MacBook Pro (MBP) 13 inch RAM Bench Tests (XBench + Geekbench) 4GB, 6GB, & 8GB DDRAM

Discussion in 'MacBook Pro' started by NickZac, Jan 21, 2011.

  1. NickZac macrumors 68000

    NickZac

    Joined:
    Dec 11, 2010
    #1
    13 inch MacBook Pro (MBP) RAM Bench Test Results for: 4GB, 6GB, & 8GB DDRAM

    Background/Introduction
    I was bored and so I decided to test some RAM out. I am curious as to the difference in 4, 6, and 8GB. I also want to see how 4GB of Crucial RAM (1 stick) compares to the factory Apple RAM (2-sticks). Let’s take a closer look at this.
    Here are the specs for my computer:
    • 13 inch MacBook Pro, late 2009 Unibody, model 5,5
    • 2.53 Core 2 Duo CPU, NVIDIA GeForce 9400 GPU, 4GB RAM (from the factory, two 2GB sticks), RAM type: DDR3 at 1067Mhz
    • 120/8 GB OWC Mercury Extreme Pro Solid State Drive: about 80GB used and 40GB free
    • No other hardware modifications other than the SSD and RAM


    Methodology: Testing Procedure
    • Reconnect the computer to an electrical outlet and restart the computer
    • First test with Geekbench
    • Quit Geekbench
    • Test with XBench
    • Shut the computer down and disconnect the power cord
    • Change out the memory to a different amount
    • Start from the top again in the exact same format

    Results:

    4GB Apple Based RAM (two 2GB Sticks, both slots used with equal RAM)
    Total XBench score: 207.78

    • Memory Test: 192.47
    • System 240.19
    • Allocate 428.89 1.58 Malloc/sec
    • Fill 183.90 8941.80 MB/sec
    • Copy 211.83 4375.22 MB/sec
    • Stream 160.56
    • Copy 152.72 3154.28 MB/sec
    • Scale 151.60 3132.05 MB/sec
    • Add 170.43 3630.58 MB/sec
    • Triad 169.47 3625.42 MB/sec

    Total Geekbench score: 3564
    • Memory Performance: 2628
    • Memory Bandwith Performance: 1831


    4GB Crucial RAM (one 4GB stick in slot 0, other slot left empty)

    Total XBench Score: 198.40
    • Memory Test 165.87
    • System 231.25
    • Allocate 428.23 1.57 Malloc/sec
    • Fill 177.21 8616.30 MB/sec
    • Copy 200.21 4135.36 MB/sec
    • Stream 129.31
    • Copy 124.56 2572.80 MB/sec
    • Scale 125.81 2599.10 MB/sec
    • Add 134.48 2864.64 MB/sec
    • Triad 132.99 2844.96 MB/sec

    Total Geekbench score: 3509
    • Memory Performance: 2544
    • Memory Bandwith Performance: 1438


    6GB RAM (One 4GB Crucial stick and one 2GB Apple-based stick)
    • Total XBench score: 208.18
    • Memory Test 191.06
    • System 237.57
    • Allocate 434.44 1.60 Malloc/sec
    • Fill 183.84 8938.68 MB/sec
    • Copy 204.65 4227.02 MB/sec
    • Stream 159.77
    • Copy 152.72 3154.28 MB/sec
    • Scale 149.94 3097.78 MB/sec
    • Add 169.63 3613.38 MB/sec
    • Triad 168.83 3611.76 MB/sec

    Total Geekbench score: 3563
    • Memory Performance: 2629
    • Memory Bandwith Performance: 1822


    8GB RAM (two 4GB Crucial sticks in each slot)
    • Total XBench score: 210.52
    • Memory Test 190.52
    • System 238.56
    • Allocate 425.19 1.56 Malloc/sec
    • Fill 183.42 8918.48 MB/sec
    • Copy 209.58 4328.77 MB/sec
    • Stream 158.58
    • Copy 151.13 3121.63 MB/sec
    • Scale 148.45 3066.92 MB/sec
    • Add 169.00 3600.07 MB/sec
    • Triad 167.97 3593.32 MB/sec

    Total Geekbench score: 3561
    • Memory Performance: 2628
    • Memory Bandwith Performance: 1827



    Other Major Differences with XBench
    • My disk test speed was significantly slower with the one 4GB Crucial stick than the original 4GB, 6GB or 8GB (10-16 point difference on overall disk score)
    Other Major Differences with Geekbench
    • None of significant value

    Study Limitations

    • How much these figures mean is debatable
    • If these figures correspond to real world performance may not be very likely
    • Obviously, the 8GB set-up will have higher performance as more physical memory is used; I did these tests from a cold-startup with no other programs running
    • I have the older and slower, Core 2 Duo processor; the iX series may be affected differently
    • I initially conducted each test 3 times, but after getting consistent results, I stopped as it was too confusing to quantify
    • This is one test for only one computer; it may affect you completely differently

    Discussion (these conclusions are drawn from the results and so they may or may not affect real world usage)
    • RAM upgrades do not make a computer ‘faster’ as is commonly implied, but RAM can, however, make a computer much ‘faster’ if you are using most of your available (free) memory
    • Rating wise, using a single 4GB Crucial memory stick showed inferior performance to the two 2GB memory sticks included from the factory and it also affected my SSDs read/write speed negatively. Subsequently the possibility exists that two 2GB sticks may be a better choice than one 4GB stick for users going from 2GB to 4GB
    • 6GB works fine on this model MBP and may be a good upgrade for people unable to afford the 8GB. Although Apple recommends using matching size pairs, this did not seem to affect performance whatsoever
    • Two different brand-name RAM sticks can be used simultaneously
    • 8GB makes no major impact on performance when you are not using significant physical memory
    • RAM speed, type, and size allocation have the ability to impact the benchmarked speeds of other computer parts, both negatively and positively
    • 2GB of RAM was not tested and probably should be in the future for a baseline comparison
    • Future studies need to test RAM with a higher speed (1333Mhz) to the standard 1067Mhz used by the MBP to see if there is any performance advantage or disadvantage



    Overall, for users considering going from 4GB to 8GB in their MBP, if they currently are not using most to almost all of their physical memory, this may not be a worthwhile upgrade at the present time. With that said, 8GB of RAM can currently be purchased for under $100 making it one of the least expensive upgrades for the MacBook Pro.
     
  2. nephilim7 macrumors regular

    Joined:
    Jun 13, 2008
    #2
    These test results fall well within the bounds of expected deviation. You could run any of these tests multiple times and get similar results.
     
  3. NickZac thread starter macrumors 68000

    NickZac

    Joined:
    Dec 11, 2010
    #3
    Each time, each RAM setting showed very similar performance figures. Certain settings caused other parts of the test to score higher/lower, which indicates RAM has an affect on it. How significant that affect is however, is completely debatable as the above numbers really don't mean a lot when you go to say 8GB or RAM is THIS good; you can't fill in the THIS portion with 8GB of RAM is 3527 good, and to most people '3527 Geekbench speed rating' has no meaning whatsoever. Even when comparing to other online benchmark stats you have issues regarding validation and not knowing any of the circumstances your 'baselines' had. So yeah, there will be variation, but I still thought it would be neat to test them to see how they differ.
     
  4. nephilim7 macrumors regular

    Joined:
    Jun 13, 2008
    #4
    I guess I wasn't clear. I just ran geekbench 10 times changing nothing and got results with a delta larger than yours. I was standing on my head during the highest one. Draw your own conclusions.
     
  5. NickZac thread starter macrumors 68000

    NickZac

    Joined:
    Dec 11, 2010
    #5
    Hence why I used 2 different benchmark systems, described my methodology precisely, AND...

     
  6. jreuschl macrumors 6502

    Joined:
    Dec 31, 2008
    #6
    Computer chipsets generally support dual channel memory, so that when two sticks are used performance is enhanced.
     
  7. ryanrich macrumors regular

    Joined:
    Nov 16, 2010
    Location:
    Cape Town, South Africa
    #7
    Your tests show the same results as mine, especially regarding 6GB and 8GB configurations being almost identical in throughput so not worrying about mixing 4GB and 2GB SO-DIMM's.
     
  8. DustyLBottoms macrumors regular

    Joined:
    Sep 8, 2009
    #8
    I for one applaud the OP's interest in measuring what so many have quantified as "ZOMG 8GBS IS SOOOOOOOOOOOOOOOOOOOOOOOOOO FAST"

    Thank you, I enjoyed your thread.
     
  9. Messy macrumors 6502

    Joined:
    Sep 5, 2010
    #9
    I appreciate you've obviously put a fair bit of work into this, but im not really sure what you were hoping to achieve?

    Im fairly confident it'd be easy enough to predict these, well, predictable results without all that testing.
     
  10. NickZac thread starter macrumors 68000

    NickZac

    Joined:
    Dec 11, 2010
    #10
    Mainly because I was bored, happened to get ahold of two 2GB 'Apple' sticks (in this case which are Memorex brand), and to see if 8GB over 4GB has any real effect at basic, 'normal' tasks. Most computer companies advertise 'Make you computer faster with a RAM upgrade!", and I wanted to see if any truth existed to this. While subsequent tests when opening unusually RAM-heavy tasks did help performance, basic tests (on paper stats) had no effect capability wise with upgraded RAM, and even a slightly negative effect depending on how the RAM was configured. I find it faster after it has been on for days and many programs are opened, but when I am running my normal programs, little difference, if any, exists.

    The tests performed indicate that RAM upgrades will only make a difference if you actually need more RAM, and the need is determined by how much of your current RAM is being used as well as virtual memory stats. For many users, purchasing additional RAM may not be worthwhile, even with the low price. Many people, not necessarily here, claim that RAM increases performance, which is not entirely true if the above tests are at least somewhat correct. The above tests did have reliable results with subsequent re-tests, but determining validity is not possible with just one computer and only 2 benchmark programs.
     
  11. Messy macrumors 6502

    Joined:
    Sep 5, 2010
    #11
    Thanks NickZac, I probably didn't use the correct words in my reply.

    I can see how this may help people confused by a lot of the bumpf and claims made by certain computer manufacturers. Fair play for taking the time to do it.
     
  12. vincenz macrumors 601

    vincenz

    Joined:
    Oct 20, 2008
    #12
    I'm no expert with number-crunching when it comes to these tests, but it sounds like 8 GB is only worth it if you're either a) in video editing b) like to keep two dozen programs/tasks open at once or c) have $80 lying around that is itching to be spent.

    I think I'll wait until prices fall to ~60 before buying then, thanks OP. In the meantime, I'll keep salivating over SSDs....

    I need prices of OWC's 240 GB to drop before plunking down the cash. :(
     
  13. trigonometry macrumors 6502

    trigonometry

    Joined:
    Jun 19, 2010
    Location:
    South Carolina
    #13
    Thanks NickZac, you make some of the best posts!
     
  14. fibrizo macrumors 6502

    fibrizo

    Joined:
    Jan 23, 2009
    #14
    This basically proves what I knew of the hardware. Thanks for the numbers.

    2 sticks of ram runs in dual channel mode, so the 2 apple sticks have higher bandwidth than 1 4gb stick. But since most things don't stress the memory bandwidth, the score did not drop by much.

    Missmatched sticks run in flex dual channel mode for the first 2gb on each stick, which leads to a full bandwidth reading, as most likely the benchmark does not touch the memory exceeding 4gb, where it will run single channel.

    2x 4gb sticks increase latency a little due to having to address more memory space, which is a little slower.

    So the net result is, if you don't use more than 4gb of ram, it does nothing for performance to add more ram.
     
  15. mulo macrumors 68020

    mulo

    Joined:
    Aug 22, 2010
    Location:
    Behind you
    #15
    my cheap 8GB nuram
    Code:
    Memory Test	341.46	
    		System	409.45	
    			Allocate	866.11	3.18 Malloc/sec
    			Fill	231.20	11241.57 MB/sec
    			Copy	541.41	11182.53 MB/sec
    		Stream	292.84	
    			Copy	293.37	6059.49 MB/sec
    			Scale	288.60	5962.32 MB/sec
    			Add	297.90	6345.99 MB/sec
    			Triad	291.63	6238.62 MB/sec
    
     
  16. NickZac thread starter macrumors 68000

    NickZac

    Joined:
    Dec 11, 2010
    #16
    Thanks, you're making me blush :eek:


    That makes perfect sense based on the findings. I think tonite I am gonna try a single 2GB stick to see what that tells us. If I can find it, I'll run two 1GB sticks as well to see a difference. I think it would be even better to have comparison stats taken when programs are using enormous amounts of memory.



    I've always heard good things about Nuram?...your RAM ratings/speed are almost twice that of mine. I guess it is because you have a faster/more effecient CPU/GPU?
     

Share This Page