RAM Upgrade- Which Slot For Best Result and Why My Score is Lower After Upgrading

Discussion in 'iMac' started by arjuna.angkasa, Jul 10, 2014.

  1. arjuna.angkasa, Jul 10, 2014
    Last edited: Jul 10, 2014

    arjuna.angkasa macrumors newbie

    Joined:
    Jul 10, 2014
    #1
    The Cause
    Hye All, So i've decided to upgrade my Ram for a better performance and speed on my Imac. I mainly use my workstation for editing purpose, heavily on Adobe Photoshop and Lightroom , After Effect& Premier.

    The Plan
    From total (2x2x2x2) 8GB to (2x2x8x8) 20GB.

    The Upgrade
    the ram.jpg
    Default Ram : 2GB Samsung DDR3 x2pcs
    New Additional Ram: 8GB Kingston KTA DDR3 pc1333 x2pcs

    The Issue
    I have spent my whole night googling , read forums to determine which combination work best , how to ensure the best performance by configuring the Dual Channel setup. Unfortunately, after too much reading, i didnt get what i was looking for. In general , i found 4 major suggestion- we have to put the ram size group, according to the :-

    1)Bank ( bank 0 & 1 )
    2)Dimm ( dimm 0 & 1)
    3)original empty slot
    4)Move default ram to empty slot and put the new one to the default slot

    By far (July 11,2014) , This is the most complete ( read:the most intense) Thread ive found >> CLICK HERE

    After finished reading to the last 5 pages, i still didnt find the answer i was looking for, So i decide to waste my sleeping time, try all the suggestion, and see what work best.

    IMPORTANT
    This only represent my own experiment and the Result MAY NOT be the same as everyone else. Please don't use this as your guide, as the test may lack all the necessary steps to ensure consistency between the result.

    Benchmarking Tools
    I use Xbench and GeekBench as my benchmarking tools

    MY PROCEDURE
    1-Replace Ram & Switch On
    2-Open System Report & Screen Cap
    3-Run GeekBench , Screen Capt Result , Close
    4-Run Xbench , Screen Capt, Close
    5-I REPEAT STEP 3 & 4 alternately for at least 3 times, and choose the best score as my final guide.
    6-Shut down and repeat step 1


    THE SETUP

    RAM SLOT GUIDE: TOP LEFT - TOP RIGHT - BOTTOM LEFT -BOTTOM RIGHT

    A- NO PICTURE ( Original default ram of 2-2-2-2 ; 8GB )


    B- 2-2-8-8 SETUP B 2-2-8-8.jpg


    C- 2-8-2-8 SETUP C 2-8-2-8.jpg


    D- 8-2-8-2 SETUP D 8-2-8-2.jpg


    E- 8-8-2-2 SETUP E 8-8-2-2.jpg


    CONTINUE BELOW
     
  2. arjuna.angkasa, Jul 10, 2014
    Last edited: Jul 10, 2014

    arjuna.angkasa thread starter macrumors newbie

    Joined:
    Jul 10, 2014
    #2
    CONTINUE ..

    THE RESULT
    I have to be honest with you,I am quite surprised and a bit disappointed with the score result after upgrading, and it makes no sense for me as the result after upgrading from 8GB to 20GB, every combination of 20GB make a JUST A TINY NUMBERS , VS the Default 8GB benchmark score.

    Some people said that Benchmarking are waste of time and we cant rely on the score. But at some point i do believe that at least, just at least-i could achieve at least double or slightly better score ;considering the amount of money & time i spent for this simple upgrade.

    So here is the result.

    1) GEEKBENCH SCORE
    RAM BANK SLOT TEST-GEEKBENCH1.jpeg



    2) GEEKBENCH SCORE (MEMORY)
    RAM BANK SLOT TEST-GEEKBENCH2.jpeg


    3) XBENCH SCORE
    RAM BANK SLOT TEST-XBENCH.jpeg


    THE VERDICT

    As you can see here, The default 8GB Ram giving a better score than all the 20GB setup. ( CAN SOMEONE PLEASE, ENLIGHTEN ME, WHY ? :mad: )

    And for the best combination of 2 pcs 8Gb and 2 pcs 2GB for a total of 20GB Memory, the SETUP B return the highest score on both GeekBench and Xbench.

    My personal conclusion is -
    Leave the default ram As is ( oh btw , my default ram as shown below- ORIGINALLY 2PCS 2GB, that afterward i add another 2pcs 2gb few years ago )

    DEFAULT SLOT.jpg

    and put the new stick to the default empty slot. [2 - 2 - 8 - 8]


    phew. So thats my current setup for the time being. So please, can somebody explain WHY this setup work best for my imac, and why the score is lower than original 8GB .


    with that, THANK YOU VERY MUCH. :)
     
  3. nebo1ss macrumors 68030

    Joined:
    Jun 2, 2010
    #3
    Try the B configuration but remove the two 2 Gig memory blocks so that you only have 16 Gig in the machine. See what results you get.
     
  4. arjuna.angkasa thread starter macrumors newbie

    Joined:
    Jul 10, 2014
    #4

    Oh yea, never thought of that before. i'll do that and update the result later. Thanks :)
     
  5. DeltaMac macrumors 604

    DeltaMac

    Joined:
    Jul 30, 2003
    Location:
    Delaware
    #5
    My own take on everything that you have reported, and, in particular, why I think some of your results appear slightly different than some of your other results:
    NSS (Not Statistically Significant). Real-world use will show little difference between 8GB and 20GB - unless you are doing specific tasks that need more memory, and then you will have better performance with more memory installed. (probably)
     
  6. casperes1996 macrumors 65816

    casperes1996

    Joined:
    Jan 26, 2014
    Location:
    Horsens, Denmark
    #6
    So here's the answer.



    The answer is actually remarkably simple.
    Both GeekBench and Xbench use very little RAM to begin with. As a matter of fact, most of their work fits inside the processor cache. This means that the amount of RAM you have won't affect those two benchmarking applications very much. There are other RAM stress tests available for this, but to be honest, they don't really work that well either. Amount of RAM only matters of you use it. If you only use 2 gigs, you'll get no speed increase from going to 32 (although OS X does archive caching in RAM, so it's not entirely true, but almost is).
    Your score in Geek- and Xbench is also likely to go down by adding more RAM on the iMac (by very very little), even though the real world performance could dramatically increase, because the CPU runs the memory on a dual channel config. I.e. each memory stick gets full access when there are only two, but has to share when there is four.
    In real world testing, you'll probably see performance increases, despite the scores reported by your benchmarking applications.

    GeekBench does test RAM, but the tests run are more about speed and latency than capacity.

    Hope this helped!
     

Share This Page