Page 1 of 2 12 LastLast
Results 1 to 10 of 13

Thread: LunarG ILO Gallium3D vs. Intel's DRI Driver On Mesa 10.3-devel

  1. #1
    Join Date
    Jan 2007
    Posts
    14,532

    Default LunarG ILO Gallium3D vs. Intel's DRI Driver On Mesa 10.3-devel

    Phoronix: LunarG ILO Gallium3D vs. Intel's DRI Driver On Mesa 10.3-devel

    The Linux graphics benchmarks we have to publish today at Phoronix are some tests of the Intel "ILO" Gallium3D driver that is independently developed by LunarG as an unofficial alternative to the classic Intel Mesa DRI driver that's officially supported by the Intel Open-Source Technology Center.

    http://www.phoronix.com/vr.php?view=20533

  2. #2
    Join Date
    Feb 2011
    Location
    France
    Posts
    189

    Default

    You should bench true games !

  3. #3

    Default

    Quote Originally Posted by whitecat View Post
    You should bench true games !
    http://www.phoronix.com/scan.php?pag...tem&px=MTUyNTU

  4. #4
    Join Date
    Jun 2010
    Location
    ฿ 16LDJ6Hrd1oN3nCoFL7BypHSEYL84ca1JR
    Posts
    1,039

    Default

    So this was mainline and not "glassy mesa" yet, I guess... I would have expected that, since both are developed by LunarG...

    Anyway, why is it that it is so far behind, even in opengl features, not only performance? Are they trying to find a way to reuse as much code as possible from intel's classic mesa driver?

  5. #5
    Join Date
    Aug 2011
    Posts
    514

    Default

    Quote Originally Posted by whitecat View Post
    You should bench true games !
    Actually, when benchmarking proprietary drivers, it is IMO best to use more unknown or niche games like FOSS shooters, because it's more likely that the drivers won't contain game/benchmark specific hacks and/or replacement shaders, so it's harder for the driver to cheat.

  6. #6
    Join Date
    Feb 2011
    Location
    France
    Posts
    189

    Default

    Quote Originally Posted by Ancurio View Post
    Actually, when benchmarking proprietary drivers, it is IMO best to use more unknown or niche games like FOSS shooters, because it's more likely that the drivers won't contain game/benchmark specific hacks and/or replacement shaders, so it's harder for the driver to cheat.
    I don't care about proprietary drivers. I'm talking about free drivers.

  7. #7
    Join Date
    Feb 2011
    Location
    France
    Posts
    189

    Default

    You benchmarked Open Arena, Reaction Quake 3, Unvanquished, Warsow and Xonotic.
    Why all these pretty useless games ?
    That would be much more intersting to only test Xonotic and Doom 3 (for instance).

  8. #8
    Join Date
    Feb 2008
    Posts
    867

    Default

    Quote Originally Posted by whitecat View Post
    You benchmarked Open Arena, Reaction Quake 3, Unvanquished, Warsow and Xonotic.
    Why all these pretty useless games ?
    That would be much more intersting to only test Xonotic and Doom 3 (for instance).
    Yep all those are id Tech 4 engine based games . Xonotic with different settings is enough to represent all those, then one id Tech 5 and one Source engine based, Cube 2, etc... .

    Actually i would like to see ioQuake, ioRTCW, ioXYZ... rend2 renderer is pretty decent (and low performer ) on those but different .
    Last edited by dungeon; 06-13-2014 at 09:44 AM.

  9. #9
    Join Date
    Dec 2011
    Posts
    145

    Default

    Quote Originally Posted by ChrisXY View Post
    So this was mainline and not "glassy mesa" yet, I guess... I would have expected that, since both are developed by LunarG...

    Anyway, why is it that it is so far behind, even in opengl features, not only performance? Are they trying to find a way to reuse as much code as possible from intel's classic mesa driver?
    Wondering the same thing...

  10. #10
    Join Date
    Jul 2013
    Posts
    204

    Default

    Quote Originally Posted by ChrisXY View Post
    So this was mainline and not "glassy mesa" yet, I guess... I would have expected that, since both are developed by LunarG...
    This has been in git for a while as its in good enough state for someone to build and use if they wanted to. As far as I understand glassy mesa is still in the research phase, LunarG are not even suggesting it should be merged yet. There are also other things that will need to be worked out with glassy mesa such as the extra dependencies it adds e.g glslang.

    Quote Originally Posted by ChrisXY View Post
    Anyway, why is it that it is so far behind, even in opengl features, not only performance? Are they trying to find a way to reuse as much code as possible from intel's classic mesa driver?
    I could be wrong but I believe the biggest obstacle in reusing code is the ilo shader backend needs to use tgsi ir which is different from the ir intel use for their shader backends.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •