Results 1 to 9 of 9

Thread: OpenCL, GLSL Back-End For LLVM May Soon Open Up

Hybrid View

  1. #1
    Join Date
    Jan 2007
    Posts
    14,526

    Default OpenCL, GLSL Back-End For LLVM May Soon Open Up

    Phoronix: OpenCL, GLSL Back-End For LLVM May Soon Open Up

    A university student that successfully wrote OpenCL and GLSL back-ends to the Low Level Virtual Machine (LLVM) is arranging to have the code open-sourced if there is interest, which already LLVM developers are requesting...

    http://www.phoronix.com/vr.php?view=OTg0MQ

  2. #2
    Join Date
    Aug 2009
    Location
    Russe, Bulgaria
    Posts
    503

    Default

    This may fitin nicely in http://www.lunarglass.org/.

  3. #3
    Join Date
    Oct 2008
    Posts
    3,070

    Default While this is interesting, i'm not sure it has any bearing on Mesa

    Mesa is trying to compile GLSL and OpenCL into a form the hardware understands, while this is doing the opposite.

    IIUC, it will allow you to write some code in C and have LLVM transform that into an OpenCL program.

  4. #4
    Join Date
    Aug 2009
    Location
    Russe, Bulgaria
    Posts
    503

    Default

    Quote Originally Posted by smitty3268 View Post
    Mesa is trying to compile GLSL and OpenCL into a form the hardware understands, while this is doing the opposite.

    IIUC, it will allow you to write some code in C and have LLVM transform that into an OpenCL program.
    I think the purpose is to use LLVM optimizers.
    I just hope TGSI will be replaced with enhanced LLVM bitcode. One that supports graphics semantics, and Gallium drives will translate LLVM->hw. For more information...the site I gave above.

  5. #5
    Join Date
    Nov 2008
    Location
    Germany
    Posts
    5,411

    Default

    Quote Originally Posted by Drago View Post
    I think the purpose is to use LLVM optimizers.
    I just hope TGSI will be replaced with enhanced LLVM bitcode. One that supports graphics semantics, and Gallium drives will translate LLVM->hw. For more information...the site I gave above.
    the LLVM stuff is for software rendering only because the LLVM can not handle a VLIW gpu architecture.

    means you can build a software pipe with that but not a gpu acceleration.

  6. #6
    Join Date
    Aug 2009
    Location
    Russe, Bulgaria
    Posts
    503

    Default

    Quote Originally Posted by Qaridarium View Post
    the LLVM stuff is for software rendering only because the LLVM can not handle a VLIW gpu architecture.

    means you can build a software pipe with that but not a gpu acceleration.
    What's the problem with LLVM and VLIW? Can someone of the devs explain briefly? Is that graphics compute load, is so much more different than non-graphics one?
    Somewhere I read about differnt memory model for graphics, than non-graphics, but didn't understant it quite well.

Posting Permissions

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