Announcement

Collapse
No announcement yet.

Red Hat Enterprise Linux 10 Dropping The X.Org Server Except For XWayland

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

  • Originally posted by t1r0nama View Post
    Does steam run on native wayland? When it will then we'll be ready. Xwayland shenaningans is just crappy solution to non existing problem!
    https://github.com/ValveSoftware/steam-for-linux/issues/4924 so Far is it not running wayland Native. But the % of gamer on wayland is rising. So the pressure should slowly get higher to work on it. www.gamingonlinux.com/index.php?module=statistics&view=trends#SessionTyp e-top (not a perfect statistic but the best we have)

    Till wine/proton have wayland native is the steam client the least concern as you need xwayland anyway for most games.

    Comment


    • Originally posted by oleid View Post

      Xvfb, which is used by xpra provides no hardware acceleration at all. In the end a video stream is transmitted to the view side. You could as well replace this by some RDP based solution.
      Sure, but Xvfb is just something you would do on a HPC platform like


      It's not something the vast majority of desktop or steam users will care about or miss from RH10, RH10 is still suitable for many embedded and mobile use cases.
      Last edited by mSparks; 30 November 2023, 06:17 AM.

      Comment


      • Originally posted by mSparks View Post
        Sure, but Xvfb is just something you would do on a HPC platform like
        https://docs.hpc.shef.ac.uk/en/lates...apps/xvfb.html
        Who is going to maintain xvfb. Remember its been Redhat staff maintaining xvfb including maintaining over 400 computers for regression testing that they are going to stop doing.



        Redhat has released drop in replacement for Xvfb based around Wayland parts. Yes xwfb-run.py based around Xwayland and other headless wayland compositors.

        This xwfb-run can in fact use local EGL GPU for accelerated opengl/vulkan and also can turn on RDP and VNC outputs as well. So xwfb-run is upgraded Xvfb.

        xwfb-run basically will not need xpra in time because seamless RDP will replace what xpra did..


        Comment


        • Originally posted by mSparks View Post
          It is funny that you linked to this cluster. The first note says:
          ShARC will be decommissioned on the 30th of November 2023, after which time users will no longer be able to access that cluster and any jobs running or queueing at that time will be cancelled. Please see our page of important info about ShARC’s decommissioning.
          I know that some badly designed applications used to require some graphic output. Back in the days I had this problem myself. Qt console applications apparently required a display variable set to create PDF files. For no reason. That's why I switched to Cairo, which didn't have this limitation. I don't know if this limitation still exists, it originates from a time when Qt didn't have Wayland output. But don't you think those applications wouldn't work with a different graphical output as well? You could eg start wlroots in headless mode. Or use xwfb-run.

          Comment


          • Originally posted by oiaohm View Post

            Who is going to maintain xvfb.

            Companies that actually have the in house expertise to build and certify HPC platforms

            Author: OpenELA Community Date: November 2nd, 2023 Source code for packages is now available (and more is on the way) Incorporation is complete, and the Technical Steering Committee has been formed to guarantee access to Enterprise Linux packages RENO, Nev., AUSTIN, Texas, and LUXEMBOURG—November 2, 2023—OpenELA today announced the public release of the Enterprise Linux source code along with important technical and governance milestones. OpenELA is a trade association of open source Enterprise Linux distribution developers originally founded by CIQ, Oracle and SUSE.


            Originally posted by oleid View Post

            It is funny that you linked to this cluster. The first note says:

            Obviously, it doesn't have cuda, all their documentation says that, they've replaced that cluster with Stanage

            Because all the HPC hottness is now


            But it doesn't matter that RH10 will only have limited Nvidia and Cuda support - these are not something the vast majority of desktop or steam users will care about or miss from RH10, RH10 is still suitable for many embedded and mobile use cases, and nvidia doesn't do much for embedded and mobile use cases .
            Last edited by mSparks; 30 November 2023, 09:28 AM.

            Comment


            • Originally posted by avis View Post

              You've skipped over the most important part of my message: RedHat can perfectly continue offering X.org without any certification.
              Then its not Red Het enterprise but something else entirely. The whole point of Red Het enterprise is that critical parts of the system are certified

              Comment


              • Originally posted by mSparks View Post
                Companies that actually have the in house expertise to build and certify HPC platforms
                Do they really. None of the openELA members have the collection of hardware required to in fact do that certification.

                Also note I said who. None of the parties that make up openELA have any current core x.org developers. So this means they are starting off with basically a zero knowledge base of how x.org server works internally this is going to be bad there are many design land mines.

                Originally posted by mSparks View Post
                But it doesn't matter that RH10 will only have limited Nvidia and Cuda support .
                This is a question that still up in the air. Nvidia support is progressing. Do remember RH10 release date is 2025.
                Linux RedHat 10 will be released in 485 days! #released.info


                So Nvidia still has a year of development on their drivers to deal with issues.


                Comment


                • Originally posted by oiaohm View Post

                  Do they really. None of the openELA members have the collection of hardware required to in fact do that certification.


                  Originally posted by oiaohm View Post
                  So Nvidia still has a year of development on their drivers to deal with issues.​
                  Very unlikely, no one is going to be installing H200 GPUs on an OS that is entirely dedicated to embedded and mobile use cases that requires everything to be rewritten from scratch like RH10.
                  Last edited by mSparks; 30 November 2023, 12:08 PM.

                  Comment


                  • Originally posted by mdedetrich View Post

                    Then its not Red Het enterprise but something else entirely. The whole point of Red Het enterprise is that critical parts of the system are certified
                    RedHat has never had qualms about that. A ton of RHEL instances run totally ... unsupported NVIDIA.

                    Comment


                    • Suse has been depending on Redhat to perform the x.org server work. That suse has got some server market share it makes no difference to the base problem.

                      The reality here does not matter if it Oracle, Suse or the other member of Openela they don't have the hardware setup to check any alteration to X11 server for issues or the personal todo it..

                      There is specialist testing equipment so you can confirm a GPU is outputing correctly. GPU testing is not as straight forwards as one would think.

                      Reason why the other vendors have left the graphical/x.org regression testing to Redhat is it a 2 million dollar buy in with a half a million dollars a year to keep the testing equipment current and then the skills to use it mSpark.

                      This is the problem Redhat has been spending lots of resources on testing and equipment to perform that testing.

                      Originally posted by mSparks View Post
                      Very unlikely, no one is going to be installing H200 GPUs on an OS
                      You have one big fatal problem with that statement. Redhat/IBM owns H200GPUs and other NVIDIA GPU as part of there platform testing hardware. So that not no one.

                      Comment

                      Working...
                      X