Linux gamers won't be affected by RX 5000/6000 series driver shift — AMD changes limited to Windows thanks to separated development

AMD
(Image credit: AMD)

AMD confused the gaming community greatly after it announced it would put RDNA 1 and RDNA 2 driver support in "maintenance mode", but later corrected that the two architectures would still be receiving game optimizations regardless. Apparently, this confusion also led some people to believe game optimizations would also stop on AMD's Linux drivers as well. Luckily, Phoronix has published an article reaffirming that Linux support has not changed with this announcement, primarily because Linux AMD drivers are developed separately from their Windows counterparts.

Linux driver support is maintained very differently on AMD GPU drivers compared to AMD's GPU drivers for Windows. Linux driver support for AMD GPUs typically dwarfs what AMD supports officially in Windows. Kernel driver support reportedly dates back all the way to AMD's GCN 1.2 architecture, which includes GPUs such as the R9 390X and R9 Fury X.

Google Preferred Source

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Aaron Klotz
Contributing Writer

Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.

  • -Fran-
    This is getting an awful lot of attention.

    I wish back when nVidia burned GPUs with driver updates they would've had this much coverage.

    Regards.
    Reply
  • Amdlova
    -Fran- said:
    This is getting an awful lot of attention.

    I wish back when nVidia burned GPUs with driver updates they would've had this much coverage.

    Regards.
    Nvidia pays well
    Reply
  • LordVile
    -Fran- said:
    This is getting an awful lot of attention.

    I wish back when nVidia burned GPUs with driver updates they would've had this much coverage.

    Regards.
    Because normally Nvidia don’t try to do it in less than 5 years from the launch of a product. Pascal has just been sunset after what 9 years? 8 for the Ti models
    Reply
  • ezst036
    -Fran- said:
    This is getting an awful lot of attention.

    I wish back when nVidia burned GPUs with driver updates they would've had this much coverage.

    Regards.
    Nvidia did hold on to support for GPUs longer than AMD's latest recent announcement(on the Windows side) but even moreso in this instance it's AMD vs AMD.

    Having open source drivers means AMD does/will have the longest actively-supported GPUs on the planet meanwhile their Windows are potentially the shortest.
    Reply
  • -Fran-
    LordVile said:
    Because normally Nvidia don’t try to do it in less than 5 years from the launch of a product. Pascal has just been sunset after what 9 years? 8 for the Ti models
    True. They try to burn your GPU as soon as possible, so you get a new one. Good point!

    Since, as a reminder, the 12VHighFailure connector is working as intended, right?

    ezst036 said:
    Nvidia did hold on to support for GPUs longer than AMD's latest recent announcement(on the Windows side) but even moreso in this instance it's AMD vs AMD.

    Having open source drivers means AMD does/will have the longest actively-supported GPUs on the planet meanwhile their Windows are potentially the shortest.
    I think the interesting bit is what nVidia does not say to me. Are Turing cards actually receiving game performance improvements or just bug fixes and they're just not disclosing their internal classification for them? From many years, nVidia doesn't say much about anything they do. Remember the paid actors in forums hyping and posing as normal people? Pepperidge Farm remembers.

    The open source point is a good one though. AMD on Linux does not "die", pretty much.

    Regards.
    Reply
  • LordVile
    -Fran- said:
    True. They try to burn your GPU as soon as possible, so you get a new one. Good point!

    Since, as a reminder, the 12VHighFailure connector is working as intended, right?
    12VHP is a bad connector but that’s not really the point?
    -Fran- said:


    I think the interesting bit is what nVidia does not say to me. Are Turing cards actually receiving game performance improvements or just bug fixes and they're just not disclosing their internal classification for them? From many years, nVidia doesn't say much about anything they do. Remember the paid actors in forums hyping and posing as normal people? Pepperidge Farm remembers.

    The open source point is a good one though. AMD on Linux does not "die", pretty much.

    Regards.
    They seem to be
    Reply
  • Notton
    I don't want to sound like a conspiracy theorist, but...
    Nvidia botched drivers news cycle ended rather rapidly, whilst AMD drivers are still in the news cycle after some people got duped by fake news.

    Nvidia drivers still aren't fixed.
    Nvidia drivers don't work with OpenGL properly and constantly crashes games that run on it. This has been going on for over a year now.
    Reply
  • BFG-9000
    On the very same month Fermi would be released, nVidia released 196.75 which disabled the fans to cause the then-current big Tesla models (GTX 8800, 9800, 275/280/285/295) to burn up

    576.02 similarly disabled fan control but only after waking from sleep, after which the temperature sensors no longer updated (NvAPI_GPU_GetThermalSettings), leading to a couple reportedly burned RTX 5090 cards from fans remaining at idle. On the bright side, not that many people tried them due to all the issues with drivers newer than 572 or 566, depending on the game. Perhaps that's why it didn't make the news here on Tom's, especially as it mostly just caused crashing problems from the overheating

    I mean if the drivers are crashing all of the time anyway, the only thing most people might notice is 576.02 would seem even less stable than usual. While modern hardware is protected so should just shut off if core temperatures really got dangerous, nVidia has a bad habit of leaving little margin in their reference designs, so some VRM component could give up instead if you are unlucky

    Open source support depends on how popular it is. For example while R300-R500 recently got new driver updates in Mesa-24 and Mesa-25.3, the unloved old R200 drivers got spun-off into Mesa-amber
    Reply