Announcement

Collapse
No announcement yet.

The Interesting Tale Of AMD's FirePro Drivers

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

  • phoronix
    started a topic The Interesting Tale Of AMD's FirePro Drivers

    The Interesting Tale Of AMD's FirePro Drivers

    Phoronix: The Interesting Tale Of AMD's FirePro Drivers

    Earlier this week we published our annual look at AMD's Catalyst driver releases from the past year. Not only did the Catalyst Linux driver this year picked up a couple new features, its driver performance had improved slightly over the past twelve months. In building up some initial test data for OpenBenchmarking.org we decided not only to do these tests on the latest consumer-grade graphics card this year, but expand it to cover the workstation performance too and to go back nearly two years in time. These results for an AMD FirePro V8700 graphics card with the monthly driver updates going back to Catalyst 9.2 are quite interesting. AMD announced twice this year optimizations to their FirePro driver software, but in reality these "optimizations" were largely unsustainable and not optimizations as much as they were attempting to address driver regressions from the past.

    http://www.phoronix.com/vr.php?view=15592

  • indezign
    replied
    User should use the newer Firepro 8.773.1 and Catalyst 10.12 drivers with the newer Y2010-Y2011 cards. I don't think ATI's current drivers are OpenGL 4.1 compliant (just compatible) yet which help advanced developers push demos and graphics to help sell their software.

    Leave a comment:


  • b15hop
    replied
    Originally posted by dnebdal View Post
    1 min? That felt more like 30 sec.

    Anyway, what I wanted to add was that I noticed the Win7 part, but still felt it worth quoting since it seems to imply that this also applies on linux (for the subset of games that runs there).
    Yes it seems that nVidia win7 drivers are in parallel with *nix. Where any update in windows is almost released on par anywhere else. Yet compared to AMD this is still not the case, even though they've come a long way.

    Leave a comment:


  • dnebdal
    replied
    Originally posted by deanjo View Post
    Grr stupid 1 minute limit.
    1 min? That felt more like 30 sec.

    Anyway, what I wanted to add was that I noticed the Win7 part, but still felt it worth quoting since it seems to imply that this also applies on linux (for the subset of games that runs there).

    Leave a comment:


  • dnebdal
    replied
    Originally posted by deanjo View Post
    Grr stupid 1 minute limit.

    When was the last time you have seen a change log on linux drivers that said something like:

    - 17% increase in performance on Nexiuz on XYZ series cards
    - 22% increase in performance on GIMP on XYZ series cards
    - 37% increase in performance on ET:QW running in multicard setup on XYZ cards.
    Not long ago at all?
    Quoting the nvidia 260.89 linux driver relnotes (from 18.10.2010):
    Performance

    * Increases performance for GeForce GTX 400 Series GPUs in several PC games vs. v258.96 WHQL drivers. The following are examples of some of the most significant improvements measured on Windows 7. Results will vary depending on your GPU and system configuration:

    GeForce GTX 480:
    + Up to 10% in StarCraft II (2560x1600 4xAA/16xAF Ultra)
    + Up to 14% in S.T.A.L.K.E.R.: Call of Pripyat (1920x1200 4xAA/16xAF)
    + Up to 16% in S.T.A.L.K.E.R.: Call of Pripyat (SLI ? 2560x1600 4xAA/16xAF)
    + Up to 6% in Aliens vs. Predator (SLI ? 1920x1200 noAA ? Tessellation on)

    GeForce GTX 460:
    + Up to 19% in StarCraft II (SLI ? 1920x1200 4xAA/16xAF Ultra)
    + Up to 15% in Battlefield Bad Company 2 (SLI ? 2560x1600 4xAA/16xAF)
    + Up to 12% in S.T.A.L.K.E.R.: Call of Pripyat (2560x1600 noAA)
    + Up to 9% in Aliens vs. Predator (1680x1050 4xAA/16xAF ? Tessellation on)
    + Up to 7% in Metro 2033 (1680x1050 noAA ? Tessellation on)
    + Up to 11% in Dirt 2 (SLI ? 2560x1600 4xAA/16xAF)
    + Up to 12% in Crysis:Warhead (SLI ? 1920x1200 4xAA/16xAF Gamer)
    + Up to 13% in Far Cry 2 (2560x1600 4xAA/16xAF)
    + Up to 12% in H.A.W.X (SLI ? 1920x1200 4xAA/16xAF SSAO Very High)
    + Up to 5% in Just Cause 2 (1920x1200 4xAA/16xAF)
    + Up to 22% in Riddick: Assault on Dark Athena (1920x1200 noAA)
    + Up to 5% in 3DMark Vantage (Extreme Preset)

    Leave a comment:


  • b15hop
    replied
    Originally posted by smitty3268 View Post
    The general architecture of the ATI cards has been pretty similar since r600 was released, so i'm not surprised that the drivers were already tuned pretty well. The new DX11 features aren't being used in any of these tests, that would be where i would expect driver improvements to show up. The new 6000 series does finally switch to a new architecture, so it would be more interesting to track that and see if the performance picks up there or not. Same has been true of NVidia, using the same architecture for a long time up until a switch occurred with Fermi.
    Well my argument is that ATi might well have more average code for their binary blob than nVidia. Simply because nVidia works better most of the time in Linux. They usually have much better binary drivers. Faster too. When compared to ATi. Things have changed for ATi in the past 3 years but I still feel that nVidia is way in front regarding their portable code management etc... Again, I have no idea as to what I'm talking about. Even if I'm an AMD fanboy. x)

    Leave a comment:


  • smitty3268
    replied
    The general architecture of the ATI cards has been pretty similar since r600 was released, so i'm not surprised that the drivers were already tuned pretty well. The new DX11 features aren't being used in any of these tests, that would be where i would expect driver improvements to show up. The new 6000 series does finally switch to a new architecture, so it would be more interesting to track that and see if the performance picks up there or not. Same has been true of NVidia, using the same architecture for a long time up until a switch occurred with Fermi.

    Leave a comment:


  • deanjo
    replied
    Originally posted by energyman View Post
    you also don't see anything in the ati changelog that happens in the shared code base and benefits both sides. So if AMD tunes fpr a certain opengl app and that influences the linux side too you won't find it in the linux changelog.
    Actually IIRC, when they improved their 2D performance in windows many of those changes were brought across and mentioned in the linux release.

    Leave a comment:


  • energyman
    replied
    you also don't see anything in the ati changelog that happens in the shared code base and benefits both sides. So if AMD tunes fpr a certain opengl app and that influences the linux side too you won't find it in the linux changelog.

    Apart from that - what did Phoronix really test? Quake-based stuff, Specviewperf - but there is a lot more stuff.

    Leave a comment:


  • deanjo
    replied
    Originally posted by energyman View Post
    never, but when was the last time you saw something like:
    - added support for kernel 2.6.XY
    - added support for xorg-server version 1.uber-leet-experimental

    in windows drivers?
    That is adding support, not performance increases, and in the case of AMD's blob drivers even your examples are not that frequent.

    Leave a comment:

Working...
X