Announcement

Collapse
No announcement yet.

Are The Open-Source Graphics Drivers Good Enough For Steam Linux Gaming?

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

  • pal666
    replied
    Originally posted by << ⚛ >> View Post

    When VSync is enabled (60 Hz typically), GPU power consumption with the theoretical 420 FPS is lower than with 360 FPS.
    not really if it is faster only because of faster clocks

    Leave a comment:


  • gamerk2
    replied
    Originally posted by F1esDgSdUTYpm0iy View Post
    Whereas it's obvious you're just trolling, I would like to respond to this remark -- For this particular machine I am using right now, without discrete GPU, Windows is the invalid platform for games. I tried, it won't let me play them properly. Too much overhead from the OS. Switched to Linux and Wine and suddenly, I'm able to play D3, HOTS and other (semi-)modern titles without too much of a hassle. On the OSS drivers no less.

    Intel or AMD? Intel's Windows driver is very poor in particular, especially on the OpenGL side. AMD simply suffers from a weak CPU. OS overhead should be minimal, especially in fullscreen, unless you are loosing performance in driver-land.

    Leave a comment:


  • M@yeulC
    replied
    Originally posted by ChemicalBrother View Post

    I have more problems running games with Catalyst than with radeonsi, so Catalyst is the actual restriction, not the open-source driver.
    For me it is quite the contrary. Plus, OSS drivers are always compatible with latest kernel and Xorg versions, and have wayland support.

    Insurgency is looking quite good here on Arch with a 6870. ~140fps on average, I would say, in the training mission, with some punctual drops to 40 fps. I might do some testing again with the gallium HUD, if I have time to do so.

    Speaking about this HUD... Isn't it possible to specify a file to output data to ? I though it was possible :/
    If not, I think we should just specify a folder, where to save data in different text files, without rendering anything on screen. This would even have the side effect of lowering the small overhead bring up by the graphs.

    Leave a comment:


  • duby229
    replied
    Originally posted by << ⚛ >> View Post

    The relevant source code is in mesa-11.0.4/src/gallium/auxiliary/hud. It would be possible to add file output there.
    That would be awesome. It would effectively mean that as long as a game has a benchmark mode any data the gallium provides could be part of the data collection. That sounds like a fantastic idea.
    Last edited by duby229; 30 October 2015, 08:43 AM.

    Leave a comment:


  • atomsymbol
    replied
    Originally posted by Michael View Post
    Not unless it can all be automated....
    The relevant source code is in mesa-11.0.4/src/gallium/auxiliary/hud. It would be possible to add file output there.

    Leave a comment:


  • Michael
    replied
    Originally posted by haagch View Post
    Michael,
    I think specifically the benchmark mode in metro redux is not good. Can you record CPU and GPU usage for the Metro Redux benchmark with mesa? I think they will both be very low. As others have said, in the actual game it performs a lot better. Can you make a short playtest on one of the PCs (R9 290 for example) and see if the actual game has better CPU and GPU usage than the benchmark mode? Maybe a screenshot with GALLIUM_HUD. These should be the interesting values:
    GALLIUM_HUD="fps,cpu,GPU-load,num-bytes-moved,buffer-wait-time,num-compilations"
    Not unless it can all be automated.... But likely after my Intel article today/tomorrow will stop these Steam tests for a while as not enough support has been generated for the amount of time invested in these articles.

    Leave a comment:


  • haagch
    replied
    Michael,
    I think specifically the benchmark mode in metro redux is not good. Can you record CPU and GPU usage for the Metro Redux benchmark with mesa? I think they will both be very low. As others have said, in the actual game it performs a lot better. Can you make a short playtest on one of the PCs (R9 290 for example) and see if the actual game has better CPU and GPU usage than the benchmark mode? Maybe a screenshot with GALLIUM_HUD. These should be the interesting values:
    GALLIUM_HUD="fps,cpu,GPU-load,num-bytes-moved,buffer-wait-time,num-compilations"

    Leave a comment:


  • F1esDgSdUTYpm0iy
    replied
    Originally posted by kir? View Post
    The article seems like a joke: linux is not a valid platform to play games.
    Whereas it's obvious you're just trolling, I would like to respond to this remark -- For this particular machine I am using right now, without discrete GPU, Windows is the invalid platform for games. I tried, it won't let me play them properly. Too much overhead from the OS. Switched to Linux and Wine and suddenly, I'm able to play D3, HOTS and other (semi-)modern titles without too much of a hassle. On the OSS drivers no less.

    Leave a comment:


  • Passso
    replied
    Originally posted by kir? View Post
    The article seems like a joke: linux is not a valid platform to play games. Everything involving graphics have problems on linux. In the background the situation is not proposable while in the foreground do you want to even play with open source drivers? Yeah you're doing it right! Use amd too and go fast! XD Good luck!
    WARNING : Troll detected! Please reinstall yourself.

    Leave a comment:


  • kir?
    replied
    The article seems like a joke: linux is not a valid platform to play games. Everything involving graphics have problems on linux. In the background the situation is not proposable while in the foreground do you want to even play with open source drivers? Yeah you're doing it right! Use amd too and go fast! XD Good luck!

    Leave a comment:

Working...
X