Announcement

Collapse
No announcement yet.

More Vega & Raven Ridge AMDGPU Fixes Set For Linux 4.13

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

  • #11
    Originally posted by wizard69 View Post
    Raven Ridge dounds very interesting. It will be intereeting to see how Vega works as an integrated GPU.
    I read somewhere that the expected performance is of a RX 560, which is awesome for a iGPU

    Comment


    • #12
      Originally posted by Soul_keeper View Post
      what is the proper method for cloning this tree ?
      When you say "this tree" do you mean the 4.13 drm-next tree that is the subject of the article, or the 4.11 staging tree which was the subject of a few posts just before yours ?

      If the former, blisbell's response looks correct; if the latter then substitute replace drm-next-4.13 with amd-staging-4.11.

      If you have already cloned the repo then you just need to check out the appropriate branch (one of the two above) - the code in "master" branch is what you see by default but IIRC that is really old.
      Last edited by bridgman; 06 July 2017, 01:59 AM.
      Test signature

      Comment


      • #13
        I'm excited about Vega... but I would like to see the DC code land upstream soon after the release of Vega. Actually I'm running a normal Arch Linux kernel with mesa-git for my RX 480 and together with RADV I'm very happy and thankful about all the latest improvements in the open source graphic stack. But personally I don't want to loose this comfort and starting to compile for myself the complete stack of a AMD staging kernel, mesa-git, libdrm, llvm-git, plus additional packages and maybe additional patches, so I hope there will be at least a rough estimation about a timeline from AMD's perspective. At the moment there is too much speculation about when we see which piece to land upstream... sure we never know if patches are accepted, but this can't be avoided :-)

        Comment


        • #14
          Originally posted by bridgman View Post
          If you have already cloned the repo then you just need to check out the appropriate branch (one of the two above) - the code in "master" branch is what you see by default but IIRC that is really old.
          Why don't you always sync master branch with the latest stable code ? this way.. you would leverage the fact that the default cloned branch is the master branch, and when someone clones that repo, they actually get the latest stable release. (in this case I guess stable means the amd-staging-* one)

          Comment


          • #15
            Originally posted by xxmitsu View Post
            Why don't you always sync master branch with the latest stable code ?
            Because that is not stable code, but next and staging code

            Originally posted by xxmitsu View Post
            ... when someone clones that repo, they actually get the latest stable release.
            Yeah, you will get latest stable release there thats for sure You might but also might not get anything stable there, these are for early public testing.
            Last edited by dungeon; 06 July 2017, 05:30 AM.

            Comment


            • #16
              Originally posted by debianxfce View Post

              Use -b drm-next-4.14-wip. drm-next-4.13 is old.
              https://cgit.freedesktop.org/~agd5f/...-next-4.14-wip
              That 4.14 is very old, drm-next-4.15 is newer but private, just sign NDA debianxfce to get access You can't even imagine how stable that is - Big Stable
              Last edited by dungeon; 06 July 2017, 05:51 AM.

              Comment


              • #17
                Originally posted by dungeon View Post

                Because that is not stable code, but next and staging code
                Originally posted by xxmitsu
                Why don't you always sync master branch with the latest stable code ? .... (in this case I guess stable means the amd-staging-* one)
                I was referring to sync master branch with latest amd-staging-* branch.

                From what I understood, this is the branch considered stable, that is even ran through internal QA.
                If some a fix reaches -next first, it will backported to this branch. I think that code that reaches here, as result, is pretty stable. So in my opinion this would be more recommended for normal user.
                Last edited by xxmitsu; 06 July 2017, 08:24 AM.

                Comment


                • #18
                  I believe the change flow is usually the other way round - commits go to staging first (after review) and then are cherry-picked to drm-next or drm-fixes. The active staging branch also feeds into branches for ROCm and the hybrid AMDGPU-PRO kernel drivers.

                  There are enough developers working on the branch that breakage will usually get noticed and fixed quickly but IMO it should be viewed as a fast-moving development branch with a lot of eyes on it rather than, say, the Linux kernel stable branches which only get bug & security fixes.
                  Last edited by bridgman; 06 July 2017, 10:32 AM.
                  Test signature

                  Comment


                  • #19
                    How is full Sea Islands based GPU support coming along ? I know that Kaveri and Carrizo APUs have Oland based SI iGPUs. However...even though Oland based as well... my Bristol Ridge APU has what's classified as a Volcanic Island ( VI ) based iGPU.

                    Comment


                    • #20
                      Originally posted by Jumbotron View Post
                      How is full Sea Islands based GPU support coming along ? I know that Kaveri and Carrizo APUs have Oland based SI iGPUs. However...even though Oland based as well... my Bristol Ridge APU has what's classified as a Volcanic Island ( VI ) based iGPU.
                      Kaveri is CI (gfx7) based. Carrizo/Bristol are VI (gfx8) based. Oland is an SI (gfx6) based sGPU.

                      Comment

                      Working...
                      X