" or "dogs" or "meters" must have time component. A gigaflop is a unit unto itself.

      --
      When information is power, privacy is freedom.
    18. Re:Simpler solution. by tomhudson · · Score: 1

      And as I point out, all such objects DO have a time component.

      Also, unlike "metres", since a gigaFLOP requires that something actually be executed (an action performed, done, or carried out from start to finish), you can't have a FLOP w/o time. A FLOP isn't a discrete object or thing, it's an action. You can't "have" a gigaFLOP, you can only "do" a gigFLOP. Sort of like you can't "have" a push-up, you can only "do" it. You can't wrap a gigaFLOP up in nice paper and send it with a gift card ...

      But it's all nit-picking anyway ... I thought the joke wasn't *that* bad (I've done better, I've done worse, try the fish :-)

    19. Re:Simpler solution. by codemaster2b · · Score: 1

      technically, according to your logic, it would be a billion floating-point operations PER

      --
      And over there we have the labyrinth guards. One always lies, one always tells the truth, and one stabs people who ask t
    20. Re:Simpler solution. by blueg3 · · Score: 1

      No, "gigaflop" is a different unit, in which the "p" comes from the second letter in "operations".

    21. Re:Simpler solution. by codemaster2b · · Score: 1

      Sorry, it would have helped if you had clarified, perhaps by using a source like this, or spelled it out with capital letters.

      --
      And over there we have the labyrinth guards. One always lies, one always tells the truth, and one stabs people who ask t
  32. BEOWULF! by dandart · · Score: 2, Funny

    But will it run Crysis? But in all seriousness, BEOWULF OF DSs!

  33. Why? by Billy+the+Mountain · · Score: 1

    Why would DARPA want this? Maybe they want a AI that can navigate aircraft or gound vehicles? BTW, I think it's ironic that autonomous operation seems easier to develop for aircraft than for ground vehicles when you consider that pilots get way more respect than the average municipal bus driver.

    BTM

    --
    That was the turning point of my life--I went from negative zero to positive zero.
    1. Re:Why? by captjc · · Score: 1

      I think it's ironic that autonomous operation seems easier to develop for aircraft than for ground vehicles when you consider that pilots get way more respect than the average municipal bus driver.

      Maybe it is because fighter pilots don't have animals and idiots who don't look where they are going running out in front of planes while the plane right behind them is flying like an asshole because the pilot would rather send text messages than watch where they are going.

      --
      Slow Down Cowboy! It's been 1 hour, 47 minutes since you last successfully posted a comment
    2. Re:Why? by EbeneezerSquid · · Score: 1

      It would be just as easy to make autonomous operation for vehicles too, if everyone filed a Driving Plan at least 12 hours prior to leaving.
      That goes for people walking, too.
      I think the incidence of DogStrike would be a little higher than the incidence of BirdStrike, but not too much.

  34. Hmmmm by the+eric+conspiracy · · Score: 4, Funny

    So now we know what the hardware requirements for Windows 10 are going to be.

    1. Re:Hmmmm by Sparky+McGruff · · Score: 1

      Maybe that will work, but you'll have to turn Aero 5 off. And, of course, you'll be limited to 3 apps.

    2. Re:Hmmmm by selven · · Score: 1

      And no, you cannot make one of the apps a virtual machine that lets you wish for three more wishes.

  35. Whoa. by Anonymous Coward · · Score: 0

    Imagine a Beowolf cluster of those things!

  36. Yeah I know... by Anonymous Coward · · Score: 0

    Comprise... proofreading fail.

  37. Government job by lagomorpha2 · · Score: 2, Funny

    Given this is the government would I still get funding if I developed a computer that was capable of 50 Gwatts per FLOP?

    1. Re:Government job by rts008 · · Score: 1

      [sarcastic humour]
      Gov't. funding? Bah!
      That would likely win you a multi-billion dollar contract to:
      1. install it to control the electrical power grid!
      2. Fail to meet the deadline, and ask for bailout bucks to fix the unforeseen bugs/security holes*
      3. ????
      4. Profit!
      (Gov't. contracts require that extra step to go from '????' to 'Profit!'...red tape, you know)

      *leave the backdoors intact, and sell them to the Chinese hackers and Bin Laden, patent this method, and then sue Diebold for bonus profits!

      Think big!...Don't sell yourself short!

      With all of that on your resume, you could find yourself in a position to be posted as Infrastrucure Czar, and have the ability to be bought my thousands of Corporations!
      [/sarcastic humour]

      BTW, I not only spewed vodka out of my nose when I read your comment, I also shamelessly giggled like a schoolgirl while doing so!
      Yes, funny can hurt, but it's worth it!

      --
      Down With Slashdot BETA!!! I've been around the corner and seen the oliphant; you can only abuse me from your perspecti
  38. Positive Development by rbmyers · · Score: 1

    The government has spent way too much money wiring up huge quantities of commercial off-the-shelf processors and issuing press releases about its great accomplishments, which mostly reflect having money and being willing to spend it with a minimum of imagination, insight, and risk. I'd love to know where they got these goals. I sense a briefing from IBM about three-dimensional chips and microfluidic cooling, and all the wonderful things they could do if only the heavens started pouring forth money. I hope someone else is in the game, but it's hard for me to imagine who.

  39. try this by GarretSidzaka · · Score: 0

    see how many intel atoms you can cram into a back pack!!!!

  40. only 60KW? by Mishotaki · · Score: 1
    I'm pretty sure it can be done easily, make robotic legs that can run... use less than 60KW.... and put a Roadruner supercomputer in a 19"X19" box that goes as high as it can... and there you go!

    get rich or die trying!

  41. Re: Good luck with that by symbolset · · Score: 1

    I can think of at least three ways off the top of my head. They're not likely to knock on my door. It's really not hard if you think about it.

    You start with the smallest core that achieves your goal, and the largest, thinnest wafer you can get. And then you stack them with through-silicon vias. Include holes for liquid nitrogen coolant or even high-rate gaseous cold helium. Get a cool interconnect like 12xQDR infiniband or something custom, but it can be done. Connecting the silicon to external interfaces requires gold wire, an electron microscope, and infinite patience but it can be done.

    How hard was that? Do we need to draw them a map?

    --
    Help stamp out iliturcy.
  42. The company who were closest have gone broke by Jacques+Chester · · Score: 3, Interesting

    A firm called SiCortex was selling just this sort of compact, energy-efficient supercomputer. They shut down a few weeks ago because an investor pulled out.

    It's a damn shame, they had really cool stuff. If I was Johnathon Schwartz I wouldn't have pissed away $1 billion on MySQL (it was worth maybe $10 and a stick of gum), I would have been out the front of SiCortex banging on the door with a chequebook.

    Oh well.

    --

    Classical Liberalism: All your base are belong to you.

    1. Re:The company who were closest have gone broke by Anonymous Coward · · Score: 0

      this presentation has been brought to you by the people at SiCortex.

      SiCortex: tomorrows venture - today!

    2. Re:The company who were closest have gone broke by SpinyNorman · · Score: 1

      The SiCortex website mentions their top of the line machine using about 900mW per 1.4GFlop processor (1.55Glop/W), but the overall system uses 20KW for a 5832 processor system, giving a system figure of (5832 * 1.4) / 20,000 = 0.40GFlop/W.

      http://www.pathscale.com/products/high_capability_system_sc5832

      Now, consider that DARPA wants 50GFlop/W - a factor of 100x improvement over the SiCortex number.

      So, I'd hardly classify SiCortex's products, cool as they may be, as "just this sort" of energy-efficient supercomputer.

      In fact the IBM Roadrunner supercomputer TFA refers to does better thsn SiCortex's system, and achieves 0.44GFlop/W.

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

      Just keepin' it real!

  43. Not one, the other by WilyCoder · · Score: 1

    Forget about computing FLOPS and instead first consider this a thermodynamic problem.

    Solve the thermodynamics, find a little space for CPU cores, solved.

    1. Re:Not one, the other by rts008 · · Score: 1

      The 'Laws' of Thermodynamics are just holding up progress....I say we repeal them!
      *apply sarcasm filter for above*

      I think you are spot on here, but I would like to introduce two points here:

      1. Typically DARPA research is often aimed at stretching, if not breaking boundaries.

      It's not always aimed at achieving that specific goal, as it is more about branching into many paths.(both with technology as well as ways of thinking/approaching problems)
      Remember how scientific research works...disproving something, as well as demonstrating that 'it can be done, but is not the best choice here' are both valuable contributions to the body of scientific knowledge.
      Knowing what does not work is just as valuable as knowledge of what does work relative of 'just getting it done'. (I see the inestimable value of challenging and testing both to further improve and refine our knowledge to improve our lives...no arguments from me here)
      The achieving of the stated goal is worthwhile, noteworthy, extremely 'lofty', and important in their own right, but more 'icing on the cake', as all of the spawned tangents can frequently overshadow the original project with numerous useful stuff.

      2. DARPA research projects trickle down into the DOD, then to the military.
      Many of these, or aforementioned tangents end up as useful things available to civilians. (GPS is a good example-many more are used by us almost daily)

      Maybe this will advance heat removal tech, or a practical way to utilize the heat removed to convert it to useful energy, or both for the win!
      Maybe something completely different to solve some/all of these problems to reaching these goals will surface.

      I work at a state university, and in the 'break room' we use is a poster about 'Alchemy' as a humorous comparison to what we refer to as modern science today.
      I often wish I could live long enough to see the time that future scientists regard our current science, as we currently regard Alchemy and other 'Sciences' during medieval days/Dark Ages.

      'twould be a wondrous world to experience, I expect, having come along for the ride. *sigh*
      I should not complain, as I have witnessed many tech marvels in my lifetime so far!(I am 51, and a NASA brat...hopefully a few more decades to go yet!)

      --
      Down With Slashdot BETA!!! I've been around the corner and seen the oliphant; you can only abuse me from your perspecti
  44. A device mostly for C3? by Kyusaku+Natsume · · Score: 1

    Coupled with the power source this thing on the field would have a huge heat signature that would work like a beacon for enemy missiles I guess, but I will gladly be corrected by any reader that served on the armed forces and have a better idea.

    --
    Mexico: 100% conservative's America now!
    1. Re:A device mostly for C3? by John+Hasler · · Score: 1

      It's heat signature will be no worse that that of a large internal combustion engine. The enemy already knows the location of division headquarters.

      --
      Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
  45. It's doable, but too expensive by bertok · · Score: 4, Informative

    This is actually probably quite doable, but would be filthy expensive.

    Most people don't realize, but digital electronics is way, WAY ahead of what you get in your home PC, if you're willing to pony up the cash.

    For example, non-Silicon based semiconductors often outperform the good old standard stuff significantly. Silicon is by no means the fastest, it's just the cheapest. Gallium Arsenide and Indium-based materials can both clock many gigahertz higher than Silicion for the same process size and power dissipation. They're toxic, fragile, and the largest wafer sizes are tiny, so not exactly mainstream, but available now.

    The real performance king though is the Rapid Single Flux Quantum process, which can go over 100 GHz easily. It's used in things like radio telescope amplifiers and high-performance DSPs for military radar. Sure, it requires liquid helium cooling, but it also only requires milliwatts per gigaflop, so it's just about the only technology that'll let you squeeze a petaflop into a box and not have it melt into slag. That still means you'd need something like a kilowatt of cryogenic cooling, which is nontrivial, but still, I'd say it's doable with a bit of engineering wizardry.

  46. Re:Going against Resistance is futile. Sidestep it by Hurricane78 · · Score: 1

    Which only shifts the heat to somewhere else, by using power, creating even more heat.
    I don't think that is what they want. ^^

    --
    Any sufficiently advanced intelligence is indistinguishable from stupidity.
  47. 50 GFLOPS per Watt = Hard by 192939495969798999 · · Score: 1

    Just for some reference, an Intel Core 2 Quad can throw down about 50 gflops, but it runs on 95-130 Watts of power. To even come close to this 50GFLOPW number, you'd be designing your own HPC chips for sure.

    http://www.pctechguide.com/26quadCore.htm

    --
    stuff |
  48. Sounds like... by Anonymous Coward · · Score: 0

    Nigel gave me a drawing that said 18 inches. Now, whether or not he knows the difference between feet and inches is not my problem. I do what I'm told.

  49. This has been done before... by jerryluc · · Score: 1

    The first computers was as big as supercomputers are today. Today these computers would fit inside a peanut.

    The day today's supercomputers will fit inside a 19" box, will be the day an everyday user will need that kind of power.

  50. Tesla does this already by Anonymous Coward · · Score: 0

    I thought super-parallelized processing in a box was already solved...

    Nvidia Tesla fits the bill I believe.

    http://www.nvidia.com/object/product_tesla_s1070_us.html

    This is 1U at about 4 TFLOPS

  51. It's 50 gigaflops per Watt, which is timeless by Jamie+Lokier · · Score: 1

    Watt = Joules per second.

    They want 50 gigaflops per Watt.

    50 x 10^9 floating point operations per second per (Joules per second).

    = 50 x 10^9 floating point operations per Joule.

    That part of the specification does not have a time dimension.

    Enjoy :-)

    (Ok, then they specify the Wattage which brings time back in...)

  52. me too by GISGEOLOGYGEEK · · Score: 1

    I want one tooo

    .
    .
    cus.gus@hotmail.com

    --
    George Bush + Linux = "I will not let information get in the way of the fight against Windows"
  53. Umm by Anonymous Coward · · Score: 0

    16 positronic brains in a cabinet woohoo
    highspeed computing here we come.

    A couple congress members a decision every few .... decision what decision???

  54. One of my favorite games! by rts008 · · Score: 1

    Don't give that robot a weapon, as it usually will shoot you or others in your party, or bystanders...thus starting a fight with the whole town(especially likely in New Reno), etc... ...or give him a gauss rifle and set his combat mode to berserk. Good fun!

    --
    Down With Slashdot BETA!!! I've been around the corner and seen the oliphant; you can only abuse me from your perspecti
  55. 24" maybe ... by HW_Hack · · Score: 1

    I could probably pull that off - 19" is fucking insane.

    They call it hardware because its fucking hard to do. You can't just recompile a motherboard or a thermal solution if you screw it up.

    --
    Its not the years, its the mileage .....
  56. Bwhahahahahaha! Muhahahaha! by rts008 · · Score: 1

    *drawing diagrams and writing equations on blackboard, puts down chalk, and posts on /. while looking over shoulder at camera, grinning maniacally*

    'And I'm working on a BeoCoyote Supercluster to finally catch that Roadrunner!
    Acme's new Multicore Coyote(tm) chips RULE!!!!!
    It will work this time!

    signed,
    Wile E. Coyote, Super-Genius

    P.S. 'Meep! Meep!' my ass!'

    --
    Down With Slashdot BETA!!! I've been around the corner and seen the oliphant; you can only abuse me from your perspecti
  57. 18 years by bill_mcgonigle · · Score: 1

    I'll bid the entire contract for $15,000. The only catch is it's going to take me 18 years to deliver and I require payment up front. Checks payable to Dr. Gordon Moore.

    --
    My God, it's Full of Source!
    OUTSIDE_IP=$(dig +short my.ip @outsideip.net)