Supercomputer built from Raspberry Pis and Lego


A team of computer scientists at the University of Southampton in the UK created a supercomputer out of 64 Raspberry Pi matchbox Linux-on-a-chip computers and Lego. The team included six year old James Cox, the son of project lead Professor Simon Cox, "who provided specialist support on Lego and system testing."

Here's a PDF with instructions for making your own Raspberry Pi/Lego supercomputer.

Professor Cox comments: “As soon as we were able to source sufficient Raspberry Pi computers we wanted to see if it was possible to link them together into a supercomputer. We installed and built all of the necessary software on the Pi starting from a standard Debian Wheezy system image and we have published a guide so you can build your own supercomputer.”

The racking was built using Lego with a design developed by Simon and James, who has also been testing the Raspberry Pi by programming it using free computer programming software Python and Scratch over the summer. The machine, named “Iridis-Pi” after the University’s Iridis supercomputer, runs off a single 13 Amp mains socket and uses MPI (Message Passing Interface) to communicate between nodes using Ethernet. The whole system cost under £2,500 (excluding switches) and has a total of 64 processors and 1Tb of memory (16Gb SD cards for each Raspberry Pi). Professor Cox uses the free plug-in ‘Python Tools for Visual Studio’ to develop code for the Raspberry Pi.

Professor Cox adds: “The first test we ran – well obviously we calculated Pi on the Raspberry Pi using MPI, which is a well-known first test for any new supercomputer.”

Engineers Build Supercomputer Using Raspberry Pi, Lego [Parity News]

Southampton engineers a Raspberry Pi Supercomputer [Press release]

(Images: Simon J Cox 2012)

29

  1. what makes this a super computer?
    is it the fact that you have so many nodes?
    As surely all modern computers are super computers compared to past.

    1. The person that wrote the press release probably just got the capitalisation wrong. We are used to talking about megabit (Mb) / gigabit (Gb) in relation to bandwidth so  confusing the two is easily done.

  2. Neat idea, and looks great. Would be interested in seeing benchmarks against a virtual equivalent machine though

    1. Right? 3 months from registration to delivery for me.  For one.

      Love mine, best £25 media center a person could want.  But how in the hell did they get 64 of them?  That said the Pi was built for educational purposes in mind, so maybe the uni got a little preference…

  3. Depending on what you want to compute this isn’t exactly good value:

    Each Pi can apparently compute at .175 GFLOP/s, so in theory this cluster could run at a max of 11.2 GFLOPS/s.

    A single £250 graphic card (eg. OpenCL on a Radeon 5870) can compute at 2.72 TFLOP/s, or 240 times faster for a tenth the price.

    Being general purpose CPUs the Pis may be able to outperform a gfx card on some types of calculations and would probably be easier to program tho.

    Interestingly a Radeon 5870 would have been the worlds fastest supercomputer in 1999 http://en.wikipedia.org/wiki/History_of_supercomputing#Historical_TOP500_table

      1. Yep!  For generic computations, graphics card programming is still a bear in terms of programming tools.  Many highly-parallel scientific applications are already written using MPI as the standard… it’s worth the speed hit to not have to reprogram.  I’m definitely looking at this!

    1.  I got the impression this was more about demonstrating *how* to set up a supercomputer cluster for university students rather than actually being something you’d actually use for something!

      Much cheaper and power friendly to use such cheap low power PCs for playing around with :)

  4. “The team included six year old James Cox, the son of project lead Professor Simon Cox, ‘who provided specialist support on Lego and system testing.'”

    I hope that kid has business cards. “James Cox, Lego Specialist.”

  5. “The first test we ran – well obviously we calculated Pi on the Raspberry Pi using MPI, which is a well-known first test for any new supercomputer.”

    They had me at the Pi on Pi action. :-)

Comments are closed.