Announcement

Collapse
No announcement yet.

Fedora 21 Drops Support For A Bunch Of Old GPUs

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

  • Fedora 21 Drops Support For A Bunch Of Old GPUs

    Phoronix: Fedora 21 Drops Support For A Bunch Of Old GPUs

    While Fedora 20 isn't going to be released until at least December, changes are already ongoing for its successor, Fedora 21. This first major Fedora Linux release of 2014 will abandon support for quite a few older graphics processors...

    http://www.phoronix.com/vr.php?view=MTQ5NDk

  • #2
    Aww man, now where am I going to run my totally sweet Trident 4MB video card from the 90's?

    Comment


    • #3
      Geode

      Anyone have more details why about the Geode change? I thought they had found a maintainer..

      Also, what about KMS based VESA support?

      Comment


      • #4
        Originally posted by meltingrobot View Post
        ... totally sweet Trident 4MB video card ...
        Trident sucks!! Matrox G200 kicks its ass big time!!

        (Yeah, let's start a mid-90's video card flame war for a change. This nvidia vs amd thing is getting boring...)

        Comment


        • #5
          Some of these are the embedded video on servers. And given "Red Hat" (yes, I know, Red Hat is *not* Fedora) propensity for *graphical only* administration tools, this could be a mistake (??).

          I know even contemporary cards made by vendors with supposed KMS support do not work with KMS, even though some people think they should (particular contemporary Nvidia's come to mind).

          I will certainly be watching this one...

          Comment


          • #6
            Originally posted by cjcox View Post
            Some of these are the embedded video on servers. And given "Red Hat" (yes, I know, Red Hat is *not* Fedora) propensity for *graphical only* administration tools, this could be a mistake (??).

            I know even contemporary cards made by vendors with supposed KMS support do not work with KMS, even though some people think they should (particular contemporary Nvidia's come to mind).

            I will certainly be watching this one...
            Someone else is free to pick them up if they care enough, even another Red Hat employee. Ajax is just saying "Its not me anymore." One very important thing that Michael did cover last time, I think, but definitely didn't mention THIS time... These old drivers? No one tests them anymore. Their maintenance is "Did it compile for this release? Yes? Good. No? Fix the build error." They could be broken at runtime, but as long a they build okay, they get shipped. So the idea of them being "supported" isnt even really valid NOW, pre-drop.

            Comment


            • #7
              The default install of fedora has been fail (either slow, glitchy, drops you to an error screen, or if you are lucky falls back to fallback mode) on older graphics cards since Gnome 3. Its about time the system requirements said 'You need a new (last 5 years) intel, amd or nvidia', and directed people with other hardware to a MATE/xfce spin.

              We still have plenty of workstations with MGA G200e cards.

              Comment


              • #8
                Nooooooooooooo!!!

                I still have around (with tons of dust collected):

                R128, Rendition, s3virge, Savage, Matrox G200 and yes a Trident !

                the good ol days

                P.S. I wonder if the open source Vga crowdfunding project is able to match any of those.

                Comment


                • #9
                  Originally posted by madjr View Post
                  P.S. I wonder if the open source Vga crowdfunding project is able to match any of those.
                  I think you can piece together something on a breadboard that can match those...

                  Comment


                  • #10
                    Originally posted by gQuigs View Post
                    Also, what about KMS based VESA support?
                    Using VESA drivers is the very definition of not having support. It's not that your card will not work at all, but it will use a highly generic, unaccelerated driver.

                    Comment


                    • #11
                      Originally posted by mrugiero View Post
                      Using VESA drivers is the very definition of not having support. It's not that your card will not work at all, but it will use a highly generic, unaccelerated driver.
                      I think he meant this: http://www.phoronix.com/scan.php?pag...tem&px=MTI4NTU

                      Comment


                      • #12
                        Originally posted by Ericg View Post
                        Indeed. The VESA driver we currently use is User Mode Set. I want them to be able to drop UMS entirely...

                        Comment


                        • #13
                          Originally posted by gQuigs View Post
                          Indeed. The VESA driver we currently use is User Mode Set. I want them to be able to drop UMS entirely...
                          There is a generic userspace modesetting driver called xorg-x11-drv-modesetting , these days. For any bit of hardware which has an in-kernel KMS driver but no specific userspace driver, this will be used as the fallback driver; this is already the case for some of the hardware mentioned, IIRC.

                          Comment


                          • #14
                            Originally posted by gQuigs View Post
                            Indeed. The VESA driver we currently use is User Mode Set. I want them to be able to drop UMS entirely...
                            IIRC the problem with KMS VESA is that with a straightforward implementation you end up having to execute real mode BIOS code in the kernel.

                            It would presumaly be possible to create a sufficiently robust "sandboxed" x86 emulator (and maybe something like QEMU is there today) for this to be considered safe but I haven't heard about anyone looking into it.

                            Comment


                            • #15
                              Originally posted by bridgman View Post
                              IIRC the problem with KMS VESA is that with a straightforward implementation you end up having to execute real mode BIOS code in the kernel.

                              It would presumaly be possible to create a sufficiently robust "sandboxed" x86 emulator (and maybe something like QEMU is there today) for this to be considered safe but I haven't heard about anyone looking into it.
                              Running an x86 emulator for an already non-accelerated driver doesn't sound like a decent performing solution, anyway.

                              Comment

                              Working...
                              X