Personal View site logo
Finally Launched: the Adapteva so called "supercomputer"
  • 43 Replies sorted by
  • Vitaliy_KiselevVitaliy_Kiselev

    @endotoxic

    Again, you are not right. And I am not stupid, so it is bad idea to make school course here.

    It has two A9 cores. But all other is custom thing totally non related to ARM.

  • endotoxicendotoxic

    This is the only company i know was able to implment so so.. the ARM to big company...failed cos i remember in 2004.. 2003 they wanted to make a complete hardware via chip. Northbridge, southbridge, video, net, cache, and periphericals inside processors, so only mainboard was used to put sokets, since all was run out from prosessor. Like the solution AMD is doig now almost 10 years later.

    Funny thing about tech is it must happen in the right time for the proper implementation, and also PEOPLE must be aware of it and use it, if not its a failure.

    I have lots of examples from this kind of projects and they usually fail.

    Good old PASEMI.

    http://en.wikipedia.org/wiki/P.A._Semi

  • kavadnikavadni

    @Vitaliy_Kiselev

    ARM A9 here is external CPU, and this "accelerator" is just connected to it

    Yes, exactly. I think I was confused by your sentence "It is not ARM"

  • endotoxicendotoxic

    Maybe the proper thing is too name it diferently. But its not my project. :D Make shure they dont call it ARM please, since ARM is based on RISC.

    RISC is open standard.

  • Vitaliy_KiselevVitaliy_Kiselev

    RISC is open standard.

    RISC is not standard at all. It is architecture.

  • endotoxicendotoxic

    Sorry its arquitecture that is right.

  • kavadnikavadni

    Hah, I am doing some reading and learning more about GPU vs CPU floating point precision than I really care to know. This things going to be a lot of fun .. however it turns out .. and teensy

  • kavadnikavadni

    Here is their answer to the 'So Called SuperComputer' issue

    Why do you call the Parallella a supercomputer?

    The Parallella project is not a board, it's intended to be a long term computing project and community dedicated to advancing parallel computing. The current $99 board aren't considered supercomputers by 2012 standards, but a cluster of 10 Parallella boards would have been considered a supercomputer 10 years ago. Our goal is to put a bona-fida supercomputer in the hands of everyone as soon as possible but the first Parallella board is just the first step. Once we have a strong community in place, work will being on PCIe boards containing multiple 1024-core chips with 2048 GFLOPS of double precision performance per chip. At that point, there should be no question that the Parallella would qualify as a true supercomputing platform.

    They probably should put that right up the top

  • kavadnikavadni

    Yay, looks like the luxrender boys are onboard .. comment from David Bucciarelli on the kickstarter page.

    --
    I'm one of the developer at (http://www.luxrender.net) and I'm very interested.

    We have also developed full OpenCL applications like SmallLuxGPU (http://www.youtube.com/watch…), LuxMark (http://www.luxrender.net/wiki/LuxMark), Sfera (http://www.youtube.com/watch…). They could run on Parallella out of the box thanks to the OpenCL support.

  • Vitaliy_KiselevVitaliy_Kiselev

    They could run on Parallella out of the box thanks to the OpenCL support.

    I have big troubles with statement. As no one ever tested this "OpenCL compatibility". And all architecture is very bad suited for rendering (that requires acess to quite big amount of data).

  • kavadnikavadni

    This is all they have so far:
    http://www.adapteva.com/support/examples/opencl-examples/efficient-matrix-multiplication-using-opencl/

    http://www.eetimes.com/electronics-news/4394513/Adapteva-ships-OpenCL-compiler-for-Epiphany

  • goannagoanna

    As of today their Parallella KickStarter program has 1,720 Backers $231,316 pledged of $750,000 goal and 19 days to go.

  • zcreamzcream

    Makes me want to set up a project on kickstarter. Discover the Unified Field Theory. Or commercialize fusion based power. Start off with a few million. Hmmm!

  • goannagoanna

    @zcream

    I look forward to helping your project get off the ground!

    Design and Technology projects that are developing new hardware or products must show on their project pages a functional prototype — meaning a prototype that currently does the things a creator says it can do — and detailed information about their experience. Kickstarter Guidelines

    image

    from http://www.squidoo.com/backyard-space-projects

  • goannagoanna

    An anti-scamming Police spokesman warned on Aussie TV last week that the most common victim of an investment scam is of a certain age, has made a few wise investments, but is under the illusion that he/she can investigate the viability of a company or investment proposition by doing a few internet searches.

    He went on to explain that it is easy for perpetrators to seed false, web-based material to passe off as credentials.

    Regarding Adapteva's credentials: When I'm doing a [paid] story I can usually substantiate or disprove the kind of vague claims we're reading here within three or four phone calls. I'd phone the head of that faculty at MIT which Adapteva says they're cooperating with. I might get a qualitative reply, referring specifically to the work they're doing - or I might find that this "cooperation" was just a free Power Point demo attended by a handful of students.

    In the former case, it would be like reading the CV of a competent job seeker who just doesn't know how to sell himself or realise that future employers will have to phone all his referees. In the latter case, you're dealing with a candidate who's just as naive but untrustworthy as well.

    Adapteva could be on the level. Or it could be a classic sting. I honestly cannot tell from web searching. And I should be able to. And that - alone, is cause to withdraw.

    Conclusion: Get more info. What is published is not enough for you to invest. If a company keeps using excuses such as "commercial confidentiality" for not showing their cards on the table, the wise position to take is that of disbelief.

    Adapteva, my cheque's in the mail. ;-)

  • kavadnikavadni

    @goanna .. interestingly they have released their documentation.

    I didn't see where they say they are cooperating with MIT, can you provide a link?

  • goannagoanna

    @kavadni

    Hope I got the Uni right. Would have to check browser history. On the road for 3 days with just my phone.

  • goannagoanna

    from CNet, April 15, 2013.

    Summary: Move over Raspberry Pi, here comes Adapteva's Parallella, a low-cost parallel chip board for Linux supercomputing.

    see http://www.zdnet.com/parallella-the-99-linux-supercomputer-7000014036/

Add a Comment