Announcement

Collapse
No announcement yet.

GCN 1.1 On AMDGPU DRM Driver To Now Default To PowerPlay

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

  • jrdoane
    replied
    Originally posted by klapeto View Post
    Maybe this change has to do with the fact that Powerplay code miraculously solved the notorious DPM bug that some of us R9 390 users had the past 2.5 years. https://bugs.freedesktop.org/show_bug.cgi?id=91880
    I wouldn't be surprised. I was plagued by that particular bug ever since I completely stopped using Windows a little over 2 years ago. I constantly had to force clocks up to keep the card stable. The moment that I enabled the amdgpu.dpm=1 kernel option, this bug practically went away with my MSI R9 390. For me, this was huge and eliminated years of headache.

    Leave a comment:


  • chithanh
    replied
    Originally posted by rene View Post
    Still some ATi AGP regression on some ppc32 Macs: https://www.youtube.com/watch?v=uYx9X92iQHM
    I think bugs.freedesktop.org is the place to report such issues.

    That being said, I noticed that there was some problems between AGP driver and libata PATA for my Powerbook G4 Titanium. Switching to IDE driver or disabling AGP solved stability problems.

    Leave a comment:


  • klapeto
    replied
    Maybe this change has to do with the fact that Powerplay code miraculously solved the notorious DPM bug that some of us R9 390 users had the past 2.5 years. https://bugs.freedesktop.org/show_bug.cgi?id=91880

    Leave a comment:


  • nanonyme
    replied
    Also AGP all in all belongs in nightmarish hit-or-miss quirky technologies. It was even worse on ppc32. You can probably get more compatible hardware from a trash can

    Leave a comment:


  • nanonyme
    replied
    Originally posted by rene View Post

    Yep, not at the Cube right now, but IIRC it is a Radeon 7500 (RV200) so basically a die-shrink of the R100 in the then new 150 nm manufacturing process. But still a regression though. Really hate that older hardware support is constantly getting broken for no good reason. Aka "never change a running system^W code", … :-/
    1) That's handled by older kernel driver radeon
    2) ppc32 had tons of quirks starting from cache coherency differences with x86 and getting bugs fixed meant significant personal debugging effort even in its prime time. The CPU+motherboard+GPU combination needed to be reproduced for proper debugging
    3) ppc32 architecture has essentially been dead since Apple moved to Intel around ten years ago. ppc64 is still very much alive but was never used in old Macs

    Leave a comment:


  • rene
    replied
    Originally posted by sandy8925 View Post
    rene - I don't think those have Sea Island GPUs.......those are pretty old GPUs.
    Yep, not at the Cube right now, but IIRC it is a Radeon 7500 (RV200) so basically a die-shrink of the R100 in the then new 150 nm manufacturing process. But still a regression though. Really hate that older hardware support is constantly getting broken for no good reason. Aka "never change a running system^W code", … :-/

    Leave a comment:


  • Guest
    Guest replied
    rene - I don't think those have Sea Island GPUs.......those are pretty old GPUs.

    Leave a comment:


  • rene
    replied
    Still some ATi AGP regression on some ppc32 Macs: https://www.youtube.com/watch?v=uYx9X92iQHM

    Leave a comment:


  • Aeder
    replied
    I wonder if that means I'll stop getting those DPM warnings on boot on a R7 M260

    Leave a comment:


  • skyrrd
    replied
    That's quite nice, but as long as hdmi-audio doesn't work on hawaii using the amdgpu-driver does leave a bitter taste.
    And yes i have tried with amdgpu.dc=1 but that doesn't work...even leads to system-hangs on shutdown and still no audio output available

    Leave a comment:

Working...
X