Announcement

Collapse
No announcement yet.

9 Year Old Mesa Bug Fixed, Addresses Stability With RadeonSI Gallium3D

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

  • #21
    Originally posted by bridgman View Post

    Dozens of developer-years of driver effort plus equivalent amount or more in the OS.
    Correct. AMD does have a very large team working on the Windows driver. On the OpenSource stack, however, I believe it's between 6 and 9 people, correct me if I'm wrong.

    Anyways, a very good catch by Marek. Thanks to everyone involved in getting this terrible bug fixed.

    Comment


    • #22
      Originally posted by bridgman View Post
      Dozens of developer-years of driver effort plus equivalent amount or more in the OS.
      I'm going to guess that the kiddos on Phoronix forum don't know about WDDM(among other things, to make indirect rendering possible for Longhorn Aero!), and why it made Vista suck for years, and made graphics not insane on Windows 7. They surely don't know how many millions of dollars were spent paying very smart people to scratch their heads about it for years of their life. ;- )

      Comment


      • #23
        Originally posted by Amarildo View Post
        Correct. AMD does have a very large team working on the Windows driver. On the OpenSource stack, however, I believe it's between 6 and 9 people, correct me if I'm wrong.

        Anyways, a very good catch by Marek. Thanks to everyone involved in getting this terrible bug fixed.
        He's referring to what it took to make it so that the Windows NT kernel was (more) isolated from serious failures in drivers, especially display drivers. Windows NT runs display drivers as isolated processes, Linux has no such thing; and even if Linux had memory-isolated drivers, it would still take many developer-years of effort to develop drivers which behave well in that environment.

        This is currently not doable on Linux the way that it is done on NT.
        Last edited by microcode; 10 December 2016, 02:15 AM.

        Comment


        • #24
          Still not in Padoka! :-)

          Comment


          • #25
            Originally posted by Qaridarium
            very good to read. and to all the people who blamed the negative people all the years to be wrong about bugs like this now you have it black on white if people experience this developers better listen.
            Huh? Marek said that the way it got fixed was someone uploaded a test case that allowed him to reproduce the error. That's exactly what everyone has been saying - don't just complain randomly on phoronix, go create a bug report and help solve it.

            Comment


            • #26
              Originally posted by microcode View Post

              He's referring to what it took to make it so that the Windows NT kernel was (more) isolated from serious failures in drivers, especially display drivers. Windows NT runs display drivers as isolated processes, Linux has no such thing; and even if Linux had memory-isolated drivers, it would still take many developer-years of effort to develop drivers which behave well in that environment.

              This is currently not doable on Linux the way that it is done on NT.
              Oooh, I see. Thanks for the info

              I do remember seeing "Hardware Virtualization coming to Linux" somewhere, though.

              Comment


              • #27
                Wow... For someone like me who plays TF2 a lot, this exact bug was what was holding me back from using/switching to Linux. Thank you a lot for your efforts Marek!

                Comment

                Working...
                X