Slashdot Mirror


University Team Builds Lego and Raspberry Pi Cluster

hypnosec writes about a neat little hack using Lego, Raspberry Pis, and Scratch to construct a "supercomputer." From the article: "A team of computational engineers over at the University of Southampton led by Professor Simon Cox have built a supercomputer using Raspberry Pi and Lego. The supercomputer is comprised of 64 processors, 1TB of storage (16GB SD cards in each of the Raspberry Pis) and can be powered on using just a single 13-amp mains socket. MPI is used for communications between the nodes through the ethernet port. The team managed to build the core of the supercomputer for under £2500. Named 'Iridis-Pi' after University of Southampton's supercomputer Iridis, the supercomputer runs software that was built using Python and Scratch. Professor Cox used the free plug-in 'Python Tools for Visual Studio' to develop code for the Raspberry Pi." Lots of pictures of the thing, and a howto on making your own.

109 of 147 comments (clear)

  1. Re:Memory by Dave+Whiteside · · Score: 1

    yeah 64* 256MB = 16GB
    less than that is usable
    even if you use the 240Mb split
    but you do get 64 processors and 48*64 gpu processors - though using those is going to be a bit hard for a while

    --
    who where what when now?
  2. Want by Sparticus789 · · Score: 2

    Gussy it up however you want, Trebek. What matters is does it work? Will the Rasperry PI supercomputer calculate large prime numbers? Because I've ordered devices like that before - wasted a pretty penny, I don't mind telling you. And if the Rasperry PI supercomputer works, I'll order a dozen!

    --
    sudo make me a sandwich
    1. Re:Want by the_humeister · · Score: 5, Informative

      Unfortunately, the term "supercomputer" isn't really being used properly. They built a cluster of computers, sure. But "supercomputer"??? Hardly. The Raspberry Pi uses a processor based on ARM v6. Lemme give a single-threaded rendering comparison (Povray 3.6 running the benchmark scene (here's what the benchmark image output looks like) with my old HTC Aria, which uses a Qualcomm MSM 7227 processor and has similar processor specs as the Raspberry Pi (ARM v6 + VFP2 floating point hardware):

      HTC Aria (MSM 7227 @ 0.6 GHz) *
          Debian 6.0(armel), gcc 4.4, -mfloat-abi=softfp -mthumb
          Parse Time: 0 hours 1 minutes 3 seconds (63 seconds)
          Photon Time: 0 hours 53 minutes 49 seconds (3229 seconds)
          Render Time: 57 hours 31 minutes 41 seconds (207101 seconds)
          Total Time: 58 hours 26 minutes 33 seconds (210393 seconds)

      For comparison, here's a faster ARM processor from my Samsung Galaxy S II:

      Exynos 4210 @ 1.2 GHz (ARM Cortex A9),
          Debian 7.0(armhf), gcc 4.6, -mcpu=cortex-a9 -mhard-float -mthumb -mfpu=vfpv3 -ffast-math -funsafe-math-optimizations
          Parse Time: 0 hours 0 minutes 4 seconds (4 seconds)
          Photon Time: 0 hours 1 minutes 33 seconds (93 seconds)
          Render Time: 1 hours 26 minutes 34 seconds (5194 seconds)
          Total Time: 1 hours 28 minutes 11 seconds (5291 seconds)

      And here's from an Intel Core i5 2400s @ 2.5 GHz:

      Core i5 2400s @ 2.5 GHz, Ubuntu 12.04, gcc 4.6, -march=corei7-avx
      Total Scene Processing Times
          Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
          Photon Time: 0 hours 0 minutes 14 seconds (14 seconds)
          Render Time: 0 hours 10 minutes 12 seconds (612 seconds)
          Total Time: 0 hours 10 minutes 27 seconds (627 seconds)

      The ARM v6 processor took more than 2 days to render something that takes 10 minutes on a Core i5. So, "supercomputer" this cluster is not.

      * You may say, "Hey, this test is running using soft-float! If you used hard float, it'd be faster!" Well, you would be right that it would render faster under hard float, but this processor still wouldn't finish rendering in less than a day, let alone come anywhere close to Core i5 or Cortex A9.

    2. Re:Want by the_humeister · · Score: 2

      Forgot to add the following:

      The image rendered is 384 x 384 pixels. MSM 7227 results are 0.70 pps and 1.17 pps/GHz. Raspberry Pi is runnying at 700 MHz, so it should theoretically get 0.82 pps. Its possible (and fairly easy) to split up the rendering among all the CPUs in this cluster with some custom scripting, so this benchmark image could theoretically render at 52.42 pps. That Core i5 2400s I mentioned above renders at 235.18 pps!

    3. Re:Want by Sparticus789 · · Score: 2, Interesting

      There's a bit of apples and oranges comparison there. You are comparing single-core processors to a quad-core processor. Of course the i5 is going to be faster. It would be better to divide the performance of the i5 by 4, to represent the performance of a single core of the processor.

      There's also a cost comparison. Just the i5 processor is ~$200, not to mention the motherboard, RAM, etc. Let's just say you can build a computer with an i5 for about $800 That's the same price as 32 Raspberry PIs. So if you take the MSM 7227 processing time and divide by 32, you get ~1.8 hours. Not stellar, by any means. However it is an interesting figure. There's also power requirements, cooling requirements, etc.

      Not saying that everyone should flock to SoC cluster computing, but the story is interesting nonetheless.

      And perhaps Celebrity Jeopardy was before your time.

      --
      sudo make me a sandwich
    4. Re:Want by the_humeister · · Score: 3, Informative

      Povray 3.6 is single threaded, so all results are single threaded.

    5. Re:Want by Sparticus789 · · Score: 1

      Processor comment retracted.

      --
      sudo make me a sandwich
    6. Re:Want by dgatwood · · Score: 1

      You are comparing single-core processors to a quad-core processor. Of course the i5 is going to be faster. It would be better to divide the performance of the i5 by 4, to represent the performance of a single core of the processor.

      Based on those numbers, the quad-core i5 processor is approximately equal in performance to 335 Raspberry Pi cores (at this type of computation). Thus, even a single-core of the i5 would still be equivalent to almost 84 Raspberry Pi cores, and costs only about $600 even if you buy it pre-built as a laptop. By contrast, 335 Raspberry Pi machines would cost almost $12,000 (or more like $16k by the time you add power supplies, power cords, and flash cards). Even dividing by four, 84 of them (with power and flash) would cost about as much as a new Mac Pro with 12 cores. That, in turn, is equivalent to somewhere on the order of a Porsche Boxster worth of Raspberry Pi units as far as computing power goes. :-)

      And they're also not remotely efficient in power-per-watt. That Mac Pro takes about as much power as only a hundred Raspberry Pi boards, assuming a mythical 100% efficient power supply for the Pi, but would outperform a couple of thousand of them. So current Intel offerings are somewhere around 20x as efficient as a chip based on an eight-year-old ARM cell design. No surprise there.

      But that's really not the point. In performance-per-dollar or performance-per-watt, clusters of several-year-old CPUs will almost never be cost-effective. What makes them interesting is that they can be a cheap platform for letting folks test cluster-computing (distributed multiprocessing) apps experimentally without investing a huge amount of money. When testing how well an algorithm scales, assuming your tests don't tie up the hardware for such a long period of time that other folks in the class can't use it, it really doesn't matter how fast the total performance of the hardware is. What matters is the number of nodes. Thus, for supercomputer research, these sorts of devices are seriously cool.

      --

      Check out my sci-fi/humor trilogy at PatriotsBooks.

    7. Re:Want by CastrTroy · · Score: 4, Informative

      Actually, Just ran a test, because I was a little amazed that the ARM 6 was so much slower than the A9. Here are my numbers. Parse Time: 0 hours 0 minutes 14 seconds (14 seconds) Photon Time: 0 hours 5 minutes 43 seconds (343 seconds) Render Time: 5 hours 58 minutes 53 seconds (21533 seconds) Total Time: 6 hours 4 minutes 50 seconds (21890 seconds) While the Raspberry Pi wasn't faster than the A9 (didn't expect it to be) it was way faster then ARM6 you tested on. Most likely due to the fact that it uses hard float.

      --

      Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
    8. Re:Want by petermgreen · · Score: 2

      You may say, "Hey, this test is running using soft-float! If you used hard float, it'd be faster!"

      Massively faster

      http://www.raspberrypi.org/phpBB3/viewtopic.php?f=9&t=4256&start=175

      --
      note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
    9. Re:Want by petermgreen · · Score: 3, Informative

      The GPs figures are off. He is using a horrible compiler setup, not only is he using the softfloat calling convention, he is using -mthumb which AIUI will prevent the code from making direct use of the hardware FPU (and I suspect he uwas using debians version of libc preventing indirect use of the hardware fpc through libc routines)at all on armv6. According to hexxeh the povray benchmark under raspbian gives the following results under raspbian on a PI.

      Total Scene Processing Times
      Parse Time: 0 hours 0 minutes 16 seconds (16 seconds)
      Photon Time: 0 hours 5 minutes 57 seconds (357 seconds)
      Render Time: 6 hours 13 minutes 57 seconds (22437 seconds)
      Total Time: 6 hours 20 minutes 10 seconds (22810 seconds)

      http://www.raspberrypi.org/phpBB3/viewtopic.php?f=9&t=4256&start=175

      Your price figures are off too. An i5 based compute node can be built for more like $500

      Similarlly the real price of a Pi node is quite a bit more than $25. Firstly the Pi you can actually buy and would want for this task (clustering needs networking support) has a base price of $35 not $25. Secondly that price excludes things like the power power supply the SD card, the network cable and the mouning hardware. The real cost of a Pi node is probablly more like $50.

      So the Pi is about 10 times lower per node than the i5

      My overall conclusion is if compute power per dolar is your goal then a smamler number of i5s is a much better bet than a larger numer of Pis.

      --
      note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
  3. Real news for this article by BooMonster · · Score: 5, Funny

    One university managed to get a hold of 64 Raspberry PI units.

    1. Re:Real news for this article by Anonymous Coward · · Score: 1

      They must have ordered from Farnell, because they sure as hell wouldn't have received them from RS.

    2. Re:Real news for this article by petermgreen · · Score: 1

      As someone who tried to order Pis through a university early on we didn't get any special treatment.

      But anyway the suppliers have been taking multi-unit orders for a while now and the farnell group have actually been filling them.

      --
      note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
  4. Not really interesting by filmorris · · Score: 1

    They should have built a Beowulf cluster. The regular one is such a cliché.

    --
    "Hello, IT... Have you tried turning it off and on again? Yeah... No problem."
  5. erm, good by Anonymous Coward · · Score: 1

    So a cluster of 64 pi boards don't exceed ~3 kilowatts... Why would you expect them to given that they are supposed to run from a 5V supply at 1A (5W * 64 = 320W)

    1. Re:erm, good by jones_supa · · Score: 2

      So a cluster of 64 pi boards don't exceed ~3 kilowatts... Why would you expect them to given that they are supposed to run from a 5V supply at 1A (5W * 64 = 320W)

      The comparison point is that a 64-node cluster of regular PC hardware couldn't fit behind a basic mains line.

    2. Re:erm, good by snowraver1 · · Score: 1

      No, but a single computer that has 16 or rmore cores, many GB of RAM and multiple TB of storage would. I'd guess it would out perform the PI cluster too.

      --
      Copyright 2010. All rights reserved. This comment may not be copied in any way including, but not limited to caching.
    3. Re:erm, good by Desler · · Score: 1

      Well, yes. It would also be magnitudes faster than this.

    4. Re:erm, good by jones_supa · · Score: 1

      Oh, that might of course be true.

    5. Re:erm, good by M1FCJ · · Score: 3, Insightful

      Of course, this being a teaching tool, having the performance would mean nothing. Having the discrete computing units do.

    6. Re:erm, good by lister+king+of+smeg · · Score: 2

      sounds like the old mainframe vs x86 commodity server argument. small mobile tech will take the place of much of the current PC's because grandma doesn't need a PC to check email, facebook, play solitaire, and watch netflix, but the PC won't disappear just like the mainframe didn't disappear because it still has many more use's. in fact there are more mainframes sold per year now then there were in the pre-PC era. what we will see come from this will be a continuation of the trend toward less and less expensive hardware at lower power. facebook for example is trying to move to arm based server farms to decrease their energy costs.
         

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
  6. Not a `Super' computer by colin_faber · · Score: 3, Insightful

    Sorry but doesn't even crack the top 10,000's in machine performance, not exactly a super computer. A cluster yes. Super computer, HPC machine, etc. no.

    1. Re:Not a `Super' computer by vlm · · Score: 2

      A supercomputer is any overall system that's IO limited not CPU limited like most machines. At least at full theoretical CPU use. Hard to define a rasp pi as anything other than IO limited, so... An alternative def more popular recently is programmer limited as in its hard to parallelize some algorithms. Either way it fits.

      --
      "Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
    2. Re:Not a `Super' computer by slim · · Score: 2

      A supercomputer is any overall system that's IO limited not CPU limited like most machines.

      So if I cripple all IO to my 486, except a 300 baud modem, I've built a supercomputer?

    3. Re:Not a `Super' computer by Desler · · Score: 1

      Actually many normal desktops are IO limited for a number of applications. That hardly qualifies them as supercomputers.

    4. Re:Not a `Super' computer by localman57 · · Score: 4, Funny

      You did? That's amazing! I'm submitting it as a slashdot article now.
      Man Builds SuperComputer from a 486 and a 300 Baud Modem

    5. Re:Not a `Super' computer by MozeeToby · · Score: 1

      That seems like a pretty poor definition to me. In fact, it seems like something could be or not be a supercomputer depending on what job you have running on it.

    6. Re:Not a `Super' computer by RaceProUK · · Score: 1

      A supercomputer is any overall system that's IO limited not CPU limited like most machines.

      Do all your computers have a 286? Last time I checked, modern PCs are always IO limited.

      --
      No colour or religion ever stopped the bullet from a gun
    7. Re:Not a `Super' computer by colin_faber · · Score: 2

      Sorry but this statement is completely wrong. HPC (Super computers) are a moving target and the term represents the top 500 (or less) machines world wide.

      This statement is also wrong in respect to I/O limitations as bottle necks have nothing to do with whether or not a system is considered a "Super computer".

    8. Re:Not a `Super' computer by ultranova · · Score: 1

      A supercomputer is any overall system that's IO limited not CPU limited like most machines.

      Isn't that pretty much any system nowadays? Most desktop computers are limited by their disk-to-memory and memory-to-registers speeds.

      --

      Forget magic. Any technology distinguishable from divine power is insufficiently advanced.

  7. For the old timers by thammoud · · Score: 4, Funny

    Can we now retire "Bewolf cluster" jokes?

    1. Re:For the old timers by wild_quinine · · Score: 3, Funny

      Call me old fashioned, but my time on Slashdot has taught me that it's much better to *imagine* a cluster than to create one.

    2. Re:For the old timers by Anonymous Coward · · Score: 3, Funny

      Call me old fashioned, but my time in jail has taught me that it's much better to *imagine* a naked, petrified Natalie Portman covered in hot grits than to create one.

    3. Re:For the old timers by dwywit · · Score: 1

      RPi cluster = Beowolf 2.0

      --
      They sentenced me to twenty years of boredom
  8. Meme overload by Plammox · · Score: 4, Funny

    Aaaargh...imagine a.....in the Soviet Union....***carrier lost***

    1. Re:Meme overload by Howard+Beale · · Score: 2

      OMG!!! PONIES!!!!!!

    2. Re:Meme overload by Narnie · · Score: 1

      So... are they mining bitcoins on that cluster yet?

      --
      greed@All_Evils:~#
    3. Re:Meme overload by avandesande · · Score: 1

      In Korea only old people eat raspberry pie....

      --
      love is just extroverted narcissism
  9. Abuse of the term "Supercomputer" by CajunArson · · Score: 3, Insightful

    Calling this thing a cluster.. fine.
    Calling it interesting for students to learn about how clusters work... fine.
    Calling it a supercomputer? Maybe if the University of Southampton got sucked into a time vortex to the early 1990's, and even then while the raw theoretical number crunching capability of the RPis would be impressive, the lackluster I/O and interconnects would mean that even supercomputers of that time would still win on many common workloads.

    --
    AntiFA: An abbreviation for Anti First Amendment.
  10. It's called a bramble, guys by Bovius · · Score: 5, Funny

    Cluster of Raspberry Pis = Bramble. Slashdot has been so drooly over every nitpicky update about these, I thought everyone here would know that by now.

    1. Re:It's called a bramble, guys by slim · · Score: 1

      A quick google validates this. But I abhor it. It's blackberries, not raspberries, that grow on a bramble.

    2. Re:It's called a bramble, guys by History's+Coming+To · · Score: 4, Funny

      Brambles are blackberries, you'd need a cluster of mobile email devices to use that name. I believe this is the more generic "Thicket".

      --
      Please consider this account deleted, I just can't be bothered with the spam anymore.
  11. Mixed feelings by gblackwo · · Score: 2

    Whenever I see "professional" projects like this use legos- I have mixed feelings. Here is another example, a lab using legos for automation

    I love to see legos doing advanced things, but for a chassis? I feel like people can be very smart, but sometimes afraid to learn how to build something with their hands. The lab example I posted above is at Cambridge University. Cambridge has a very competent engineering department, why not reach out to them?- It could have made for an excellent project for some engineering students.

    I'm reminded of the very cited researcher who reinvented some calculus instead of simply reaching out to someone in another department for help

    1. Re:Mixed feelings by RaceProUK · · Score: 1

      I love to see legos doing advanced things, but for a chassis? I feel like people can be very smart, but sometimes afraid to learn how to build something with their hands. The lab example I posted above is at Cambridge University. Cambridge has a very competent engineering department, why not reach out to them?- It could have made for an excellent project for some engineering students.

      Aside from your Lego-assembling robot, Lego is always assembled by hand. It's also cheaper and faster to build a Lego chassis than get the engineers to weld up one from mild steel.

      --
      No colour or religion ever stopped the bullet from a gun
    2. Re:Mixed feelings by xaxa · · Score: 1

      "The rack for the supercomputer has been built using Lego under the guidance of Professor Cox's son James Cox (aged 6)."

      Also "reached out" is a stupid phrase.

    3. Re:Mixed feelings by Dishevel · · Score: 1

      Engineers do not weld.
      They use a computer to let others know that they need to weld those two pieces of steel together.

      --
      Why is it so hard to only have politicians for a few years, then have them go away?
    4. Re:Mixed feelings by jones_supa · · Score: 1

      Plastic is always a bit problematic case material for larger-size electronics. It might build up some static electricity, and you can't create grounding and RF shielding. However I agree that LEGO is simply practical for sculpting quick prototypes.

    5. Re:Mixed feelings by gblackwo · · Score: 2

      It is this kind of thinking that makes people build something out of legos instead of trying to learn a new skill.

    6. Re:Mixed feelings by Anonymous Coward · · Score: 1

      I love to see legos doing advanced things

      So would I, whatever "legos" might be. Sounds Greek.

      Meanwhile all we have are these LEGO bricks. That being the plural of LEGO brick.

  12. Re:So this is why... by Ice+Tiger · · Score: 1

    Your's wasn't the only one they had!

    --
    "Because we are not employing at entry level, offshoring will kill our industry stone dead."
  13. Re:Memory by slim · · Score: 4, Insightful

    SD cards are memory. Wikipedia: "Secure Digital or (SD) is a non-volatile memory card format for use in portable devices.".

    Hard disks are memory too. So are tapes. So are CDs.

    Just because you might habitually use "memory" as shorthand for RAM, doesn't mean it's the only meaning of the word. And why do you think we need the "Random Access" disambiguation?

  14. Computer does things a computer can do! by slim · · Score: 5, Insightful

    I'm a big fan of the RP project. But I'm a bit bored of seeing news items in which someone does something with this Linux box, which obviously a Linux box can do. Raspberry Pi compiles C! Raspberry Pi controls a robot! Raspberry Pi runs MAME! Well of course it does, it's a little PC, and that's what PCs can do.

    1. Re:Computer does things a computer can do! by Selfbain · · Score: 1

      Last I checked PC == Personal Computer regardless of architecture.

      --
      Well, it has never been successfully tested.
    2. Re:Computer does things a computer can do! by slim · · Score: 1

      Did the definition of PC change. Last I checked PC == x86.

      The meaning of PC was never really solid. It just means "Personal Computer". In this case I just meant "something with a CPU, RAM, a filesystem, keyboard, monitor and ethernet. (although the keyboard and monitor aren't relevant to this particular project).

    3. Re:Computer does things a computer can do! by mapkinase · · Score: 1

      I agree. The original post also omits relevant number of Lego blocks used for construction of the cluster.

      --
      I do not believe in karma. "Funny"=-6. Do good and forbid evil. Yours, Oft-Offtopic Flamebaiting Troll.
    4. Re:Computer does things a computer can do! by the_humeister · · Score: 1

      PC = personal computer. Also, running Windows isn't theoretically impossible, just very very slow if via emulation x86 (not least of which is due low memory and swap thrashing). Or you could get the source code for the ARM version of Windows 8 (which I'm sure the academic person could get) and hack it to work too.

    5. Re:Computer does things a computer can do! by TheTrueScotsman · · Score: 1

      Or they could just use a desktop PC with 64 VMs.

    6. Re:Computer does things a computer can do! by JosefSit · · Score: 1

      If you put an armada of cute kittens in front of a horse carriage, you do not get a super carriage - just a super cute, very slow, high maintenance carriage...

    7. Re:Computer does things a computer can do! by aliquis · · Score: 1

      Now you're thinking inside one box!

  15. Re:Memory by Anonymous Coward · · Score: 1

    memory is shorthand for RAM? Which one is easier to say, and write?

  16. Recommendation by puddingebola · · Score: 1

    I would like to recommend the red and white suited astronaut lego people to maintain the server, or to work as sysadmins. They seem very dependable. If not them, then maybe the Lego people from the 70s that didn't have the smiley face painted on them. They seem more analytical and inclined to this type of work. Anybody remember them?

    1. Re:Recommendation by Ambvai · · Score: 1

      Okay, now those guys are just plain creepy and surreal: http://en.wikipedia.org/wiki/File:Lego-minifigs-old.jpg

  17. Re:Memory by K.+S.+Kyosuke · · Score: 1

    Just because you might habitually use "memory" as shorthand for RAM, doesn't mean it's the only meaning of the word.

    Exactly, everybody has known for at least half a century that we are supposed to call that one "core storage".

    --
    Ezekiel 23:20
  18. That's a supercomputer? by MetalliQaZ · · Score: 1

    Wouldn't a dual Xeon server be able to easily out-muscle that "supercomputer"?

    --
    "Here Lies Philip J. Fry, named for his uncle, to carry on his spirit"
    1. Re:That's a supercomputer? by bWareiWare.co.uk · · Score: 1

      A Core 2 era dual-core Xeon would out-muscle them unless they can use the propitiatory GPU. With the GPUs they could spank any single chip.

  19. Re:Memory by slim · · Score: 1

    I work with some people who persist in calling hard disks DASD -- even on PC hardware.

  20. Re:Supercomputer? by Kupfernigk · · Score: 3, Interesting
    Yes, you are missing something (though I have slight reservations about the 16 cores to a die CPUs you claim to be using). There's this thing called education...your large server running loads of VMs is not going to be nearly as useful or informative at getting the ideas across as a rig like this. There is a big difference between working with virtual networks and seeing the hardware of a real network, as well as being able to program the thing with "small" languages without monster frameworks just to make anything happen.

    However, you do win a "Miserable git" award for being unpleasant about Prof. Cox.

    --
    From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
  21. Re:Memory by SJHillman · · Score: 1

    When it's absolutely necessary to differentiate, use "primary storage" (typically RAM) and "secondary storage" (typically hard disk). After all, VRAM is essentially just using a secondary storage device for primary storage. RAMDISK is essentially using a primary storage device as secondary storage. The lines get pretty blurry if you try to say a specific physical device is only used as primary or secondary storage.

  22. SuperBe(rry) by SimplexBang · · Score: 1

    Supercomputer as in ' Super Structure' not as in ' Super Man'

    --
    Avoid your fears , or wonder at the past
  23. More power than a 1U machine, most likely by SuperBanana · · Score: 1

    I just did the math. The Pi community supposedly recommends a minimum of 1A@5V if you intend on using any peripherals, including ethernet. 700mAh is the minimum draw with *nothing* connected. 5W x 64 = 320W. That's quite close to the max capacity of the power supply for the dual-socket machine I mentioned. The E5-2620 processors have a max TDP of 95W each. Now, that doesn't count the auxiliaries - but there's still a 120W difference between typical power usage for the Pi, and MAXIMUM power usage for the Xeons, and I haven't even counted the power loss from the AC-DC power supply against the Pi (the rackmount machine's supply is ~95% efficient.)

  24. Re:waste of money / publicity stunt by slim · · Score: 4, Insightful

    Mmm, but it *is* a nice environment for *students* to experiment with the *principles* of parallel computing in a tactile manner.

    I began learning to code on an 8 bit 2Mhz CPU, with 32KB of RAM. If I wrote an inefficient loop, I'd often notice the slowness without benchmarking. If I was careless with memory, my program would crash. On my quad core laptop today, I only notice issues like that if I benchmark or do deliberate load testing. So working on low-spec systems is instructive.

    Likewise, working with clusters of low-powered units on a slow comms bus is going to teach these students a lot about optimising parallel programs. They're going to have to deal with race conditions, memory ceilings, etc. which might not even show up on faster systems.

  25. "Miserable Git" Awards by FilmedInNoir · · Score: 1

    Can I get nominated for that award? I'm creeped out by his Mini-Me son.
    All parents that dress their children as tiny doll versions of themselves actually.

    --
    Sig. Sig. Sputnik
  26. Re:waste of money / publicity stunt by hattig · · Score: 3, Insightful

    Exactly, whilst the system isn't powerful, it is instructive in cluster design and programming, which is very relevant at a university.

    They won't be running "real applications doing real calculations" on this thing. They'll be writing student-level clustered applications. For the price paid, it's probably a really instructive system for the university to have installed, if they make use of it in student courses and/or projects.

  27. Re:waste of money / publicity stunt by M1FCJ · · Score: 1

    So.. You have a little server there. Good luck with using it for teaching a bunch of students about how scalable clustered software works, how to write the software, what are the pitfalls and else.

    Good luck running 64 separate VMs on your small server (not saying it's not impossible but I really wonder which one is faster to set up) and you won't be able to test any of the very different interconnects that easily.

  28. Re:waste of money / publicity stunt by Simon+Brooke · · Score: 4, Insightful

    WOOOOOOOOOOOOOOSH!

    There is a whole lot of point missing going on here. Yes, you could build a faster computer cheaper using other hardware. But it wouldn't explain the concepts to children (and to first year CS students, which is pretty much the same thing) nearly so well. Throw together a heap of little itty-bitty boards each of which, individually is, as everyone knows, relatively low power, and knit them together with ordinary cat5 cable, and get out of the collection high compute performance, and you have something which will intrigue children|students and get them thinking about how it works. Show them an anonymous 1U box doing exactly the same job, and you won't get them thinking, because they can't immediately see and understand what it comprises and how it's put together. This is a teaching machine, not a practical machine. It's job is to teach students. It teaches students by being perspicuous.

    It's not (yet) a requirement for getting a Slashdot account to demonstrate that you have an IQ slightly south of that of a stick of used chewing gum, but some of you clearly haven't yet got that message.

    --
    I'm old enough to remember when discussions on Slashdot were well informed.
  29. You forgot BitCoins!, the MPAA, and Unity, etc. by sirwired · · Score: 5, Funny

    You clearly need to turn in your Slashdot commenter license... to REALLY entice editors to post a story, work BitCoins into the mix. Oohh... better yet, work in references to the MPAA, And Ubuntu, and whatever else can be stirred into the pot. References to MAME are old school... (although that can be forgiven, Mr. 4-digit UID.)

    How does this sound? "Raspberry Pi used to mine BitCoins to help pay an MPAA Lawsuit Fine. However, due to a security hole in Ubuntu caused by the new Unity interface, the new coins were stolen from the user by someone claiming to be affiliated with Anonymous. Wil Wheaton offers to sponsor a live D&D game played with Arduino-programmed robotic miniatures to make up for the lost funds."

    Did I miss anything?

    1. Re:You forgot BitCoins!, the MPAA, and Unity, etc. by rochrist · · Score: 1

      You need to blame either Apple or Microsoft.

    2. Re:You forgot BitCoins!, the MPAA, and Unity, etc. by Rotag_FU · · Score: 2

      Yep, you forgot about how kick-starter was used to fund the creation of the robotic miniatures. Also, the Raspberry Pi was actually running MineCraft which had a working implementation of an 8-bit processor that was doing the actual BitCoin mining. Researchers were observing the operation of the MineCraft processor using the Oculus Rift headset. In the future, the designers plan to port this all to the upcoming Ouya.

      There, I think that has it covered. :)

    3. Re:You forgot BitCoins!, the MPAA, and Unity, etc. by Razgorov+Prikazka · · Score: 1

      I think that your biggest contribution is putting in the words: OLD SCHOOL in your reply, although I'm almost completely convinced that you didn't mean it as a school that has been there for a long time.
      What I am really missing is an article about a school (especially in the 3rd world). Wasn't this whole RPi concept about bringing tech lessons to schools (in the 3rd world) in the first place? Yet I haven't seen 1 single contribution (and 1 would suffice) about a school in say, Angola, with kiddos all huddled around a RPi trying to get "hello world" on a tv-screen. That would be more impressive than a professor with years and years of academical skill's and drills getting Deb to run IMHO. That would be cool wouldn't it? (yeah I know, you've all been there, done that and got a "hello world"t-shirt, but for those kids it is a different ball-game)

      So to conclude, I think you are funny AND analytical! Congrats on that!

      --
      rm -rf --no-preserve-root / ...and let /dev/null sort them out...
    4. Re:You forgot BitCoins!, the MPAA, and Unity, etc. by dkf · · Score: 1

      You need to blame either Apple or Microsoft.

      And then, one day later, submit an almost-identical story that just blames the other one.

      --
      "Little does he know, but there is no 'I' in 'Idiot'!"
  30. Re:waste of money / publicity stunt by TheDarkMaster · · Score: 1

    I thought the same thing. It's much easier to notice errors in your code and how the whole thing works when using a machine that is slow or purposely slow.

    --
    Religion: The greatest weapon of mass destruction of all time
  31. So what are they going to use it for? by Guillaume+le+Btard · · Score: 1

    Calculate the billionth digit of pi?

  32. Re:waste of money / publicity stunt by slim · · Score: 1

    Good luck running 64 separate VMs on your small server (not saying it's not impossible but I really wonder which one is faster to set up) and you won't be able to test any of the very different interconnects that easily.

    Very easy indeed, and almost certainly quicker/easier to set up than the physical way, either using something like Vagrant or by rolling your own scripts to drive VirtualBox.

    However, I think it's instructive for students to do it the physical way first. By analogy: first understand LANs, then learn about VLANs.

  33. I love it...that got Many PI.... by haplo21112 · · Score: 1

    ....I can't even get one! I've been in the Queue since before initial release and still have yet to receieve mine, and even got an email two weeks ago about further delays!

    Still a really great accomplishement though.

    --
    Power Corrupts,Absolute Power Corrupts Absolutely, leaving one person(group)in charge is absolutely corrupt.
    1. Re:I love it...that got Many PI.... by ctid · · Score: 1

      RS components have had serious delays. I cancelled my order with them and instead ordered from CPC. It took approximately 48 hours for RS to refund my money and almost exactly the same amount of time for CPC to deliver.

      --
      Reality is defined by the maddest person in the room
  34. I would have got there first! by polyp2000 · · Score: 1

    But unfortuntately they would not sell more than one per customer - Unless you purchased one from both RS and Farnell.
    I'd even bought a rack mount case to house the cluster :(

    N...

    --
    Electronic Music Made Using Linux http://soundcloud.com/polyp
  35. Lego + Raspberry Pi by csumpi · · Score: 1

    This story has built out of Legos, and Raspberry Pis, so it's definitely worthy for the slashdot front page. But it could be better, like they called the order in from their Nokia phone and paid for it using Bitcoins.

  36. Re:Supercomputer? by silas_moeckel · · Score: 1

    You mean you have reservations about stock shipping AMD server procs? If you want education you want to be able to do things like artificially inflate the latency of the linking network that's easy to do on VM's. Test the effectiveness of different storage methods vs the type of workload. Looks at nodes with different processing capabilities. Honestly I find it amazing hard to fathom that it took a whole group of people to stack 64 SBC's load them with an OS and connect them up to a switch. This is a mornings work for an intern.

    Thanks for the miserable git been awhile since somebody called me names on the internet I'll try and be offended. I did not talk about the prof outside of mentioning that it was cool that be built it out of legos with his son.

    --
    No sir I dont like it.
  37. Re:waste of money / publicity stunt by SuperBanana · · Score: 1

    There is a whole lot of point missing going on here.

    You're the one missing the point here. I can fit in 1U what used to take an entire rack.

    When you can fit that kind of power into 1U, and given the massive leaps in computing power per core, traditional nodes-connected-by-networks clusters are applicable for far fewer people these days. What they should be teaching is proper multithreaded programming techniques.

    get out of the collection high compute performance

    Were you not paying attention when I said that 64 700mhz ARM nodes connected via USB (which requires enormous CPU overhead, on a processor with virtually no cache and slow busses, which means lots of out-of-cache memory access and context switching) with shitty, slow storage - does not make "high compute performance"? That cluster probably struggles to match one single 6-core Xeon.

    It's not (yet) a requirement for getting a Slashdot account to demonstrate that you have an IQ slightly south of that of a stick of used chewing gum, but some of you clearly haven't yet got that message.

    http://yourlogicalfallacyis.com/ad-hominem

  38. Re:Memory by Langalf · · Score: 1

    Heh, when I first came to work at my present job (30 years ago, ack!) One of the systems came with a drum storage system. It was never actually fired up; we immediately replaced it with a ramdisk emulation. Of course that system also had honest-to-god core memory. Those core stacks were WAY cool.

  39. Re:waste of money / publicity stunt by SuperBanana · · Score: 1

    There is questionable benefit to having that tactile experience a)extend beyond a few nodes - certainly nowhere near 64 b)have it on hardware which resembles embedded systems, not real compute nodes. In short, it's teaching 15 year old HPC technology in an era where you can fit 64 cores into 1U for a couple grand.

  40. Re:Memory by Hognoxious · · Score: 1

    Do they say TSO ehrn they're talking about a shell prompt?

    --
    Confucius say, "Find worm in apple - bad. Find half a worm - worse."
  41. Re:waste of money / publicity stunt by CastrTroy · · Score: 2

    When I was in university, I took a parallel computing course and we used MPI, same as these guys. Back then, all the personal machines were single core. If we were lucky we could test the program out by remote logging into the quad processor SUN machine. Guess what? We were able to learn quite a bit just running 64 different processes on the same box, even with just a single processor core. It would have been nice to have a machine around with 64 actual cores on it to see how things worked one everything was truly running in parallel, but we were able to do quite a bit with just a single machine.

    --

    Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
  42. Why would you build it yourself? by aliquis · · Score: 1

    Does it really have more computing power than a $2500 PC / worth of PC hardware (they did say core didn't they? So feel free to get 3-4 regular pieces of PC hardware in that case which it surely won't beat.)

    Nothing super about it. Just multiple pieces.

  43. Re:Memory by sootman · · Score: 1

    It's field-specific. If it doesn't forget everything after power cycling, we in the computer business call that "storage."

    In drag racing, quick and fast are different. In metallurgy, hardness and toughness are different. In typography, readability and legibility are different. In astronomy, revolve and rotate are different. Etc etc etc. Every field has terms like this and yes, it matters, and they should be used correctly when within that field. (Which we currently are: on a tech site, talking about technology.)

    --
    Dear Slashdot: next time you want to mess with the site, add a rich-text editor for comments.
  44. Re:Memory by houghi · · Score: 1

    I forgot. Sorry. My memory isn't what it used to be.

    --
    Don't fight for your country, if your country does not fight for you.
  45. 64 is a bad number for switching ugh. by PenguinJeff · · Score: 1

    Finding switches to do this would just suck. Most home switches are 8 port. So if you buy 9, 8 port switches and one 16 port 10 gig switch plug 7 in each of the first 9 switches that would get you to 63 that would leave one to plug into the hub. Plus each 8 port would need to be gig unless you can buy switches with single gig. or 4, 16 port gig switches with a ten gig uplink to a 10 gig switch any way you need enough to plug in all 64 +1 for the head node. any which way this is a switching nightmare. The head node needs a 10 gig so that it can handle all 64 nodes at 100 mbps. Each set of 8 gets close to a gig network traffic for the head node. so 64 nodes is 6400 mbps that is why I say 10 gig. Maybe I'm missing something but it seems complicated.

  46. not even close by Chirs · · Score: 1

    Even on modern PCs things like transcoding, compiling, etc. are still cpu-limited.

  47. memory is not really "I/O" by Chirs · · Score: 1

    While technically RAM is not part of the CPU itself, I think you'll find most people don't consider memory access when calling something "I/O limited". That's more along the lines of network ports, hard drive, USB, firewire, thunderbolt, etc.

  48. Re:waste of money / publicity stunt by rusty0101 · · Score: 1

    Something that I don't thing got much play in the article is that each of the 64 Pi boards has a SOC processor that in addition to the general purpose processor also includes a 48 core processor optimized for graphics. And yes in http://www.raspberrypi.org/archives/1967 they note that there is already code that can use those processors for graphics. I have little doubt that someone looking at the code can port one of the gpu processing libraries to make use of these processors for other numerically intensive purposes.

    So don't forget to add in sufficient video processing to provide 3072 cores of processing equivalence to your rig. I suspect that you can figure out how to do that as you've already calculated what you feel is the equivalent general processing equivalent (or better) for your sample system. I'm not sure that the resulting system would still fit in a 1u case, but it might.

    --
    You never know...
  49. Re:Memory by slim · · Score: 1

    It's field-specific. If it doesn't forget everything after power cycling, we in the computer business call that "storage."

    Er, in my CS textbooks (admittedly from the early 90s), memory and storage are interchangeable terms, and RAM counts as "storage" too.

    In those books, RAM is "primary storage" and hard drives are "secondary storage".

    By a different definition, coming from a different part of the computing field, primary storage is what you get from malloc() and secondary storage is what you get from fopen() -- but we all know how blurry *that* can get.

  50. waitlisted by spxZA · · Score: 1

    I've been waiting since 26 June for my Pi. And this guy has 64, just for kicks! WTF?

  51. Re:Not at all by Joce640k · · Score: 1

    I've got two...

    --
    No sig today...
  52. Nice to see someone do it... by coofercat · · Score: 1

    Quite seriously, I wondered about making a cluster of Pis to replace a desktop PC I have running in the loft. It really just runs some web servers, PHP, Mysql and a few other fiddly things. I wondered if I could potentially even dynamically boot up Pis to cover load (ie. spin up some extra web servers when load increases). My big problem is the DB though - I mainly use Drupal, so don't have separate read and write DB handles, so I can't scale MySQL horizontally. Also, the ethernet isn't very fast, so the interconnects probably wouldn't work very well either. I thought about maybe using Beaglebones or Pandaboards. Whilst that have faster CPUs, they still only have 100MB networking, so probably a little less than ideal. After working through all that, I don't suppose I'd save that much power (or space?) from the desktop.

  53. Re:Memory by heefeneet · · Score: 1

    memory is shorthand for RAM? Which one is easier to say, and write?

    Its also short for Bob.

  54. USB problem affect ethernet connectivity by Douglas+Goodall · · Score: 1

    How do you build a supercomputer out of processor modules that cannot reliably communicate with each other. The ethernet connectivity of the pi is based on a small module that attaches to the USB. I don't get it...

  55. Re:Memory by sootman · · Score: 1

    It varies, of course, but the most common current practice is "memory" (RAM) and "storage" (disk or solid-state -- the main thing is "long term" or non-volatile.). From the Wikipedia page on computer memory (which the GP did not quote):

    The term "storage" is often (but not always) used in separate computers of traditional secondary memory such as tape, magnetic disks and optical discs (CD-ROM and DVD-ROM). The term "memory" is often (but not always) associated with addressable semiconductor memory, i.e. integrated circuits consisting of silicon-based transistors, used for example as primary memory but also other purposes in computers and other digital electronic devices.

    --
    Dear Slashdot: next time you want to mess with the site, add a rich-text editor for comments.
  56. Costs, Currencies and Comparisons by andersh · · Score: 1

    Is your comparison fair? I don't work with hardware, so I claim no expertise here, but a quick shopping list shows:

    - $549 at NewEgg for one AMD Opteron 6262 HE Interlagos 1.6GHz 16-Core Server Processor
    - 4 x $549 = $2196 for just the CPUs?
    - £2500 (GBP) ~ $4047 (USD)
    - $4047 - $2196 = $1851

    Is it really that much cheaper or better for their purpose? They're not looking for power/$ ratios. Clearly you would have a few Dollars left over to buy the other required parts for your server.

    Please keep Purchasing Power Parity (PPP) in mind. Prices in the UK are not the same as in the US, simply using currency exchange rates does not give a meaningful comparison.