1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Multi Site HD7950 comparisons

Discussion in 'AMD / ATI' started by the54thvoid, Feb 1, 2012.

  1. the54thvoid

    the54thvoid

    Joined:
    Dec 14, 2009
    Messages:
    3,485 (1.89/day)
    Thanks Received:
    1,698
    Location:
    Glasgow - home of formal profanity
    This is not a flame thread and this is NOT a criticism of W1zzard in any shape or form whatsoever.

    The purpose of this thread is solely to illuminate the quite significant results from different reviewers of the HD7950 versus the GTX 580. I came across various reviews that state the HD 7950 is better than the 580 and I am very cool with that. TBH, if Kepler isn't here by March (and it's not as good) I'm probably going crossfired, watercooled and overclocked 7950's :rockout:

    I've looked at 5 sites, Anandtech, Guru3D, Hexus.net, Hardware Canucks and TPU and compared the 580 and 7950 results. I've noted down drivers used and in general settings are shown. Pretty sure all sites are i7 quad cores at high frequencies (no bottlenecks). Most resolutions are 1920x1200 or 1920x1080. (Not all sites do 2560 res).

    Using BF3* (again due to it's commonality) there are some major differences between reviewers. What interests me is what factors cause such differences and how it affects real world scenarios.

    * - One site has stated that BF3 causes probs for ATI cards for some reason. I also must point out that in most other games I believe the 7950 wins over the GTX 580 - (even in NV games).

    This is a technical question - it would be futile to spout personal opinions, it's simply of interest to me (and others?) how reviewers get such differing results.

    (And finally, I'm not interested in maths lessons, my fractions were Leader-Loser=Diff, Diff/Loser= result)

    [​IMG]
     

    Attached Files:

  2. Xaser04

    Xaser04

    Joined:
    May 15, 2007
    Messages:
    743 (0.27/day)
    Thanks Received:
    106
    Most variances in performance can be traced to the differences in how the benchmark is run.

    Levels played, settings used and benchmark timeframe will all have a drastic effect on overall performance.

    Alas some of the big review sites don't actually mention what section of the game they are testing on making comparisons very hard to make (even to your own tests).
     
  3. Delta6326

    Delta6326

    Joined:
    May 21, 2008
    Messages:
    3,843 (1.59/day)
    Thanks Received:
    670
    Location:
    Iowa, USA
    I would say you have to look at the Test Setups most likely all of these have different Ram, CPU, HD/SSD etc.
     
  4. m1dg3t

    m1dg3t

    Joined:
    May 22, 2010
    Messages:
    2,247 (1.34/day)
    Thanks Received:
    513
    Location:
    Canada
    That would be my guess. I think W1zz know's how to run a review....
     
  5. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    14,152 (4.44/day)
    Thanks Received:
    7,358
    Location:
    Edmonton, Alberta
    #1...not all sites use same drivers. This is a key point. You can only compare results using the same components, same drivers, and same tests. This has not happened here.

    #2...not all sites test BF3 in the same way, as there is no built-in benchmark. Could even be different levels.

    #3...not all systems have the same config.


    Otherwise, seems about normal to me. HWC and TPU use different test areas for BF3, yet the results are very similar(Although you choose HWC graphs with no AA, I might mention).

    Doesn't mean that any of these results are "wrong"...Some sites report avg, some use max, some use minimum, there are many things that are vaired here.

    One of the most important lesssons when studying "Statistics" in College/University...numbers can easily be manipulated, and still be valid. I can change motherboards, and get different 3D results, so something as simple as that can have a profound effect on the numbers.
     
  6. theoneandonlymrk

    theoneandonlymrk

    Joined:
    Mar 10, 2010
    Messages:
    3,414 (1.95/day)
    Thanks Received:
    573
    Location:
    Manchester uk
    guys bang on right i can get upto 80 fps with ultra everything ingame via some catalys tweeks and disabling by force vsync on 2x5800 ,yeh bit,, bit of tearing now and again i can also get 40-50 fps out of the same setup ,no cat tweeks but max quality set in catalyst, on 5800s and i assume all radeons if you go with defaults some settings are compromised/optimsed by catalyst for performance, nvidia do similar

    im about quality tho obv fps does make me smile, its all just more solid init

    so many configs and hardware options
     
  7. Wrigleyvillain

    Wrigleyvillain PTFO or GTFO

    Joined:
    Oct 13, 2007
    Messages:
    7,678 (2.92/day)
    Thanks Received:
    1,776
    Location:
    Chicago
    Yeah BF3 really needs a better, uniform benchmark method. And something that can gauge multiplayer too would be ideal as it's a different, more demanding beast.
     

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page