Announcement

Collapse
No announcement yet.

The Current State Of Radeon Power Management

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

  • phoronix
    started a topic The Current State Of Radeon Power Management

    The Current State Of Radeon Power Management

    Phoronix: The Current State Of Radeon Power Management

    It's been a while since last looking at the state of power management for Radeon GPUs, but here's an updated look at the various options surrounding power management for modern ATI/AMD graphics processors and their effectiveness. Various drivers, graphics cards, and tuning options are compared.

    http://www.phoronix.com/vr.php?view=16793

  • Hamish Wilson
    replied
    Originally posted by sstp View Post
    Recently one of the developers said that people doesnt care much about power management, so its not in high priorities. Can we do something about it? May be some poll? There to many work(opengl, glsl, video acc, new chipset) to see usable pm even in year.
    Where did they ever say that? I heard them say that it is a lot of effort to get fixed, but never that they do not care. And keep in mind many of the people who are working on those other priorities may not be able to work on power management as they do not know how or in some cases may be told to work on certain things from their employers. As it stands though, the new chip-set enablement definitely seems more important than the others, with the others being more iffy.

    Though that is my opinion, and I do not have any of the skills needed to change the situation, so I am just going to walk off and shut up now.

    Leave a comment:


  • Ansla
    replied
    Actualy, this limitations were mentioned in the article.

    Leave a comment:


  • DaemonFC
    replied
    He forgot to mention that dynpm is more or less unusable on a lot of Radeons because it causes display flickering as it cycles through the available clock speeds and it only works when you only have one display. If you have two or more displays it falls back to profile:auto which tends to work fine other than using more power.

    Leave a comment:


  • Death Knight
    replied
    I have HD6850, using open source stack, but it consume ~60-80W on idle!. Not acceptable for 7/24 open case...
    I can set "low" my GPU manually, than have problems when I require more power for short bursts (like compiz animations...)
    Also on "low GPU" state, card consumes ~10 more Watt than "catalyst"...

    Does it require much work to implement proper auto or dynPM settings? I think it's shame for AMD. Is there any DRM issue on power management of GPUs?
    I simply can't use my card due this. I just detached my radeon 6850 from MB and using integrated 790GX.

    Leave a comment:


  • AnonymousCoward
    replied
    Originally posted by sstp View Post
    Recently one of the developers said that people doesnt care much about power management, so its not in high priorities. Can we do something about it? May be some poll?
    http://www.phoronix.com/scan.php?pag..._results&num=1 for example? Where power management isn't far behind the other two useful options (2d is mostly fast enough anyway these days, and Gallium and KMS are just means to an end, not that interesting in themselves for end-users).

    Leave a comment:


  • DanL
    replied
    Michael, the PTS specs on page 2 are terrible to look at. Is this one of those things that's better with a premium subscription? Despite the tablet trend, some of us still have relatively large desktop monitors..

    Leave a comment:


  • curaga
    replied
    Missing results on the last graph of page 3, Cayman GPU voltage.

    Where's low from the graph? It's mentioned in the text though.

    Leave a comment:


  • sstp
    replied
    Recently one of the developers said that people doesnt care much about power management, so its not in high priorities. Can we do something about it? May be some poll? There to many work(opengl, glsl, video acc, new chipset) to see usable pm even in year.

    Leave a comment:


  • AnonymousCoward
    replied
    And the one interesting thing to compare, how dynpm stacks up against the 'auto' profile, is sorely missing from the test. On my system, dynpm is still far worse, getting 5C higher just idling on the KDE desktop.

    Leave a comment:

Working...
X