No announcement yet.

LLVM Back-End For Gallium3D Almost There

  • Filter
  • Time
  • Show
Clear All
new posts

  • LLVM Back-End For Gallium3D Almost There

    Phoronix: LLVM Back-End For Gallium3D Almost There

    While Tungsten's Gallium3D architecture is still a ways out from being used by most open-source graphics drivers and then being picked up by end-users, it continues to pickup new technical features. Corbin Simpson and Stephane Marchesin that work on the Radeon and Nouveau projects, respectively, have been working to building LLVM back-ends for Gallium3D...

  • #2
    OK, not exactly "almost there" but almost.


    • #3
      Thats good, then GLSL should work.


      • #4
        "it now builds, but it ends with a segmentation fault."

        lol.. good luck to them and debugging what's causing that segmentation fault


        • #5
          segfault implied the CPU is simple executing garbage...


          • #6
            Originally posted by FunkyRider View Post
            segfault implied the CPU is simple executing garbage...
            mmm... you should probably learn some things before talking... segmentation faults is way more complicated then "executing garbage". As always wikipedia is a good place to learn things you don't know:


            also, if it was so easy to fix they would have done it immediately..don't you think?

            Last edited by bulletxt; 12-28-2008, 10:05 PM.


            • #7
              Wow. Maybe wait until there's something useful to report before reporting? "Almost there"

              Come on now. I know there isn't much going on, but the article is grasping at strays.


              • #8
                Mac OS X uses LLVM for graphical stuff, so there's must be some reason.

                I hope Linux gets more advanced in the graphical stuff and can compete in this arena with both Mac OS X and Microsoft platforms in performance and support some day...
                Last edited by timofonic; 12-29-2008, 01:23 AM.


                • #9
                  A compiler backend is not a trivial piece of code, and getting it integrated isn't trivial either. I expect the reason for the segmentation fault will be tracked down and fixed soon (likely revealing several more bugs, but that's always the way it is with new code).


                  • #10

                    No, it's not even that.

                    An article was written about a piece of code that is 'barely started' as if it were 'Almost there'

                    See more of the IRC log:

                    <EruditeHermit> MostAwesomeDude: looks like you made the rounds already
                    <MostAwesomeDude> EruditeHermit: Aw, dammit, and for something that's only barely started.