Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Handhelds Supercomputing Hardware

Researcher Discusses iPod Supercomputer 108

Posted by ScuttleMonkey from the melts-in-your-hand-not-in-your-rack dept.
schliz writes to mention that in a recent interview with ITNews researcher John Shalf explained the purpose and some of the technical details of the newly-announced "iPod supercomputer." "Microprocessors from portable electronics like iPods could yield low-cost, low-power supercomputers for specialized scientific applications, according to computer scientist John Shalf. Along with a research team from the US Department of Energy's Lawrence Berkeley National Laboratory, Shalf is designing a supercomputer based on low-power embedded microprocessors, which has the sole purpose of improving global climate change predictions."
This discussion has been archived. No new comments can be posted.

Researcher Discusses iPod Supercomputer More

Researcher Discusses iPod Supercomputer

Comments Filter:

  • sure, this is how it starts (Score:5, Funny)

    by Anonymous Coward on Friday May 09, 2008 @01:04PM (#23352946)
    But sooner or later, they come after you claiming you haven't legally purchased your global climate change predictions, or that you've been sharing them with your friends online.

    • Re: (Score:3, Funny)

      by Peet42 ( 904274 )
      "We're sorry, we can only authorise you to share this data with four other nodes. Have a nice day."

      • "We're sorry, we can only authorise you to share this data with four other nodes. Have a nice day."
        I think you're confusing this with the Zune super computer... and it doesn't "share" it "squirts [portableplanet.co.uk]"
        • REDMOND, Seattle, Wednesday (UNN Technoporn) -- Microsoft CEO Steve Ballmer today announced a new era at the Seattle software company, announcing their entry six^Wnine^Wtwelve months hence into the cell phone market with the exciting new Zune Z-Phone, to finally get the company properly into the rapidly changing digital media landscape.

          Ballmer, speaking to a group of trained-monkey analysts and cynical bloggers at the company headquarters today, unveiled mockups^Wprototypes of the Z-Phone, which combines the Zune music player (with wifi for "squirting" songs), a CDMA cell phone, a PDA, an eight gigabyte hard disk, a camera, a laser pointer and a bottle opener into one semi-portable device. It will also allow you to "squirt" music to and from your Windows Vista Service Pack 1^W2 Media Center computer.

          The product underscores the shift the company has attempted to make in recent years from an office supply company to a consumer electronics darling as it aims not to become utterly obsolete in the digital future. "And even Linux fanboys admit our hardware is pretty nice," Ballmer said before the somewhat sullen and cynical crowd. "It's definitely the best music player we've ever made."

          Ballmer called the Z-Phone a revolutionary device that will leapfrog current technology. He said the company expects to sell about 100 million of them next year. "Maybe two hundred million. This is so the coolest music player ever." Unlike the MP3 player market, which the iPod has dominated even with the entrance of Microsoft's Zune two months ago, the cell phone market is much more fragmented. "There is not one device that everyone buys," said completely independent analyst Rob Enderle, "but this fabulous device should trounce all comers. I've ordered three already in anticipation."

          Weighing in at only 15 ounces (425 grams), with a 5-inch 640-by-480 pixel screen, the $498 (with three-year $80/month contract) Z-Phone, a rebadged version of the LG Smart Display from 2003 with new firmware, looks like a Classic Brown Zune (to come in mission, chocolate, corduroy and meconium) with a phone touchpad in place of its imitation scroll wheel. It runs Windows Mobile, Pocket Internet Explorer, Pocket Microsoft Office, Pocket Solitaire and Pocket Pool. MSN will supply e-mail, mapping, search and other Internet services to the Z-Phone. It also features an amazing 1.3 megapixel (300,000 pixels interpolated) black and white camera. Battery life is estimated at up to four hours in Microsoft tests.

          To better work with its content partners and ensure that you, the user, can rest safe in the knowledge that the artists and their representatives have been paid properly for all their hard work, Microsoft has limited "squirtable" songs to encrypted WMA files purchased from the Zune Music Store, which can be listened to three times or within three days before automatically being deleted from both the Z-Phone and the Media Center computer. Songs may also be "squirted" between two Z-Phones (though not the original Zune) if both are registered with Microsoft as being linked to that installation of Media Center. Users are advised to purchase Microsoft Zune Secure Headphones ($129), which encrypt the signal between the Z-Phone and your ears, as playback quality is degraded on conventional "analog hole" earphones or when playing back unencrypted MP3 files. Phone calls may be made to or received from any number on the network carrier you bought the Z-Phone from, with only a 99-cent charge for humming a song to someone you call or are called by on the phone or ten cents per use of the camera, laser pointer or bottle opener. Microsoft will also pay $20 from each Z-Phone sold to Universal Music. In addition to the ability to "squirt" songs, the user may "squirt" his calls, which are stored on Microsoft Zune Live servers and cost $40 per month to access.

          In other news, Ballmer said that Microsoft had reached over 600 music downloads since introducing its Zune Music Store, selling over 70 songs a month. To keep those numbers rising, Ba

  • Oblig. misleading title (Score:5, Insightful)

    by Ethanol-fueled ( 1125189 ) * on Friday May 09, 2008 @01:05PM (#23352952) Homepage Journal
    From the article:

    "Using the embedded microprocessor technology used in mobile phones, iPods and other consumer electronic devices, the boffins propose a cost-effective machine for running complex computational models."

    In other words, all be damned if they decide to implement this monstrosity using actual iPods when they could use their talent to design and build greater efficiency through Spice/HDL, manufactured boards, and a pick-and-place.

    Gee, A mesh of dedicated machines, hardcoded for more efficiency than a cluster of bloated pc's designed for MS office is actually more efficient? Geddouttahere!

    [/sarcastic rant]

  • iPhone (Score:3, Insightful)

    by Philomathie ( 937829 ) on Friday May 09, 2008 @01:07PM (#23352966)
    Excuse me if I'm wrong, but would this not be more specifically a mobile microprocessor supercomputer than an iPhone supercomputer? I mean, its not as if only the iPhone uses mobile processors.

    • Re: (Score:3, Funny)

      by Philomathie ( 937829 )
      Hahah, oh dear I misread the summary, this is about iPod microprocessors not iPhone microprocessors... excuse me I am quite drunk :P I will see you all in another life, when we are all cats!

      • Re: (Score:2, Funny)

        by Anonymous Coward
        I'm already a cat, you insensitive clod!
  • but eventually they come after you saying you haven't legally purchased your global climate predictions, or that you're sharing them with your friends online
  • ...will be devoted to DRM?

    Inquiring minds want to know.

  • Image a ... (Score:5, Funny)

    by Anonymous Coward on Friday May 09, 2008 @01:13PM (#23353032)
    Beowulf cluster of ...I for one welcome our iPod overl....in Soviet Russia, iPods ....does it run....

    Please stop hitting me!

  • Send this article to that Jonathan Zittrain idiot who thinks that putting the web in the hands of everyone everywhere is a failure of technology and will stifle innovation.

  • It's about the bandwidth, not the MIPS (Score:5, Interesting)

    by morcheeba ( 260908 ) on Friday May 09, 2008 @01:23PM (#23353156) Journal
    There's a famous quote about supercomputers that says that supercomputers are really good memory systems, with a bit of CPU tacked on. The hard part isn't adding more MIPS -- we've done that with the massively parallel connection machine -- or even increasing speed. It's about shuttling the data around the computer efficiently so that all ALU's are constantly fed. During the cold war, Control Data had a supercomputer that came in two variants -- one for domestic use, one for export. The difference between them? Same ALU speed, but the domestic one had a scatter/gather memory access capability that sped up big matrix operations.

  • self-fulfilling prophecy (Score:5, Funny)

    by fpgaprogrammer ( 1086859 ) on Friday May 09, 2008 @01:23PM (#23353158) Homepage
    The observer effect: the more energy we consume studying the effect of energy consumption on climate change, the more we'll have to incorporate this factor into our models.

    Positive feedback: if the results of these studies are striking enough to merit funding for more research, we'll no doubt consume even more energy to determine the effects of energy consumption on climate change.

    Self-fulfilling prophecy: if this positive feedback between funding for climate change research and supercomputing energy consumption is not counteracted by efforts to reduce supercomputing power consumption for climate change research then we're damning ourselves by studying it.

    • The observer effect: the more energy we consume studying the effect of energy consumption on climate change, the more we'll have to incorporate this factor into our models.
      Conversely, if we ignore the problem, eventually it will go away.

    • Re: (Score:3, Funny)

      by Chris Burke ( 6130 )
      Well, they said they're trying to increase the accuracy of global warming predictions. I'd think that if that's what they wanted, using low power processors is exactly the wrong idea. I say use the least efficient, highest-power-sucking processors they can find, and guarantee that their results are accurate. "Global warming is a sure thing! And it's centered around our data center..."
    • Excuse me, Mr. fpgaprogrammer, but your bringing critical analysis to this situation has consumed far too many calories......

      Two brilliant & original suggestions:

      (1) Why not build a super computer the size of the Earth, and name it Deep Thought?

      (2) Oh...I forgot it....

  • Is that a supercomputer in your pocket or are you just happy to see me?

  • facepalm (Score:5, Insightful)

    by blhack ( 921171 ) on Friday May 09, 2008 @01:35PM (#23353260)
    What the hell does this have to do with ipods? They're building a supercomputer out of low-power MIPS procs..

    embedded processors were, believe it or not, NOT invented by apple. I don't know if its true or not (i doubt it) but I've also heard that there were portable electronics BEFORE the ipod.

    This is really cool, but slashot, come-on...most of us here are geeks, we don't need to have the word "ipod" tacked onto the end to indicate that we're talking about something small.

    • Re: (Score:3, Insightful)

      by Idbar ( 1034346 )
      Better yet, why do they have to talk about iPods and not cellphones. There are plenty more cellphones in the world than iPods, all of them, connected to networks. Most of them, running already Java applications. The platform is already there.

  • BlueGene/QCDOC vs this? (Score:4, Interesting)

    by Beale ( 676138 ) on Friday May 09, 2008 @01:37PM (#23353302)
    BlueGenes, and their predecessor, QCDOC [wikipedia.org] supercomputers, already use slightly modified low-power embedded system chips. How is this any different?
  • My "global climate change prediction", sans iPod:
    HOT

  • its about time we had a climate model (Score:5, Funny)

    by museumpeace ( 735109 ) on Friday May 09, 2008 @01:41PM (#23353336) Journal
    that could predict global warming without first causing it.
  • I am just making up a good excuse to buy 100 ipods using some grant money (I'll use most for "research" and have some spares---which in the meantime I can use: one for me, one for wifey, one for daughter, ah! and lets not forget our nice nephew... he said he'd mow the lawn a few times for free too). Pathetic.
  • Well then hopefully they'll factor in the amount of CO2 given off from generating the electricity to run these processors. Just because they're more power efficient doesn't mean they run on magic. This is like taking a bus ride across the country to protest all the CO2 given off by vehicles.
    • Not to forget the CO2 given off by their bodies and the bodies of the animals whose meat they might eat.
  • Man, operating that thing with the click-wheel is going to be a bitch.
  • The last time I was called a "boffin" was by a Reme sergeant in 1982. I remember it well. The sergeant's precise words were "You, Sir, are quite sensible for a boffin" and I asked if I put it on my cv would he sign it?

    Onto the serious bit. This proposal is basically a reinvention of the Transputer, lots of little blobs with cpu, memory, and fast communication links. Is this because:

    • [ ]It's now possible to write software to run on massively parallel machines effectively
    • [ ]The idea just keeps getting reinven
  • This reminded me of blade servers [wikipedia.org]. I wondered why they didn't just order a bunch of blades with RAM and CPU only.

    I read the article, and they are planning to have special CPU chips fabbed: CPUs tailored specifically to the needs of climate modeling. I guess this will provide the lowest possible operational cost--the least electrical consumption and heat dissipation possible to solve their problem.

    Quote from John Shalf:

    We have something that automatically tunes the software after we make a hardware change,

    • So... they're trying to solve the problem while being as small a contributor to the problem as possible. Makes sense to me.
    • The BlueGene series of super-computers follow this model - although less specific to climate modeling.

      iirc they have paired PowerPC chips with a floating point accelerator as processing nodes, with i/o nodes connected via some sort of bus - something like 16 or 32 processing units/notes per board, with a handful of i/o nodes providing connectivity to storage & i/o layers.

      Much higher density than blade servers, but the ram/cpu only blade server idea is still pretty popular; at a previous company I worked
    • Oh, come on. It's just a very minor difference between "ASIC" and "iPod". It surely was just a typo.
    • Sure! Right. Why don't they just order a bunch of loaves of bread and hook them up with Vegemite. :) .o.
  • my iPod is wasting valuable cpu cycles that could be saving the world!
  • Well at least the shuffle play might be truly random.
  • Outlaw all electric heaters that heat using a resistance rather than microprocessors. Make heaters work with a WAN connection only.

  • Crap (Score:1, Insightful)

    by Anonymous Coward
    Yeah, 32 bit ARM processors that probably use soft float (soft float is a guess!) are the way forward in super computing. You better believe it, Ripley.
    • Many embedded processors have FP and DSP hardware in addition to an ARM core, because their applications demand that they perform non-trivial DSP. Smartass.
  • Of course by the time you get enough low-power processors running to solve the the problem you'll be a major contributor to climate change yourself.
  • A RAID array of zip drives?
  • Steve Circia did that in the 80's with a bunch of 8051's.

    Funny how all we ever do now is run in circles, where is the REAL innovation?
    • Commodore has been doing this since the 1970's. Each standard 1541 disk drive has a 1 MHz CPU and 2K of RAM, and communicates via a daisy-chained serial bus. Aside from the usual purpose of the drive's CPU/RAM, there have been one or two programs in the past that would use a drive as a second, general-purpose computing node, so any C64 with a few drives could be called a cluster if programmed for that purpose. There's a limit to how many drives the serial bus can handle (due to signal degradation), but
      • er, I should qualify that by saying that the "1970's" part refers to PET machines and 8050 drives, which had a similar amount of raw CPU power.

  • Riiight (Score:3, Informative)

    by 4D6963 ( 933028 ) on Friday May 09, 2008 @03:38PM (#23354998)
    Not just the title is misleading but the idea associated with it. Last time I checked (ok, that was a while ago), iPods came with ARM7 cores clocked at 80 MHz. Thing is, these CPUs don't have a floating point unit, so unless they write they weather simulators in fixed point arithmetic (lol, right) or go ahead with software floating point emulation (which would slow things down several times) they're not going to use these, they'd rather use more sophisticated stuff like ARM11s or Cortex A8s.
    • The ARM7s are RISC CPUs, they do only a few instructions very fast and efficiently. That means that if you want an FPU you include one in software and if you don't then you also don't have to pay for one or have to power one saving you a pack of cash.

      • Re: (Score:2)

        by 4D6963 ( 933028 )
        Yeah and the software FPU is awfully slow, which was my point.
        • Really? And why is that? I can't think of any reason for a software FPU to be tremendously slower, sure there's a bit more overhead but on the other hand a RISC processor will typically execute instructions far faster then a CISC. (faster in this case referring to the number of cycles needed to complete an instruction rather then the amount of actual time needed)

          • Re: (Score:2)

            by 4D6963 ( 933028 )

            Well, we the people who develop for the GP2X (a handheld console with an ARM920T core) avoid floats because -msoft-float is so slow. Can't tell you why, only can tell you it is.

  • Okay, so this should probably be on halfbakery.com, but - after my initial non-reading of the article, and my assumption that this had nothing to do with iPods, and my scoffing of the notion of a supercomputer of iPods... hmmmmmmmmm...

    So just for the Friday afternoon fantasy's sake, I am envisioning a series of flat grids of iPods, communicating through their dock adapter. More like discrete workers - here's a work unit, there's your output, etc. Built in UPS (battery), ability to pause a simulation and mov
  • Did anyone else read that and see John Shaft [imdb.com] instead of John Shalf?

    Who's the computer scientist who's a sex machine to all the chicks? Shalf!
    He's a complicated man and no one understands him but his supercomputer based on low-power embedded microprocessors.

  • Not to burst any ones funding bubble, but this has been done. Take a look at SiCortex [sicortex.com]. They did it, they are shipping product, and it works quite well. And it runs Linux.

  • This would be so much more useful and immediately advantageous then these silly SETI and FOLDING@HOME projects that people are running.

    I ran both of these for a while, but here's the thing about them, particularly SETI. It makes a lot more sense to do these calculations in 20 years, rather then now. The computing effort required compared to the available world computing power is HUGE, but it won't be nearly as much in 20 years given Moore's law. Why suck up all the electricity now, when what we do for t
  • The results will be easily available at 99 cents on iTunes. Of course, if it involves video, it will go for 1.99 cents.
  • I'm not some wackjob who believes global warming doesn't exist, but this strikes me as a great money making endeavor. Hear me out.

    1 - Buy lots of expensive computer equipment
    2 - Write some software that models climate change, but adds a bit of extra warming each time
    3 - Run simulations
    4 - Release results.
    5 - Clearly, more information is needed on climate change!
    6 - Receive new grants. (aka Profit!)
    7 - Increase fudge factor, repeat.

    For bonus points, let the devices you use be so inefficient (iPhones? c'mon)
  • Comment removed based on user account deletion

Slashdot Top Deals

The truth of a proposition has nothing to do with its credibility. And vice versa.

Close