Announcement

Collapse
No announcement yet.

AMD Linux Graphics: The Latest Open-Source RadeonSI Driver Moves On To Smacking Catalyst

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

  • #31
    I still remember the times when Michael didn't miss any opportunity to point out (=bash) how horrible and useless the open-source drivers are.

    Comment


    • #32
      Originally posted by Linuxhippy View Post
      I still remember the times when Michael didn't miss any opportunity to point out (=bash) how horrible and useless the open-source drivers are.
      Maybe that motivated opensource driver developers to improve them....

      Comment


      • #33
        Originally posted by dungeon View Post



        Or maybe this is all about that some things behave weird on Skylake?
        Yeah, I wonder what's up with that. There has to be some explanation that makes sense. Some cause that can be described. It doesn't seem like Skylake or Broadwell are stable.
        Last edited by duby229; 01 September 2015, 06:21 PM.

        Comment


        • #34
          If that work and performance keep improving I think my idea of ditching AMD will go away. A 4 GB R9 290 may look as a nice upgrade over my 2GB 7870

          Comment


          • #35
            Originally posted by marek View Post
            whitecat: I'm sure Valley doesn't use tessellation, but I don't know if it uses a compute shader. The log only tells us what the engine "can" use, not what the engine is using.
            Well that's not too hard is it? Someone just needs to run it through apitrace on fglrx, and it should reveal exactly what the game is doing.

            Comment


            • #36
              Originally posted by Ancurio View Post

              Well that's not too hard is it? Someone just needs to run it through apitrace on fglrx, and it should reveal exactly what the game is doing.
              There is no difference, beside that usual midshader dissalowed.

              Better if someone can made video with render comparision with fglrx and radeon, and why not even nvidia... because if this is true that R9 290 with opensource driver matches GTX 980 with nvidia driver, this should be news which does not happen every day - maybe that will be first time we smashed that green blob

              But yeah it is very hard to believe without further approval

              Comment


              • #37
                Can anyone post a screenshot showing the difference?

                I guess it's probably compute shaders, since it appears to be checking for that extension.

                Comment


                • #38
                  Originally posted by smitty3268 View Post
                  Can anyone post a screenshot showing the difference?
                  Ask Michael i can't reproduce this with current git mesa/llvm even with kernel from linus's tree... let alone that Catalyst also works fine for me on Kabini and Bonaire with kernels 4.2, 4.1, 4.0, 3.16... whatever i tried it works, and that is on Debian Sid which should currently be more broken, but it isn't
                  Last edited by dungeon; 01 September 2015, 11:24 PM.

                  Comment


                  • #39
                    The other thing to keep in mind with these benchmarks is that I'm pretty sure the newer fglrx driver came with a bunch of performance boosts compared to this older version. But it's fantastic to see Mesa running about on par with it.

                    Comment


                    • #40
                      Originally posted by dungeon View Post
                      Ask Michael i can't reproduce this with current git mesa/llvm even with kernel from linus's tree... let alone that Catalyst also works fine for me on Kabini and Bonaire with kernels 4.2, 4.1, 4.0, 3.16... whatever i tried it works, and that is on Debian Sid which should currently be more broken, but it isn't
                      Did you pick up a newer .drirc when you updated mesa ?

                      Apparently the app settings for Unigine * and others are different starting with Mesa 11, looks like the main change is taking away the hacks that were added for 3.0 Unigine demos.
                      Last edited by bridgman; 02 September 2015, 12:15 AM.
                      Test signature

                      Comment

                      Working...
                      X