Making the Most of What You've Got

By Julianne Dalcanton | October 17, 2007 2:14 pm

I love stories like these:

Suffering from its exorbitant price point and a dearth of titles, Sony’s PlayStation 3 isn’t exactly the most popular gaming platform on the block. But while the console flounders in the commercial space, the PS3 may be finding a new calling in the realm of science and research.

Right now, a cluster of eight interlinked PS3s is busy solving a celestial mystery involving gravitational waves and what happens when a super-massive black hole, about a million times the mass of our own sun, swallows up a star.

As the architect of this research, Dr. Gaurav Khanna is employing his so-called “gravity grid” of PS3s to help measure these theoretical gravity waves — ripples in space-time that travel at the speed of light — that Einstein’s Theory of Relativity predicted would emerge when such an event takes place.

It turns out that the PS3 is ideal for doing precisely the kind of heavy computational lifting Khanna requires for his project, and the fact that it’s a relatively open platform makes programming scientific applications feasible.

CATEGORIZED UNDER: Computing, Gadgets
  • Freiddie


  • Sean

    Black holes, bah. Everyone knows the true purpose of PlayStation is to teach plays to college football quarterbacks.

  • Michele Vallisneri

    Unfortunately, it’s not quite _ideal_. It’s much faster at single-precision than double-precision math, because the latter is not so important for games and graphics…

    So either Gaurav is optimizing his code by making do with single precision, or it’s not running as fast as it could, had the creators of the Cell processor been more sympathetic to scientific applications. It is hoped future versions of the processor will remedy this.

  • Andy

    Stanford has a distributed computing project called Folding@home that includes PS3s. Like Seti@home, people give up spare cycles on their home PS3 to crunch protein folding problems. See this article from the SF Chronicle back in March.

  • bswift

    I thought about putting in an NSF proposal for 1,001 PS3s for doing cosmological simulations or something. 1,000 for the number crunching, and 1 for something to do while the other were running.

    Too bad I didn’t think about the corporate sponsorship angle!

  • H of L

    Yes, yes, but do the sims wield anything resembling a true stainless-steel broad-sword?

    Actually, anything to overcome a “Cover 2” defense would be pretty high in my list … or anything Appalachian State used in their defeat of Michigan, for that matter …

  • Quasar9

    lol Julianne, are you saying those guys and gals in the hollywood movies about adventures on board the USS Enterprise starships were walking around playing with their PS3s. And I thought it was all science fiction or play acting and make believe.

    So What’s the name of the game they are trying to ban or have banned in the uk?

  • Luis Sanchez

    Now, I am wondering… what will happen when the simulations are ended? I have come to the inevitable conclusion that the researchers are going to end making some really lively gaming tournaments.

    So seems quite a cool deal, you get your cluster and a bunch of gaming machines for playing Resistance: Fall of men while you are supposed to work. The next level on procastination, I suppose.

  • Ian Paul Freeley

    I can just see this progressing to a simulation with a bunch of networked Nintendo Wii. It runs pretty fast, but requires a line of grad students all waving the controllers around to generate the gravity waves…

    There could be a nice thesis project on what happens when you include time-dilation in Mario Cart.

  • Pablo

    @ Ian

    LOL. Men, If just I had had this wonderful idea before I started my stupid PhD thesis about such an useless thing as autonomic computing . . .

  • Abelian

    I always knew all those night spent playing games were not in vain…

  • Steinn Sigurdsson

    Graphic Processing Unit science is a big deal this year
    workshop at the IAS on it next month.

  • Tyler DiPietro

    I’m glad that someone else mentioned GPGPU. Apparently the primary targets of the new HPC products (e.g., NVidia’s “Tesla” line) are those involved with scientific computing.

    Using PS3’s is pretty makeshift though. The CELL may have a high theoretical peak performance (especially with flops), but you have 7 asymmetric coprocessors with all kinds of different address space. Optimizing the code is a bitch, refactoring code from other platforms is a nightmare. Can’t imagine it would be ideal as a scientific computing solution.

  • Jason Dick

    Yeah, with either a CELL or a GPU, pretty much the only way you’re going to get good utilization is if you have a small number of algorithms that take up all of your processing time (e.g. FFT’s), and can spend some time optimizing those few algorithms for maximal use of the parallel processing elements.

    So, for scientific apps, the ideal situation would be to have a CELL or GPU-optimized library where you just call a few commands to make use of the CELL or GPU as an external processing device. As a result, I think that GPU’s are a bit more usable in this regard, as the PPE is a bit underpowered compared to desktop processors. The difficulty, though, is that modern GPU’s don’t even support double-precision floating point, though nVidia has promised support with hardware released late this year (this was as of last year).

  • Tyler DiPietro

    “pretty much the only way you’re going to get good utilization is if you have a small number of algorithms that take up all of your processing time (e.g. FFT’s), and can spend some time optimizing those few algorithms for maximal use of the parallel processing elements.”

    Exactly. The reason why GPU’s became a massively SIMD arch in the first place was because graphics has so many inherently parallel processes (e.g., ray tracing, scan conversion, voxel based computations, etc.). Similar processes in scientific computing would make GPU’s very useful (as soon as they natively support double precision flops).

    But there is another limitation I can think of: BUS speed. This is a huge bottleneck in games, and I’d imagine sharing info between the CPU and the GPU would be even worse for scientific apps when considering the huge amount of data that has to scheduled and transfered.

  • Jason Dick

    But there is another limitation I can think of: BUS speed. This is a huge bottleneck in games, and I’d imagine sharing info between the CPU and the GPU would be even worse for scientific apps when considering the huge amount of data that has to scheduled and transfered.

    Well, that’s just a function of how much processing you need to do on the data between bus transfers, such that any time your operation scaling with data size is greater than linear it’s likely to be a benefit over CPU processing.

  • Neil B.

    It’s great how common “personal” computers and game players got powerful enough to do good scientific computing – they’re up with the supercomputers of old. Well, a bigger computer should still be even better, so what’s out there good in various ranges for those purposes?

  • Astronomy-Page

    I own a Web 2.0 astronomy oriented website with webapps and stuff and I wanted to ask whether or not you would care for a link exchange. I browsed you links section, and I read the message telling us to contact you, but I couldn’t find any email anywhere, so sorry for posting the comment here. My website is
    Astronomy-Page: Astronomy 2.0
    Please reply by email or with another comment.

  • Chris H.

    I thought I would point out the following
    paper :
    “Lattice QCD as a video game.” As I understand it, cost
    is an important factor here. The video game makers sell
    their game boxes at a loss and make up the money selling software.

  • Rob MacRiner

    Rob MacRiner ,

    Answer to Question: Why does time seem to exist only in a forward direction?

    Time seems to only exist in a forward direction because the universe is expanding. If the Universe reaches Critical Velocity and starts to contract ….then time, as we measure time will reverse according to the Big Bang / Big Crunch Theory. The reason for this is that time does not exist without change or movement….. (change or movement of particle matter or energy as we know it). If matter has no movement either expanding or contracting then time does not exit for that matter. However Time can exist around non moving particle matter if something is either expanding or contracting around it.

    If the expansion of matter increases as in the case of our universe, or an expanding object, or even light…then time increases relative to the rate of expansion. Example: if carbon A is heated and expands faster than carbon B (which is not heated) then time increases in carbon A relative to carbon B…However as Einstein pointed out…time is relative to the observer…and you need something of contrast to make that comparison….fortunately our universe offers lots of contrast …otherwise we would have a very difficult time figuring this out. Time being relative to the observer can exist at different speeds based on the rate of expanding matter. If you are on riding on a beam of light than time is much different than your friend riding on a sound wave. Of course time is relative to the observer, therefore your time is much faster only to him, or any body else who is not on a beam of light.

    If matter contracts or condenses then time actually reverses…as in the case of a contracting universe…so Planks Quantum would be measured as zero time for the entire Universe…and time starts at the point of the Big Bang (once matter is on the move again)… In the case of a Black Hole, relative to our expanding universe)… there is also no time. (except for matter being sucked into a Black Hole….this matter would be reversing in time, until at which point it becomes part of the Black Hole mass, then time (in a Black Hole) as in Planks Quantum is zero….which is odd because the Universe is still expanding around the Black Hole…but it is consistent with the theory that. Time can exist around “non moving matter” if something is either expanding or contracting

    Time as we know it is measured in a forward direction and will continue until the point of critical velocity…at which point time starts to reverse…and for a brief moment…the point where the Universe changes from expanding to contracting…time will again be zero…as in Planks Quantum. However…during the forward direction of time…(while the Universe is expanding)…Black Holes are continuing to suck up matter…and should in theory at some point converge with other Black Holes….Therefore…as the universe is expanding from the Big Bang…there is multitude of matter which is not expanding (Black Holes)…which might well be unexploded Planks Quantum matter from the Big Bang…and the Black Holes with their massive gravitational force are sucking up matter which was attempting to expand but was not able to overcome the stronger force of the Black Hole…like mini-Plank Quantum’s converging within the universe …When the Universe reaches Critical Velocity and then all matter in our Universe starts to contract…heading towards the Big Crunch….the multitude of Black Holes converging (up to that point) should in theory rapidly increase the speed of reverse time …acting as an accelerant force of a contracting Universe with their collective gravitational force …So the reverse of time.(the journey the contracting Universe is taking towards the Big Crunch)…should happen much quicker than the time it took for the Universe to go from the Big Bang to Critical Velocity…That is of course Time relative from the Big Bang to Critical Velocity ……in contrast to …….Time Relative from Critical Velocity to the Big Crunch. Mathematically it might be possible to estimate this time…. Rob MacRiner Nov 2007


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

See More

Collapse bottom bar