New barefeats Graphics card test.

Discussion in 'Mac Pro' started by macgoblin7950, Mar 25, 2013.

  1. macgoblin7950 macrumors member

    Joined:
    Mar 22, 2013
    #1
    http://www.barefeats.com/gpu7950c.html

    In the latest test Barefeats has a bunch of PC cards in mac OS X. But he does not say how hes powering them. Does anyone have a clue? Im especially interested in the GTX 690. :D
     
  2. GermanyChris macrumors 601

    GermanyChris

    Joined:
    Jul 3, 2011
    Location:
    Here
  3. lssmit02 macrumors 6502

    Joined:
    Mar 25, 2004
    #3
    I know, and it has me questioning whether to sell my Mac Pro + ACD setup and update to the iMac with it's nice display, rather than spending $479 on a card that is no better on computer missing USB 3, with an older monitor. I think I can sell my Mac Pro and monitor and cover most (if not all) of the cost of the update.
     
  4. GermanyChris macrumors 601

    GermanyChris

    Joined:
    Jul 3, 2011
    Location:
    Here
    #4

    thats your call but thats not a trade I'd make..
     
  5. SDDave2007 macrumors regular

    Joined:
    Apr 12, 2007
    #5
    All Mac Pro GPUs above were in a in a 'Mid 2010' Mac Pro 3.33GHz Hex-Core. The GeForce GTX 680MX is the only exception -- embedded in the 'Late 2012' (27") iMac 3.4GHz Core i7

    Second paragraph of the article
     
  6. DanielCoffey macrumors 65816

    DanielCoffey

    Joined:
    Nov 15, 2010
    Location:
    Edinburgh, UK
    #6
    Could still have needed external power though. That was not actually mentioned.
     
  7. brentsg macrumors 68040

    Joined:
    Oct 15, 2008
    #7
    The 690 is a dual GPU card though. I wonder if he's basically getting single GPU scores out of it.
     
  8. MacVidCards Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #8
    I do know that the 580 3GB Classified that he tested required external power.

    I can not speak for how he did others.

    Only that and the GTX690 and 7970 have plugs that aren't supported by the Dual 6 pins in a Mac Pro.

    The GTX570 2.5, the 7950, and 5870 all run on Dual 6pins, as does the Quadro 6000
     
  9. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #9
    I agree... if you've got an aging Mac Pro like I do, and you want a modern GPU and a new 27" display to go with it, you could buy this new 7950 and a 27" Cinema Display and be out of pocket about $1500. OR for only another $1000 you can get a really nicely equipped 27" iMac with a boost in performance all around along with USB 3 and TB. It's very tempting. :eek:
     
  10. dpny macrumors regular

    Joined:
    Jan 5, 2013
    #10
    That's the worst possible X-Plane test imaginable.
     
  11. xav8tor macrumors 6502a

    Joined:
    Mar 30, 2011
    #11
    You said it. Go back through previous tests. They're always changing and on a given test one GPU spanks the others, and on the next one, the results are completely different.
     
  12. dpny macrumors regular

    Joined:
    Jan 5, 2013
    #12
    For those out there who don't use X-Plane: all else being equal, the single biggest fps killer in X-Plane is object density. X-Plane has a very high object budget per frame and uses instancing to get it that high. Flying over the ocean means no objects and means the GPU isn't being used for much more than the shaders which make the water look reflective. Cloud density can impact frame rate, but without telling us their rendering settings, it's a useless test.

    I can tell you from personal experience that X-Plane will bring a 5870 to its knees.

    It's as if you tested mpg by driving cars on the freeway with no traffic and a tailwind.
     
  13. MacsRgr8 macrumors 604

    MacsRgr8

    Joined:
    Sep 8, 2002
    Location:
    The Netherlands
    #13
    Exactly.
    Real life testing of X-Plane 10 is hard to do.

    The default "game" with no added scenery won't test a good grfx card. Nowadays with auto-gen buildings it does more "stressing" than before without hardly any buildings, but nothing beats highly detailed payware scenery like EHAM or KJFK.

    One thing you can see with these tests is the relative performance vs. other cards. 72 FPS doens't say anything, but 50% more FPS than another card does.
     
  14. Umbongo macrumors 601

    Umbongo

    Joined:
    Sep 14, 2006
    Location:
    England
    #14
    He absolutely is getting single GPU scores. It's GTX 670 performance and I think this article could have done with some explanation of the differences between the 500 and 600 series and what the 690 is to give some context.
     
  15. tomvos, Mar 26, 2013
    Last edited: Mar 26, 2013

    tomvos macrumors 6502

    tomvos

    Joined:
    Jul 7, 2005
    Location:
    In the Nexus.
    #15
    I would like to see a standard Geforce GTX 680 in this benchmarks. Because most of the results of the GTX 690 seem pretty low compared to the GTX 680MX. While the 680MX is no slow GPU, it’s still lower specced than the desktop GTX 680. And the GTX 690 is two times the GTX 680.

    Yet in some tests the 680MX scored better than the 690. So I think that there is exactly zero benefit from the two GK104 GPUs on this graphics board. Maybe Mac OS X has no idea how to address the NF200 bridge chip on the GTX 690 properly … thus leading to severely lacking performance. :mad:

    Update: I run the heaven benchmark with the same settings as barefeats. My Mac Pro 5.1 (w3680) with a standard EVGA GTX 680 2GB scored 45 FPS.
     
  16. dpny macrumors regular

    Joined:
    Jan 5, 2013
    #16
    X-Plane has frame rate testing built into the app via a command line instruction which will easily compare across different GPUs and will test the driver for things like instancing, etc. It is a much better test than the Barefeats one, which is embarrassing.

    From the X-Plane forums, testing 10.8.3 drivers on a GTX 670 in a Mac Pro 4,1:

    Code:
    1
    FRAMERATE TEST: time=93.0, frames=4267, fps=45.88 (fps=42.27 with forced instancing)
    GPU LOAD: time=93.0, wait=3.7, load=4.0%
    1024x768 2.1 NVIDIA-8.10.33 304.10.65f01 (210/0)
    
    2
    FRAMERATE TEST: time=93.2, frames=2963, fps=31.78 (fps=27.15 with forced instancing)
    GPU LOAD: time=93.2, wait=2.6, load=2.8%
    1024x768 2.1 NVIDIA-8.10.33 304.10.65f01 (210/0)
    
    3
    FRAMERATE TEST: time=94.1, frames=2411, fps=25.62
    GPU LOAD: time=94.1, wait=2.3, load=2.4%
    1024x768 2.1 NVIDIA-8.10.33 304.10.65f01 (210/0)
    
    4
    FRAMERATE TEST: time=93.8, frames=2562, fps=27.33
    GPU LOAD: time=93.8, wait=2.3, load=2.5%
    1024x768 2.1 NVIDIA-8.10.33 304.10.65f01 (210/0)
    
    5
    FRAMERATE TEST: time=92.8, frames=5129, fps=55.28
    GPU LOAD: time=92.8, wait=5.2, load=5.6%
    1024x768 2.1 NVIDIA-8.10.33 304.10.65f01 (210/0)
    This is a much better test, as it tests the entire system and X-Plane is more often CPU-bound than GPU.

    I know I'm going on about a small point, but I think it's an important one. Don't use Barefeat's X-Plane GPU tests.
     

Share This Page