Results 1 to 8 of 8

Thread: Another Open-Source OpenCL Project Pops Up

  1. #1
    Join Date
    Jan 2007
    Posts
    14,327

    Default Another Open-Source OpenCL Project Pops Up

    Phoronix: Another Open-Source OpenCL Project Pops Up

    This week has been busy with OpenCL news with the release of Portable OpenCL and libclc, but now there's been another project brought up and that's a German university research project to create a high-performance OpenCL driver for the CPU...

    http://www.phoronix.com/vr.php?view=MTAwMzQ

  2. #2
    Join Date
    Jan 2009
    Posts
    1,591

    Default

    i would be glad if they had put all the time and efford in making Clover (btw denis seems to have resumed his work on it) work on HW


  3. #3
    Join Date
    Nov 2008
    Location
    Old Europe
    Posts
    904

    Default

    Denis seems to be a quite gifted person.
    If I'm right, he's just 18/19 years old.
    Some of the comments in his blog, from well-known developers, clearly
    stated that they are really impressed by what he accomplished during his GSoC project.

  4. #4
    Join Date
    Feb 2008
    Posts
    205

    Default

    Quote Originally Posted by entropy View Post
    Denis seems to be a quite gifted person.
    If I'm right, he's just 18/19 years old.
    Some of the comments in his blog, from well-known developers, clearly
    stated that they are really impressed by what he accomplished during his GSoC project.
    There's a big bunch of smart young boys in the FOSS community, it's a great opportunity for them to show their real skills in a meritocratic way and not get discriminated against age any other reason. It's the benefit of the hacker ethic

  5. #5
    Join Date
    Sep 2009
    Posts
    313

    Default What is the point?

    I understand the possibility of code commonality but beyond that why OpenCL on the CPU? If you are going to run on the CPU why not simply use your normal C/C++ code & development tools?

  6. #6
    Join Date
    Oct 2007
    Location
    Toronto-ish
    Posts
    7,385

    Default

    OpenCL gives you a relatively easy way to spread work across a lot of CPU cores, in addition to the code commonality you mentioned earlier.

  7. #7
    Join Date
    Mar 2008
    Location
    Milan, Italy
    Posts
    103

    Default Any benefit for end users?

    Even if OpenCL would be fully implemented in a way or another, how many applications are ready to use it?
    I mean, for the general desktop.
    AFAIK Mac OSX has been sporting it for 1 year already, in a official way and in a well integrated environment. Is there any major application taking advantage of it?
    I'm just asking out of curiosity, not criticizing. I admit I know little or nothing about it.

  8. #8
    Join Date
    Sep 2009
    Posts
    313

    Default OpenCL is used extensively.

    You have good questions, it is good to be curios.

    Quote Originally Posted by TeoLinuX View Post
    Even if OpenCL would be fully implemented in a way or another, how many applications are ready to use it?
    A loaded question! People need to understand that some (many?) apps will never benefit from OpenCL. GPUs are only capable of acceleratining certain types of calculations. However there is a catch here.

    That catch is that Apple uses OpenCL deep inside of Core Image (I believe, could have been quartz) so in a sense all apps benefit from GPU acceleration. Maybe I should say the system benefits as the GPU effectively reduces CPU load.
    I mean, for the general desktop.
    AFAIK Mac OSX has been sporting it for 1 year already, in a official way and in a well integrated environment. Is there any major application taking advantage of it?
    Certainly there are apps taking advantage of the capability. However you need to understand what types of apps those are. They are not run of the mill desktop apps.
    I'm just asking out of curiosity, not criticizing. I admit I know little or nothing about it.
    That is not a problem. There is extensive mis understandings out there with respect to OpenCL and GPU computing in general. The nature of the GPU beast (right now) is that it is only viable for acceleration of highly parallel calculations. The ability go the GPU to accelerate common desktop apps is extremely limited.

    So you need apps that have or can be arraigned to have data sets that can be processed in parallel before GPU computing becomes of value. Given that OpenCL is perhaps one of Apples biggest success stories with respect to software initiatives. The adoption is wide spread on any number of platforms.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •