1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

News Google unveils Android 5.0 Lollipop

Discussion in 'Article Discussion' started by Gareth Halfacree, 16 Oct 2014.

  1. Gareth Halfacree

    Gareth Halfacree WIIGII! Staff Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    12,153
    Likes Received:
    1,674
  2. GuilleAcoustic

    GuilleAcoustic Ook ? Ook !

    Joined:
    26 Nov 2010
    Posts:
    3,277
    Likes Received:
    71
    Nice news. I so want a Tegra K1 SBC smaller than the Jetson board, something around the pico itx at most.
     
  3. Gareth Halfacree

    Gareth Halfacree WIIGII! Staff Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    12,153
    Likes Received:
    1,674
    I basically just want Nvidia to stop playing silly beggars and unlock OpenCL on the Jetson TK1. The chip can do OpenCL, but the Jetson will only do CUDA - 'cos, obviously, Nvidia wants to lock you in to its ecosystem. It makes comparisons to other boards very difficult, too: I've got an Adapteva Parallella that can run (a sub-set of) OpenCL and the Jetson that can run CUDA, and no way of accurately comparing the two. Bah!
     
  4. GuilleAcoustic

    GuilleAcoustic Ook ? Ook !

    Joined:
    26 Nov 2010
    Posts:
    3,277
    Likes Received:
    71
    Yes, that a point. Been tempted by a parallela, how does it hold itself ? I need an AIO solution, as small as possible, Linux friendly and capable to display a desktop environment (I'd be happy with openBox / XFCE).

    OpenCL would be a huge bonus as I could off-load calculation for stuff like zip/unzip. OpenGL, non-ES, would be appreciated but not mandatory.

    The best I found, at the moment, would be i.MX6 based (quad core version). Maybe I could interface with an FPGA or maybe someone know of a ARM + FPGA all-in-one solution.
     
  5. leexgx

    leexgx CPC hang out zone (i Fix pcs i do )

    Joined:
    28 Jun 2006
    Posts:
    1,346
    Likes Received:
    8
    not that i know a lot about this

    OpenCL for it to work on Nvidia they do OpenCL to CUDA to make it work (but its slower then using CUDA directly and requires Tweaks to make OpenCL it work more optimally under Nvidia OpenCL)

    most just go native CUDA on Nvidia and OpenCL on ATI (but still nice to have even if its not as speedy)
     
  6. Gareth Halfacree

    Gareth Halfacree WIIGII! Staff Administrator Super Moderator Moderator

    Joined:
    4 Dec 2007
    Posts:
    12,153
    Likes Received:
    1,674
    I'll be giving it a proper run-through next week, but it seems good so far: an OpenCL-enabled version of popular password cracker John The Ripper gets 1,200K/s bcrypt32, compared to 800 (single-threaded) on my AMD A10-5800K APU's CPU core. Bare in mind that OpenCL for the Parallella (or, rather, the Epiphany) isn't really OpenCL; it's a subset of OpenCL features and not formally certified. Don't expect OpenCL stuff to just work out-the-box on it, in other words.

    In the next week or so I'm also going to be taking delivery of a Cubieboard 4, which may be of interest: eight-core (four high-speed high-power Cortex-A15s, four low-speed low-power Cortex-A7s in an ARM big.LITTLE implementation) with decently powerful graphics supporting OpenGL ES and OpenCL. No full-fat OpenGL, though.
     
  7. GuilleAcoustic

    GuilleAcoustic Ook ? Ook !

    Joined:
    26 Nov 2010
    Posts:
    3,277
    Likes Received:
    71
    That is very nice. I was excpeting not out-of-the-box openCL, hopefully that will get me out of my x86 boredom lol. Anyway, that won't be a main rig, only a fun dev rig for a bored developper (me).

    The big.LITTLE looks nice too. It's a good way to reduce the power consumption.
     

Share This Page