Announcement

Collapse
No announcement yet.

Trying R600g On Mesa 11.3, Cayman GPU On Linux 4.7 Radeon DRM Yields Problems

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

  • #11
    Originally posted by Mez' View Post
    It's less than 5 years old hardware, that for the most part does have OpenGL 4.3 compliancy. It is omnipresent in 4 years old laptops of that period. Not everyone buys a new laptop every two years regardless of the generated waste. Plus, these cards still have enough in the bag to play games with an acceptable comfort. And above everything, people invested their money in these laptops, and they have the right to get some return out of it.
    Hence, whether you want it or not, this has to be worked on, independently of the support for newer hardware.
    Sure you are right, but is the hardware good to play new demanding games, well not my Apu, I can play well well directx 9 games, but not directx 11 games, therefore for my hardware I think OpenGL 3.3 is all I need. Yes currently it's support 4.1, but tesselation is too much. But it might not be true for the owners of the HD 69x0.

    Comment


    • #12
      The one-line-of-code to fake fp64 support isn't too horrible an idea for now, especially if combined with something like a driconf option to advertise fake fp64 support for games which require a GL 4.x context but which have been confirmed to not actually use fp64 capabilities. An interested person could probably whip up that patch fairly quickly using the patterns established by existing driconf options.

      Comment


      • #13
        OK, I finally got Manjaro Gnome to boot (at the cost of Ubuntu going from 5 seconds to 2 minutes to reach lightdm ), and thus to try r600g and DRI_PRIME on a different system.

        I have to say it runs definitely smoother than on Ubuntu. I can testify an increase in fps and smoothness of 20 to 30%. Anybody has already noted such a difference? Is it linked to Compiz >< Mutter?

        Comment


        • #14
          The improvements in synthetic tests are due to asynchronous page flip support.

          As for the regression in other tests, somebody with an affected system please bisect and file a bug report with the result.

          Comment

          Working...
          X