Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Handhelds Hardware

A Cluster Of Pocket PCs 119

Posted by timothy from the can-this-really-be-the-first dept.
Don Stratton writes "This is the coolest thing I have seen anyone do with a Pocket PC... ever! Well-known Pocket PC developers SPB Software House, located in Russia, have come up with a very interesting spin on computing clusters. The short version is they connected 12 Pocket PCs together in the first known 'supercomputer cluster' of its type and had it calculate the old '3n + 1' problem. It was just done for fun, and not intended to seriously compete with desktop computers, but it does point out some interesting possiblilities for the future of handhelds with wireless connectivity working in ad-hoc computing clusters."
This discussion has been archived. No new comments can be posted.

A Cluster Of Pocket PCs More

A Cluster Of Pocket PCs

Comments Filter:

  • Happy, aren't you? (Score:5, Funny)

    by gooru ( 592512 ) on Wednesday October 08, 2003 @01:56PM (#7165554)
    Is that a beowulf cluster in your pants, or are you happy to see me?

  • omg (Score:5, Funny)

    by blake8087 ( 688462 ) on Wednesday October 08, 2003 @01:56PM (#7165560)
    in soviet russia... i mean... imagine a beowulf... um... no comment
    • You could hook up these PocketPCs to your speaker bracelet...
    • I think it is perhaps too much of coincidence to have a story that is both about Russia and clusters. Somebody purposely sought out such a story I bet. What is next, making a profit on Russian clustering technology by helping companies filter out goatse gifs?

  • How do you? (Score:2, Interesting)

    by koali ( 175176 )
    Parallelize 3n+1? It looks like it cannot be parallelized. On the other hand, they could try several different starting n's at once, but that is not very interesting...
    • You just have each node use a different starting point, continuing until it hits the 4,2,1 loop, and ensure that when a node picks a new starting point it doesn't pick one that has already been picked by another node.
      • A further optimization is that when you get to a number that's known to go to 1, you don't need to go the rest of the way. So you can prove the conjecture by crossing out each number which goes to 1, and continuing until there are no numbers left.

    • Re:How do you? (Score:5, Interesting)

      by Ass, Ltd. Ho! ( 714400 ) on Wednesday October 08, 2003 @02:14PM (#7165677)
      The answer to your question is really kind of interesting. I came across it while searching through some japanese documents at the university of Nagasaki.

      instead of taking the "multiple-serializing" approach you suggest, you begin by using the output of your first transform filter as a sort of "seed" to feed groups of parallel "second-tier" transform filters. Once you've done that, you can see how it is easy to fan those outputs out and iterate into third or fourth-tier (obviously, that's a LOT of pocket PC's!) transform filters. As the filters each work their way through seed data, they each begin to converge toward (notice i didn't say TO) a solution. You then use what the japanese called a "recompositing Reimann filter" to sort of sift your way from the multi-tiered parallel seed results back into a final solution.

      It's very elegant, IMHO.

  • Wow! (Score:4, Funny)

    by inteller ( 599544 ) on Wednesday October 08, 2003 @01:59PM (#7165578)
    now I can open up my contacts with blazing speed!
    ...if I just didn't need a backpack to carry it all around with me.
    Chalk up another nerdgasm that got posted to /.

  • Break out the mini racks (Score:5, Funny)

    by antimith ( 683310 ) on Wednesday October 08, 2003 @01:59PM (#7165580)
    IBM is introducing their lates server line, the IBM ButterKnife series, featuring a fully hot swappable cluster of up to 200 MS Pocket PC's. On good days you'll get the output of a screaming 386!
    • IBM is introducing their lates server line, the IBM ButterKnife series

      Wouldn't that be something like the iServer uSeries?

      cluster of up to 200 MS Pocket PC's

      I thought IBM and MS didn't like each other any more? IBM is pushing Linux on all their other hardware, why would the iServer uSeries be any different?

  • Too obvious? (Score:3, Funny)

    by Burpmaster ( 598437 ) on Wednesday October 08, 2003 @01:59PM (#7165583)
    In Soviet Russia, Pocket PCs imagine a beowulf cluster of YOU!

  • Sounds like a dot-com VC funding pitch... (Score:3, Interesting)

    by Anonymous Coward on Wednesday October 08, 2003 @02:00PM (#7165587)
    it does point out some interesting possiblilities for the future of handhelds with wireless connectivity working in ad-hoc computing clusters


    That or report for a MBA marketing buzzword course.

  • The "3n + 1" Problem (Score:4, Informative)

    by LittleGuy ( 267282 ) on Wednesday October 08, 2003 @02:00PM (#7165592)
    The "3n + 1" problem involves starting with a particular integer n, and repeatedly performing the following operation:

    If (n is even) divide n by 2;
    Else multiply n by 3, and add 1.

    For example, starting with the number 6, we get the following series:

    6, 3, 10, 5, 16, 8, 4, 2, 1, 4, 2, 1...

    The 4,2,1 loop repeats over and over, so it's usually convenient to terminate the process once it is entered. All numbers tested so far eventually hit this loop, although it has not been proven that all numbers do.
    • The 4,2,1 loop repeats over and over, so it's usually convenient to terminate the process once it is entered. All numbers tested so far eventually hit this loop, although it has not been proven that all numbers do.

      -1, -2, -1, -2, ...

    • Re:The "3n + 1" Problem (Score:1, Funny)

      by Anonymous Coward
      I applaud your ability to cut and paste from the linked article.

      Well done! Well done!

      Especially as the site is not even feeling the /. effect.
    • "Mathematics is not yet ready for such problems."
      --Paul Erdo"s, talking about the 3n+1 Problem.

    • Here's a quick rendition of that in Python: #!/usr/bin/python values = [] N = 100 while (values[-3:] != [4,2,1]): print N, values.append(N) if (N % 2 == 0): N /= 2 else: N = N * 3 + 1
    • numbers tested so far eventually hit this loop

      Natural Numbers-- If a number is not an integer, it would be perpetually multiplied by three and thus would not follow the 4,2,1 pattern. Previous posts have established that if the number is an integer 0) tested do conform to this is property.

      If you would like to prove this property, you will obviously need the principle of mathematical induction. You would start with the base case of saying that the property holds for 1, the first number in the serie

  • this only says... (Score:2, Insightful)

    by mantera ( 685223 )
    expect things to come out of russia, china, india, and eastern europe in years to come.
  • I'm talking to a cubiclemate about it and said "Do ya wanna see the computational horsepower of the server we just turned off?"

    And held up my old iPaq and new iPaq. (The server was a dual PPro 200.)

  • they must read /. (Score:5, Funny)

    by PhiberOptix ( 182584 ) on Wednesday October 08, 2003 @02:06PM (#7165628)
    building a cluster of pocket pcs is fun and all, but I bet that the most fun part of the project was posting about it here on /. just to see the flood of beowulf cluster in soviet russia jokes.
    • I bet that the most fun part of the project was posting about it here on /. just to see the flood of beowulf cluster in soviet russia jokes

      Not everyone on Slashdot makes the same remarks over and over again, you insensitive clod!

  • Hmm.. (Score:3, Funny)

    by DarkBlackFox ( 643814 ) on Wednesday October 08, 2003 @02:06PM (#7165632)
    Imagine the number of Beowulf comments this story draws.

  • At last! (Score:1, Funny)

    by SoTuA ( 683507 )
    We can stop imagining a beowulf cluster of pocket PCs. It's here!
  • Since GRID computer has really come alone these past years, this latest venture is just too cool. As a participant and user of the Grid, I can definitely see potential. Imagine plugging your Pocket PC into a Grid to check your jobs or participate. Now imagine millions of other people walking down the street or at a coffee shop, or car, etc.. participating in a big Grid.

    There is going to be a time where everything will be plugged into the Grid. Man, I can't wait!

  • Similar to blue gene... (Score:4, Interesting)

    by gnuadam ( 612852 ) on Wednesday October 08, 2003 @02:17PM (#7165700) Journal

    This sort of reminds me of what they're doing with blue gene [ibm.com]. Instead of using hot, ultra fast processors, they're using what amounts to 2 embeded processors per node and depending on kick ass networking to carry the load.

    For parallel problems, sometimes faster processors is not better...

  • Wireless Clustering (Score:4, Insightful)

    by Rich0 ( 548339 ) on Wednesday October 08, 2003 @02:18PM (#7165706) Homepage
    Come to think about it, having something in between WI-FI and bluetooth might make for big wireless networking capabilities in a handheld. If the market penetration were high enough, you could route a packet from anywhere to anywhere as long as you had a high enough TTL.

    The challenge would be in organizing the routing tree. You'd have an advantage in that generally two nodes that are close to each other would tend to stay that way over moderate periods of time. Even on the road your handheld would stay close to the others in your same lane of traffic.

    If you could set up these devices to be able to share CPU automatically when idle it would mean that your handheld could utilize the CPU and RAM of the handheld in the briefcase belonging to the guy in front of you on the plane.

    Certainly none of this is ready for prime time, but it does raise some interesting possabilities.
  • I'm manipulating my pocket cluster.

    Wait, put down the phone. Stop dialing 911. Wait, stop!

  • It's been done before (Score:5, Informative)

    by smartin ( 942 ) on Wednesday October 08, 2003 @02:25PM (#7165720)
    Information here [handhelds.org], pics here [handhelds.org].
    • I'm pretty sure the innovation and difference in this new PocketPC cluster is the fact that it transmits data through the cluster through the wireless spectrum (irDA).

      The iPaq model that is shown in the parent uses pretty bulky pcmcia card expansion slots and a lot of wires to connect everything together. The wireless spectrum opens up the possibility to compute a much broader range of jobs through the limited brains of pocketPC's .

      I'm kind of interested to find out what OS this was created under. Did the
  • The most interesting application I see, is the bluetooth/wifi/whatever is actually being adopted wireless protocol. Then allow the machines to unload threads ala the way MOSIX allows nodes to be added and removed, just need to figure a way around the master scheduler. Maybe each machine runs it's own scheduler.... more over head, anyway, just a really neat idea.
  • I fugure 24 of them should more or less keep track of most of the worlds time zones.

  • Okay, I'll give it a shot too... (Score:1, Funny)

    by Anonymous Coward
    Ehem. ...can you imagine a beopuppy cluster of these things?

    Thank you.

  • Wireless or not... (Score:4, Insightful)

    by SharpFang ( 651121 ) on Wednesday October 08, 2003 @03:05PM (#7165838) Homepage Journal
    ...a "cluster" of more or less randomly distributed and connected computers isn't such a bad idea.

    On universities, 99% of computers run with nearly zero CPU load for most of the time. People read emails, surf the web, but for most of the time the computers idle. And then someone has some work reaoznajeszdy, waits in queue for a month, throws data on the university campus dedicated cluster, waits for results for a week and receives results that are invalid due to some mistake in input data, so whole procedure must be repeated all over.

    Now imagine, we install a "cluster server" on all networked computers. Assign certain resources to the project and let our PC participate in that cluster. It loads a custom computational module for given task, loads data from some anonymous dude on the other end of the world and computes his project. Heaviest "daily" stuff gets finished within few hours. It doesn't really disturb you - works as "idle task", just like SETI@home or such. But, say, you're a raytracer. You prepare a nice animation in LightWave and would leave it overnight to render. Just upload it to the net and have it rendered in 5 minutes on the worldwide cluster. Cool, eh?

    Of course the system could be abused. I think some "credit system" would be in order, so people who provide more, get better priority. Plus some way of authoring the "modules" so it couldn't be used to take over the computer. And of course this would be the first step to creating a self-conscious AI, good or evil :) But I think it would be worth a try.
    • -On universities, 99% of computers run with nearly zero CPU load for most of the time.

      Every once in a while I walk past a totally idle server and I say 'take THAT!' and I pick a massive data directory at random and zip the contents onto a share on another server. Pegs the CPU, shakes up the RAM utilization, hammers the local hard drive and jacks up my network utilization.

      Works with people too. I learned that at PHB school.

  • Other mini clusters (Score:2, Informative)

    by JimmyQS ( 690012 )
    A few weeks back I submitted an article about some mini-clusters we made at the Institute for Simulation and Training at the University of Central Florida. Here's a link: http://helios.engr.ucf.edu/beowulf/miniature.phtml " [ucf.edu]
  • Can you imagine the Windows CE license fee of a cluster of these?
  • this one is even smaller, and has no moving parts:

    http://pages.cpsc.ucalgary.ca/~mirtchov/p9/cluste r / [ucalgary.ca]

    the laptop is used as a file- and authentication server (frontend to the cluster).

    there's virtually no limit to how much it can grow (using Plan 9 as the underlying OS is key here).

    also note the cute switch.

    at USENIX this year this baby stole the show at the LinuxBIOS and Plan 9 BOFs.

  • Next: Wireless wristwatches! (Score:3, Funny)

    by CrystalFalcon ( 233559 ) on Wednesday October 08, 2003 @04:25PM (#7166256) Homepage
    I can't WAIT for the first 802.11g-enabled wristwatch, imagine being able to link up ad-hoc to a dozen other watches and be able to... ...uhm... ...to tell time twelve times as efficiently!

  • Spontaneous supercomputer. (Score:3, Funny)

    by YouHaveSnail ( 202852 ) on Wednesday October 08, 2003 @04:27PM (#7166281)
    For some geeky reason, the notion that you could spontaneously assemble a fairly powerful "machine" just by getting together with a bunch of suitably equipped friends really appeals to me.

    Mark my words: this meme will eventually find its way into a movie. A bunch of people will be trapped together. and the obvious geek will hook a bunch of their PDA's together in order to decode a message or open some sort of cypher lock which will disarm the deadly hazard and free the people with mere seconds to spare. And, of course, he'll get the girl.
  • of an idea I had in high school to cluster TI83's. We just couldn't figure out how to get them to talk to more than one other unit without cable switching...
    • And here's the limit on your calculator (YES, I have WAY too much free time on my hands ;-))
      9999999999999999999999999999999999999999999999999 9 99999999999999999999999999999999999999999999999999 99999999999999999999999999999999999999999999999999 99999999999999999999999999999999999999999999999999 99999999999999999999999999999999999999999999999999 99999999999999999999999999999999999999999999999999 99999999
    • Throwing negative numbers into this thing gives us more interesting output.

      I like it.
  • This is not new...except that the code is running on top of PocketPC OS. Cactuscode.org [cactuscode.org] used the 3600 series iPaqs, 340 MB microdrives, Linux and I am assuming a PC Card Ethernet connection for the networking. They actually have code on this site. So, the only thing new that Spb is doing is using the IR Ports and using Pocket PC as the os du jour. I would rather see BT connections being used for this and setting something up at a conference where nodes could be added as people walk in and out of an Expo
  • There is so much computing power around me, not just in my organizer and in my desktop pc.

    Of course, many microcontrollers have an unchangable firmware, but not all of them:

    camera (PPC860 @ 66mHz)

    mobile phone (80C166 @ 16mhz)

    car (engine electronics, SAB80C517 @ 12mhz)

    ISDN phones

    washing machine. (No kidding.)

    And these are only the devices, where I know for sure that they allow firmware upgrades.

    Where can I find Linux for washing machines?

    Markus

  • The article didn't speak to any redundancy gains that come with the cluster. Nor did it speak to cost comparison. This looks like the beginning of a major direction in future development.
  • Back around 1997 my group built a cluster of laptops running MPI for kicks. We laughed at how small it was stacked so we talked about making things even smaller. We talked about getting a number of Palms together or some other such tiny form factor machines and selling "6-packs" of clustered computing power - even to the point of laughing about making flanisters to put them in or cardboard "beer-like" holdersr Of course, back then, the aggregate computing power of a "6-pack Cluster" was about equal to th
  • Imagine a Beowulf clust- oh... wait a minute...

    Um.

    Oh! In Soviet Russia, the clust- ah forget it... the moment has passed. :(

  • Excellent, They clustered a bunch of expensive management toys together and achieved about performance than a cheap desktop computer that costs less than the price of one of these toys?

    Give them a medal - AND a slashdot leading story!

    What next? how about a cluster of abacus? hey - wouldn't that be great! I feel excited already!

    This is not some uber-hack, it is a trivial bit of work that a couple of half drunk people with a CE development system could throw together in a few hours!

    How about some better i
  • You just really have to run SETI on the road.
  • 12 are called a cluster. Just one is called a piece-of-crap. I should know I own one.

  • The end is near!! (Score:3, Funny)

    by gbulmash ( 688770 ) <semi_famous.yahoo@com> on Wednesday October 08, 2003 @05:43PM (#7166977) Homepage Journal
    What happens when enough of these pocket PC's network and an intelligence develops that determines humanity to be a threat to its existence?

    I envision nightmarish scenes out of "Small Soldiers [imdb.com]"... lilliputian armies of talking Barbies, chasing people down and carving them up.

    I don't know about you, but the day my Palm Tungsten's calendar shows I've got a 4 p.m. meeting with Death, I'm a headin' for the hills.

    - Greg

  • 1: Welcome our beowulf cluster in soviet russia overlords,
    2: but can they run Linux???
    3: Profit!!!

    Serious point - what is the origin of the beowulf/soviet/overlord/profit!!/etc jokes? Anyone done an FAQ? Anyone care?
  • How is this different from several people playing multiplayer Pocket Quake [pocketmatrix.com] over WiFi, simulating in parallel a virtual 3D environment on their Pocket PCs?

    Pocket Quake is almost 3 years old [slashdot.org].

    Dan East
  • with a better operating system running on these, you'd have to wonder would a cluster of zauruses be able to do better? or is the limit of pda style clustering what we've seen here. (at least with the available computing power right now)
  • all we need is a little wifi and people could converge for a brief time to create an uber cluster and the wonder off to sip coffee and the local coffee shop.
  • From the article:
    It turns out that even a small cluster consisting of twelve nodes is still slower than a mainstream Pentium II-class desktop PC.
    Last I checked, a P-II wasn't anybody's idea of a supercomputer. Ever.
  • In the amount of time it takes for a successful transmission of just a few bytes over infrared, the pentium machine would have probably ran several thousand iterations. Infrared is horribly slow, regardless if you're just sending a few bytes at a time.
  • You take 11 PPC handhelds @ $800 each for a total of $8800 and build a beowulf cluster. Yet it's still runs slower than single $1500 laptop.

    Only real world application I see is the defense department. Same people who are buying $800 hammers and $1200 toilet seats.

    Man Holmes
  • it does point out some interesting possiblilities for the future of handhelds with wireless connectivity working in ad-hoc computing clusters

    If I had a handheld with wireless connectivity, I'd rather have it belonging to a cluster of "real" computers...

  • "future of handhelds with wireless connectivity working in ad-hoc computing clusters."

    Isn't this what Cell is supposed to be? Only with like, toasters and PS3s?
  • These are the guys who ported WTL to Windows CE [pocketpcdn.com].
  • 'If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens?' - Seymour Cray
  • If ALL pc devices could be part of one massive cluster, it would make things a lot more efficent.

    You just 'plug-in' with your device, and you get instant large scale power and storage.

    It would reduce costs of devices, and size.. would make palmtops as generic as a stickypad you just lay around.

    Though we arent quite there in the wireless technology department to make it practical.... yet..

  • Did anyone else read Spb Software as SBD Software? Or was it just me :-)

Slashdot Top Deals

Your password is pitifully obvious.

Close