Announcement

Collapse
No announcement yet.

Single GPU Vs. SLI Benchmark (No difference)

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

  • Single GPU Vs. SLI Benchmark (No difference)

    Good afternoon,

    I benchmarked my PC with a single GPU both at stock CPU speeds and overclocked. The FPS were slightly higher in the 3D test after OC roughly by about 8 FPS.

    Then I installed my 2nd GPU in SLI mode and there was absolutely no difference in the final baseline score or in the 2D and 3D categories.

    I made sure according to the NVIDIA control panel that SLI was active.

    I monitored the temps of both cards and all is normal.

    My question is... does Passmark recognize SLI mode? Is it hit and miss depending on the system / drivers etc etc?

    Also, the only game I've had the opportunity to play is Tropico 3 and it seems that after the install of the second card, it began to "stick" or "hang".

    I will have to try and re-install the EVGA SLI patch for this game and also re-install the game itself to see if that helps.

    But either way, Passmark did not give a higher score for SLI mode.

    To assist, below are my specs:

    EVGA 790i FTW Motherboard
    (2) EVGA GTX 260-216 Superclocked GPU's
    Q9650 (3Ghz) overclocked to 3.4Ghz (377 FSB x 4 = 1511)
    8GB DDR3 Patriot Memory (1511 FSB)
    Memory and CPU Synced
    1000 Watt Antec PSU
    1.35 Volts to CPU
    Windows 7 64-bit

    Any assistance will be greatly appreciated.

  • #2
    All this sounds normal.

    Overclocking can have a small positive impact the the frame rate.

    To get SLI to work the application needs to be aware of, and optimised for SLI, and the video card device card device driver needs to be optimised for the application.

    So in the release notes for nVidia's device drivers you find comments like this,
    "New in Version 196.21: Adds SLI and multi-GPU support for many top new gaming titles including Avatar Demo, Battlefield: Bad Company 2, City Bus Simulator, Dirt 2..."

    So SLI only works for the small number of games that nVidia select.

    We did what we could to optimse for SLI, and then contacted nVidia, but they never bothered getting back to us.

    SLI also has a well know reputation for increasing the instability of a system.

    Comment


    • #3
      Thanks for the reply. I was hoping it was nothing in my settings or hardware configuration.

      Since this is a well known problem, I'm assuming it will take a joint effort by NVIDIA and software developers to find a comfortable medium.

      Thanks again and I hope I can contribute more in the future.

      Comment


      • #4
        Is this still the case Anno 2019? That Performance test has no difference between Single GPU vs SLI?

        Comment


        • #5
          SLI Situation in 2005
          https://www.passmark.com/forum/perfo...formance?t=234

          SLI Situation in 2007
          https://www.passmark.com/forum/perfo...=9308#post9308

          SLI Situation in 2012
          https://www.passmark.com/forum/perfo...3964#post23964

          SLI Situation in 2014
          https://www.passmark.com/forum/perfo...6179#post26179

          The V9 release in 2016 also added a DirectX12 test, which does benefit from SLI. Also the GPU compute test can use more than 1 GPU in V9. So you should see some overall benefit with SLI.
          But our position hasn't really changed much over the years. SLI was mostly a marketing gimmick to sell twice as many video cards. With the performance improvements being pretty minimal on average (which the exception of a few games that nVidia optimised for by hand, to help with their marketing).

          More background reading: Death of SLI?
          https://www.pcgamesn.com/amd-crossfire-vs-nvidia-sli




          Comment

          Working...
          X