Getting higher than expected Geekbench results 27" 2017 3.4 i5

Discussion in 'iMac' started by AAtte, May 16, 2019.

  1. AAtte, May 16, 2019
    Last edited: May 30, 2019

    AAtte macrumors newbie

    Joined:
    Jun 4, 2014
    #1
    Hello,

    I'm getting over 5000 from single core and around 14500 from multi core. The Mac Geekbench site says that this model gets only 4753 from single and 13768 from multi.

    My results are what the 3.5 model should get.

    Here are my results: https://browser.geekbench.com/v4/cpu/13158985

    It's still bone stock.

    Is my iMac "broken"? ;)

    Edit: Now the points went to 5123 / 15102 as I upped the ram to 16gb.
     
  2. chabig macrumors 603

    Joined:
    Sep 6, 2002
    #2
    Sounds like you're getting what Geekbench says you should.
     
  3. AAtte thread starter macrumors newbie

    Joined:
    Jun 4, 2014
    #3
  4. chabig macrumors 603

    Joined:
    Sep 6, 2002
    #4
    First, you never told us what machine you have. You only said that Geekbench results for your model averaged 4753/13768. Then you said your specific machine reported 5000/14500, which is within 5% of the averages. So your machine is pretty close to average, as one would expect since they are all made the same.
     
  5. trsblader macrumors 6502

    Joined:
    May 20, 2011
    #5
    Actually, if you do a search, there's dozens of pages of results of people with your setup all getting the same scores as you.
     
  6. casperes1996 macrumors 68040

    casperes1996

    Joined:
    Jan 26, 2014
    Location:
    Horsens, Denmark
    #6
    This is part of the chip lottery. There's a margin of error both within the testing, and from chip to chip. When so many people report scores, some might be running a lot of background tasks that draw down the average, but there's also natural variation from chip to chip, even within the same exact model of chip.
    Some chips don't need as much power to run at a specific frequency and can thus boost more or just maintain lower heat. This does rely on a chip that can dynamically set voltage based on need, and it's not always as flexible as the CPU perhaps could've allowed, but they need to make sure the less ideal chips in the range are still stable.
     
  7. AAtte thread starter macrumors newbie

    Joined:
    Jun 4, 2014
    #7
    You didn't read the headline? It says 3.4 i5 27" 2017. I didn't know that 5% difference is normal. I'm happy with my +5%. :D
     
  8. benshive macrumors regular

    Joined:
    Feb 26, 2017
    Location:
    United States
  9. AAtte thread starter macrumors newbie

    Joined:
    Jun 4, 2014
    #9
    Mine also.
     
  10. casperes1996 macrumors 68040

    casperes1996

    Joined:
    Jan 26, 2014
    Location:
    Horsens, Denmark
    #10
    It's important to keep in mind that while the GeekBench team does their best to make tests comparable, they aren't across different platforms; It's kinda impossible. There will be a difference in background tasks running between the systems for one thing. But also the app model is very different accross the two; They'll be implemented with different frameworks, probably different compilers.
    Furthermore, GeekBench is way more a test of burst performance than it is sustained performance.

    I am not trying to downplay the impressive performance of the modern A-chips from Apple. They are truly remarkable. But they don't cheat physics either, and the power budget for them is not high enough to compete with a desktop chip in sustained threaded tasks
     
  11. AAtte thread starter macrumors newbie

    Joined:
    Jun 4, 2014
    #11
    Now the points went to 5123 / 15102 as I upped the ram to 16gb.
     
  12. Zdigital2015 macrumors 65816

    Joined:
    Jul 14, 2015
    Location:
    East Coast, United States
    #12
    Yes...that too is to be expected. More DRAM will increase the score and a faster storage system, say SSD instead of a Fusion Drive, will have some bearing on the final averaged numbers when the CPU test finishes. You will also see some variances due to the version of macOS you are running. For example, the GTX 780M in my 2013 27" iMac has considerably higher Metal scores under macOS Mojave that it ever did under macOS Sierra.
     

Share This Page

11 May 16, 2019