Announcement

Collapse
No announcement yet.

NVIDIA Publishes Vendor-Neutral GL Dispatch Library

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by liam View Post
    I didn't know bumblebee was still around.
    What is the point of using that when we have the shared buffer mechanism that airlied created?
    DMA-BUF needs DMA-Sync (not yet mainlined AFAIK) to function PROPERLY. If you don't have DMA-Sync then everything gets throttled and synced by the CPU. Also not everything is running a modern enough kernel / drivers to have access to DMA-*, ALSO there's no window hint yet to say "RUN THIS APP ON THE DEDICATED CARD" so the optirun utility from Bumblee is useful.
    All opinions are my own not those of my employer if you know who they are.

    Comment


    • #12
      Originally posted by johnc View Post
      I don't think NVIDIA has claimed to be working on EGL at all.


      Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

      Comment


      • #13
        I stand corrected.

        Comment


        • #14
          Originally posted by johnc View Post
          I don't think NVIDIA has claimed to be working on EGL at all.
          They have to be, though, at least with respect to their mobile drivers. This will be increasingly the case as they converge upon a single architecture (perhaps as early as T5).

          Comment


          • #15
            Originally posted by scionicspectre View Post
            I was also wondering if this has anything to do with a future method to do this. It sounds related. Of course, if it does, I would expect Michael to be a bit more explicit about it.
            Explicitly: No. At least not yet.
            Read the README file: https://github.com/NVIDIA/libglvnd/b...ster/README.md

            It's pretty clear that NVidia concentrates on GLX but Mir and Wayland are based on EGL. So Mir and Wayland support may be byproducts of this but NVidia's development focus for this clearly is X11 ? no surprise there as the Linux distributions that actually matter to NVidia (RHEL etc.) will stay with X11 for quite some time.

            Comment


            • #16
              "Bryan Nguyn"

              Two misspellings in a single name.

              Comment


              • #17
                Mesa

                I was under the impression that Mesa already shipped a library that does exactly this as libgl: where am I mistaken? What does Mesa actually do instead?

                EDIT: Looking at the readme they say "which is really just a thin wrapper around Mesa's glapi", so yes, it does use Mesa's lib for this.

                Comment


                • #18
                  Originally posted by Awesomeness View Post
                  Explicitly: No. At least not yet.
                  Read the README file: https://github.com/NVIDIA/libglvnd/b...ster/README.md

                  It's pretty clear that NVidia concentrates on GLX but Mir and Wayland are based on EGL. So Mir and Wayland support may be byproducts of this but NVidia's development focus for this clearly is X11 ? no surprise there as the Linux distributions that actually matter to NVidia (RHEL etc.) will stay with X11 for quite some time.
                  Yes it was developed for Xorg as intel said a long time ago they wanted to remove GLX from Xorg

                  Comment


                  • #19
                    You would think having a good way to work with multiple drivers that implement the same api.
                    You would think being able to use api's conveniently would be a higher priority.

                    Comment


                    • #20
                      this means ATI (catalyst) + Nvidia in same machine? (hybrid like windows)

                      Comment

                      Working...
                      X