Announcement

Collapse
No announcement yet.

Intel Has Good DRM Driver Changes In Linux 3.9

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

  • Intel Has Good DRM Driver Changes In Linux 3.9

    Phoronix: Intel Has Good DRM Driver Changes In Linux 3.9

    The Intel DRM graphics driver will feature a number of user-facing improvements within the Linux 3.9 kernel...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    - Stolen memory support from the Fastboot patch-set. This work is to ultimately try to avoid mode-setting unnecessarily by the Intel DRM driver when it's loaded if the BIOS-configured mode is in good shape.
    - VT/Switchless resume support for a cleaner experience when restoring your system from a suspended state.
    Once a feature list contains entries like these, then that means the driver is in pretty good shape. It's pretty good to see that the Linux kernel can provide enough infrastructure to make GPU support top notch.

    Aside from that. The first comment on his blog mentions a leak. This is a userspace bug which I have seen as well on all nouveau driven boxes (three of them).

    So, finally the driver support is getting pretty good. Userspace bugs start to pop-up lol...

    Comment


    • #3
      Isn't Ironlake Gen5 instead of Gen4?

      Comment


      • #4
        Originally posted by AlbertP View Post
        Isn't Ironlake Gen5 instead of Gen4?
        The bug in handling relocations that require synchronisation and pagefaults whilst being interrupted by signals, and is common to all generations. However, it is pretty much restricted to only being hit when using UXA on gen4/5, due to its many sub-optimal design choices exposing the race.

        Daniel mentions that the bug is prevalent on both gen4/5, but it became widely known as the infamous Ironlake bug when Dave Airlie, Greg Kroah-Hartman and then Linus Torvalds encountered it on their Ironlake boxes, aside from the many less vocal victims. In the article, Michael is trying to contract the information about that bug fix and further relocation improvements into a single sentence, and mixed up the codenames.

        TL;DR: yes.

        Comment


        • #5
          I learned my lesson:


          With intel if your shit is working: NEVER EVER EVER FUCKING UPDATE LIKE EVER




          for one thing they improve, they glitch 3 or 4 others
          Last edited by Pallidus; 18 February 2013, 05:32 PM.

          Comment

          Working...
          X