GPU Benchmarking Score Variations

Discussion in 'Benchmarks' started by wheres_my_beard, Nov 1, 2010.

Share This Page

  1. wheres_my_beard

    wheres_my_beard New Member

    Joined:
    Oct 29, 2010
    Messages:
    10
    Likes Received:
    0
    How much variation should there be when testing without settings changing?

    The only reason I ask is that some graphics setting can lead to small gains, which others can result in larger ones.

    However if the scoring / performance fluctuates from one test to the next anyway, is it possible to get an accurate idea of these gains and loses?
     
  2. Digital Doctor

    Digital Doctor Loving AMD's new AM1 CPU

    Joined:
    Dec 6, 2009
    Messages:
    1,903
    Likes Received:
    0
    Thats just it, benchys like 3dmark don't really tell you much about real world gaming, i only use things like 3d mark etc. to test if the cards are working properly as it stresses them to the max.

    Vsync should be off in all cases of benchmarking if youre into that sort of thing, otherwise your fps will only go as high as your monitors refresh rate. (60Hz normally=60fps)
     
  3. Fat Jez

    Fat Jez Master of the High Five!

    Joined:
    Apr 15, 2003
    Messages:
    4,405
    Likes Received:
    0
    As DD says, software like 3DMark is only really of benefit on your own rig to track changes after tweaks. Most benchmarking by review sites is done on default settings for consistency. The differences between one pass and another (without tweaking) should only be around 1%, to allow for statistical errors.