Announcement

Collapse
No announcement yet.

AMDGPU DC Pull Request Submitted For Linux 4.15: Finally The New Display Stack

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

  • #51
    Originally posted by Klassic Six View Post
    Everything came with AMD GPUs are experimental, first powerplay (amdgpu.powerplay=1) and now this (amdgpu.dc=1). Why they can't publish a code that works fine at the first day?? although I'm happy that finally DC become acceptable.
    Because it's not tested enough. Reading and understanding seems hard these days....

    Comment


    • #52
      Originally posted by airlied View Post
      so should I merge it or not?
      Dave.
      Definitely yes!
      Once upstream, the last quirky code bits will be fixed by a lot more helping hands.

      Comment


      • #53
        Originally posted by airlied View Post
        so should I merge it or not?
        Dave.
        Oh dear God, yes, please, yes.

        Don’t listen to Linus

        Comment


        • #54
          I very much appreciate the work that was invested into the amdgpu driver "display code", and I would certainly like to see this code included into the mainline kernels (instead of having to run some exotic amd-staging-drm-next tree forever, just to be able to drive my display at all).

          However, everybody should still be prepared to experience lots of weird behaviour from this driver - the bug tracker has lots of such reports, and the number of new reports being opened still seems to exceed the number of bug reports closed.

          What I kind of wonder about is how priorities are set within the team working on the driver. Judging from the commit messages, lots of seemingly very minor issues are being addressed and minor optimizations implemented while there are still several reports of show-stopping bugs that will hit people while just using their GPUs for most mundane things (like browsing the web...). Sometimes it seems to me as if the "cryptocurrency mining hype" has shifted focus away from "making drivers great to display images" to "implementing whatever is needed to mine".

          Comment


          • #55
            Originally posted by cb88 View Post

            Or.. you could just install the DC kernel from a PPA... its a bit buggy but did give me HDMI audio on my RX560 (on one TV at least).
            DC kernel is completely fine, it's just AMDGPU-PRO which is completely useless if you want to use a modern distro.
            ## VGA ##
            AMD: X1950XTX, HD3870, HD5870
            Intel: GMA45, HD3000 (Core i5 2500K)

            Comment


            • #56
              FYI for anyine interested, I tried the new parameter against the older DC for the 4.12 kernel series - amdgpu dosen't report anything back (no acknowedgement or error), so I setup Grub2 to default it on all future kernels on boot (so I won't have to play around if and when it makes it into the offical Kernels).

              Comment


              • #57
                Originally posted by RussianNeuroMancer View Post

                IF abstractions stuff is gone then - please merge it.
                If not - nah, it's need more work.
                Lots is gone. Some still remains, as you would expect when you're talking about 130,000 lines of code.

                Comment


                • #58
                  For anyone at AMD reading this, I think I speak for many of my fellow customers, past and future, when I say thank you for persevering with this project. Once the new code is available in major distros, it seems like AMD will be the clear choice over Nvidia for Linux users, especially if it delivers on the goal of improving performance by getting the driver more directly connected to the hardware developers. Particularly for those who want to use rolling release distros or otherwise have strong objections to proprietary kernel modules, the choice between the proprietary nvidia driver and the subpar nouveau is an unappealing one, and it's great to see so much effort going into the open source AMD driver. I for one like to vote with my wallet and buy from companies who support Linux, for the sake of making my life easier as well as principle.

                  Comment


                  • #59
                    automorphism Don't forget Wayland support! Proprietary NVIDIA driver will only work with Wayland on GNOME.

                    Comment


                    • #60
                      Originally posted by airlied View Post
                      Lols, tinydrm had seen lots of compilers, Linus was just being an idiot. but I don't get your logic (possibly because the concept of logic it outside your grasp).
                      Seeing you don't seem to get why you pissed off Linus by passing on code you hadn't as much as compile tested and wouldn't even compile I'm not sure lead maintainer is the right job for you. The same way CSO wasn't quite the right job for the person who just quit from that position at Equifax.

                      Your proof that I can't judge DAL code quality is that I merged tinydrm? Even though DAL code quality on any level is way worse than tinydrm.
                      If your idea of code in a state ready to be mainlined includes code that throws up nasty compiler warnings and doesn't even compile I don't think your judgment on what's good code and what's bad code is all that reliable...

                      so should I merge it or not?
                      Dave.
                      Judging by the state in which you pushed TinyDRM to Linus is anything to go on you should probably let someone else determine if this code should be mainlined or not.
                      Last edited by L_A_G; 28 September 2017, 03:37 AM.
                      "Why should I want to make anything up? Life's bad enough as it is without wanting to invent any more of it."

                      Comment

                      Working...
                      X