Announcement

Collapse
No announcement yet.

AMD Kabini HD 8210: RadeonSI Gallium3D Can Outperform Catalyst

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

  • curaga
    replied
    Originally posted by verde View Post
    libtxc-dxtn-s2tc0 is installed. My hardware is ok on Windows so... In my my bug report someone faced similar problems with Radeon 6850 and that leeds me to believe that the problem is wider than my GPU.

    https://bugs.freedesktop.org/show_bug.cgi?id=73739
    There's a distinct lack of apitrace in that bug, FYI.

    Leave a comment:


  • djdoo
    replied
    Originally posted by verde View Post
    libtxc-dxtn-s2tc0 is installed. My hardware is ok on Windows so... In my my bug report someone faced similar problems with Radeon 6850 and that leeds me to believe that the problem is wider than my GPU.

    https://bugs.freedesktop.org/show_bug.cgi?id=73739
    The 32 bit version of the lib is needed... The 64 bit one was installed for me too here on opensuse 13.1 x64.

    Leave a comment:


  • verde
    replied
    Originally posted by brosis View Post
    HyperZ, Dxtn(s3tc) or use s2tc, floating point textures must be enabled in mesa, and finally - make sure you use 32bit Z buffer in game. But it also could be defect hardware, crushed ball pins can result in such artifacts.
    libtxc-dxtn-s2tc0 is installed. My hardware is ok on Windows so... In my my bug report someone faced similar problems with Radeon 6850 and that leeds me to believe that the problem is wider than my GPU.

    Leave a comment:


  • brosis
    replied
    Originally posted by verde View Post
    I did try to disable HyperZ and it didn't help. In my case it is true. I've tried many games and the result rarely was acceptable. In Unigine benchmarks like Tropics, there are basic textures missing and flickering is everywhere on shadows.. I can record a video to show you an example.
    HyperZ, Dxtn(s3tc) or use s2tc, floating point textures must be enabled in mesa, and finally - make sure you use 32bit Z buffer in game. But it also could be defect hardware, crushed ball pins can result in such artifacts.

    Leave a comment:


  • bridgman
    replied
    Originally posted by Veerappan View Post
    I believe that Tim Writer took over the OSS driver team and that bridgman is now working on HSA and related projects. That being said, I don't work for AMD, and I've only had 1/2 a cup of coffee this morning.
    Right. My other jobs (including the ones related to IP issues like this) have new owners as well.

    Leave a comment:


  • Veerappan
    replied
    Originally posted by TAXI View Post
    I'm not sure if I understand you correctly: Does that mean you aren't allowed to talk about that or that you're the wrong one to talk to? I guess it means the second but then the question is who are these nice folks (are they active here on phoronix, too) ?
    I believe that Tim Writer took over the OSS driver team and that bridgman is now working on HSA and related projects. That being said, I don't work for AMD, and I've only had 1/2 a cup of coffee this morning.

    Leave a comment:


  • V10lator
    replied
    Originally posted by GreatEmerald View Post
    3 years? S3TC has been around since 1997 ("S3" – created for Savage 3D cards), so that's 17 years. And the current holder is indeed HTC.
    I was talking about blocking this features in Mesa cause of patents, not how long the pantents itself existed (and this info was just from a quick google search). Anyway, do you have a prove that it's really HTC? I'm asking cause of http://www.phoronix.com/forums/showt...703#post239703

    Originally posted by bridgman View Post
    *bridgman hides his face and looks at the nice folks who took over the open source graphics and IP areas from him
    I'm not sure if I understand you correctly: Does that mean you aren't allowed to talk about that or that you're the wrong one to talk to? I guess it means the second but then the question is who are these nice folks (are they active here on phoronix, too) ?

    Leave a comment:


  • bridgman
    replied
    Originally posted by TAXI View Post
    Could somebody from AMD (I'm looking into your direction bridgman) please share some insights on this?
    *bridgman hides his face and looks at the nice folks who took over the open source graphics and IP areas from him

    Leave a comment:


  • dungeon
    replied
    So OK we can just wait for 3 years more and s3tc issue will be resolved .

    Leave a comment:


  • GreatEmerald
    replied
    Originally posted by TAXI View Post
    I wonder why Intel has no problem enabling floating-point while AMD has: http://www.phoronix.com/scan.php?pag...tem&px=MTI3OTY

    All this patent problems seem to hang here since around 2011, so 3 years! And the last info I could find was that people aren't even sure the s3tc patent is valid or who owns it atm (HTC or AMD) so... well... something has to be done about this.

    Could somebody from AMD (I'm looking into your direction bridgman) please share some insights on this?
    3 years? S3TC has been around since 1997 ("S3" – created for Savage 3D cards), so that's 17 years. And the current holder is indeed HTC.

    Leave a comment:

Working...
X