(Some results) What's the fastest stable memory for an early 2011 13" i7 MBPro?

Discussion in 'MacBook Pro' started by adb1973, May 11, 2013.

  1. adb1973, May 11, 2013
    Last edited: May 14, 2013

    adb1973 macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #1
    Hi folks,

    After reading numerous posts and many (many!) contradicting stories and facts I just thought, let's post my question on this forum.

    The basic question is simple:
    -What's the absolute maximum speed memory working STABLE on an early 2011 MBP 13-inch with an i7-2620M) processor. (i7 dual core 2.7GHz).

    The extra question: Is it true that the somewhat slow IGP (Intel HD3000) benefits from the higher spec'd memory (if those are indeed recognized and work @1600 or 1876)

    Facts:
    - The Macrumors Buying Guide, Intel-ARK and Apple say PC10600 1333 (CL9) is the only thing possible.
    - Geekbench2 shows a total of 7500 results for ddr3-1333 memory, 375 for ddr3-1600 en 38 for ddr3-1867 with the same bios and mainbord as I have.

    My estimated guess: roughly 4% of the MBP2011-13" i7 users, successfully uses higher spec'd ram.
    So in contradiction to what Apple en Intel state, it IS possible, is it? (We've all seen that before too)

    Big remaining question: If at least 400 people did it, what are their experiences... and what are the brands/models/specs they used. That something GeekBench doesn't tell.

    Since the first question raised is often:
    Why not stick to the advised-certified-guaranteed 1333 CL9 Apple tested-approved-kissed-by-steve-jobs memory? I'll answer that one right now.

    1. because i've seen in GeekBench it's possible and it gives higher benchmarks, realistically al least 4%
    2. realworld performance is better too since the intel HD3000 is heavily depending on memory bandwidth and could use some help.
    3. because higher spec'd memory is sometimes even cheaper than lower spec'd stuff.
    4. because i'm stubborn

    Thanks in advance for your (success?) stories, warnings and knowledge!

    Arjan
     
  2. negativzero macrumors 6502a

    Joined:
    Jul 19, 2011
    #2
    Those are just benchmarks. You won't really experience Amy real-world difference with the faster ram.

    Besides the 1333MHz ram is rated at CL9 whole the other 2 are rated at higher latency CL11.
    http://www.tomshardware.com/forum/314892-30-cl11-what-difference
    Comparisons can go on forever but it's just not justifiable going with the higher speed ram.
     
  3. johnnnw macrumors 65816

    johnnnw

    Joined:
    Feb 7, 2013
    #3
    If you use faster RAM and it happens to work it will just be throttled back anyway.

    Don't bother, just go with what's recommended and avoid problems.
     
  4. adb1973, May 12, 2013
    Last edited: May 12, 2013

    adb1973 thread starter macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #4
    @Johnnnw that's not true, GeekBench shows memory in this machine is NOT always clocked back to 1333. In fact hundreds of people use 1600 memory or even 1876 in this machine.

    Which apart from GB results has real world advantages for the IGP, since it shares internal memory (like over clocking memory would for a dedicated GPU)

    Anyone wants to share his EXPERIENCES with higher clocked ram in this machine?

    Did or Do you use 1333 with a faster CL than 9
    DDR3 1600 with a 9,10 or 11 (the 'special' Mac modules use 11)
    Even DDR 1867? Is it stable?

    Thx! Arjan
     
  5. Mr. Retrofire macrumors 601

    Mr. Retrofire

    Joined:
    Mar 2, 2010
    Location:
    www.emiliana.cl/en
    #5
    You can buy cheap Mac-certified memory from Kingston (Elpida & others) or Crucial (Micron). No problem.

    I used this RAM
    CT102464BF1339
    in my 2.2 GHz, Quad-Core, Early-2011 17" MBP. Comments on crucial.com and other sources recommended this RAM. Cost in March 2013 for 2x8 GB: ≈ 106 US$ (not via crucial.com). Tested with AHT (the long RAM test).

    The module specification says 1.35 V and 1.5 V. If this RAM runs with 1.35 V (Ivy Bridge or newer), then this is a sign that Micron uses more energy efficient RAM chips for these modules. And btw. 1333 MHz CL9 RAM runs cooler and provides a longer battery life compared to faster RAM.

    Do you use your machine for benchmarks or for real work!?

    I doubt that. If the HD 3000 is slow for you, then faster RAM does not make the calculations within the IGP faster. And, AFAIK Apples GPU drivers are not really optimized.

    Good excuse. ;-)

    I agree. :p
     
  6. Mr. Retrofire macrumors 601

    Mr. Retrofire

    Joined:
    Mar 2, 2010
    Location:
    www.emiliana.cl/en
    #6
    The higher CL for faster RAM is necessary, because a lower CL increases the error rate on the RAM module.

    Higher error rate = SLOWER RAM

    If you want stability AND faster RAM, buy the 1600 MHz/CL11 versions. They should be cheap, because Ivy Bridge uses the same RAM.
     
  7. cube macrumors G5

    Joined:
    May 10, 2004
    #7
    I use the 1600MHz CL9 2x8GB Kingston HyperX kit on my early 2011.

    I used the 1333MHz CL7 2x4GB Corsair Vengeance kit on my early 2009.

    Be sure not not buy XMP versions.

    No prob.
     
  8. AppleMacFinder macrumors 6502a

    AppleMacFinder

    Joined:
    Dec 7, 2009
    #8
    1600 MHz CL11 has the same performance as 1333 MHz CL9.

    Although 1600 MHz CL9 will be stable in your notebook,
    it will work at slower speed - 1333 MHz CL9. However,
    if you want a future-proof RAM, go for 1600 MHz CL9
     
  9. cube macrumors G5

    Joined:
    May 10, 2004
    #9
    My early 2011 shows my RAM as 1600MHz in "About this Mac".
     
  10. adb1973 thread starter macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #10
    Thanks all for your answers so far...

    @ Mr RetroFire: the real work I do on this MBP is some ad hoc FCP 7 and X editing if my iMac or Company-sets are not available ( I know it's not ideal)

    Thanks for the info on battery life and the non-optimized Apple drivers for the HD3000.

    The i7-2620M I have IS Sandy Bridge AFAIK, isn't it?

    @cube thanks for your answers, this is what i assumed, but couldn't get confirmed. Can you verify that you use the same 13 inch MBP as I do? Or do you have a 15 or 17 inch?
     
  11. cube macrumors G5

    Joined:
    May 10, 2004
  12. adb1973, May 12, 2013
    Last edited: May 12, 2013

    adb1973 thread starter macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #12
    Okay thanks, I own a 13" according to this MacRumors guide your 2011 machine indeed is fully capable of using the 1600 ram (in contradiction to what Apple states)

    The 13" isn't, but then again these GeekBench results show otherwise...

    For those who want to know, here is the AnandTech article about memory in an system with an Integrated Graphics Processor (in fact the HD4000 that's in newer MBP's).

    I ordered 1600 memory today and I'll keep you posted!
     
  13. cube macrumors G5

    Joined:
    May 10, 2004
    #13
    It seems the RAM table for the original unibody models was removed from the guide

    The Vengeance was there.
     
  14. adb1973 thread starter macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #14
    In this thread ATOMAC tried to give his Mac Mini a speed boost, basically what i try to do for my 2011 MacBook Pro.

    His Mac Mini has the same 2620M officially not capable of using anything higher than 1333MHz memory.

    He used:
    8GB Corsair Vengeance 1600, and
    16GB Kingston HyperX 1866/1867 memory

    Both were recognized and were working at there advertised speed.
    He seemed to have some trouble with the Vengeance but not with the Kinston

    Benchmarks and real life performance were a lot better than stock too

    (BTW: i've chosen an 256 Crucial M4 for my MBP)
     
  15. cube macrumors G5

    Joined:
    May 10, 2004
    #15
    The Vengeance for the original unibody is the 1333 CL7, not the 1600.
     
  16. adb1973 thread starter macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #16
    First results for installing:

    8GB DDR3 CL9 Corsair Vengeance SODIMM memory

    Recognized immediately, running @ 1600 MHz no beeps, heat problems or kernel panics (so far).

    https://docs.google.com/file/d/0B13VWIqxyQpOeDBqM2dHaHZTTkU/edit?usp=sharing
    https://docs.google.com/file/d/0B13VWIqxyQpOTldYQjNqV2R5Skk/edit?usp=sharing

    Used memory type:
    - CMSX8GX3M2A1600C9 (2x4GB kit)

    Geekbench score 7764 against 7564 with stock 4GB 1333 Samsung ram, 2,64% gain

    https://docs.google.com/file/d/0B13VWIqxyQpOeGtKOUtSOC1UbTA/edit?usp=sharing

    XBench results 2,46% lower see links, 2 remarkable facts:
    - CPU faster (about 5%)
    - Memory faster (especially streaming speed, in total 8%)
    - Quarts Graphics faster 1%
    - OpenGL much slower (about 10%)
    - User Interface a little slower 3,75%

    https://docs.google.com/file/d/0B13VWIqxyQpOWnlOb3NaczVtTVU/edit?usp=sharing

    Graphic Benchmark (non synthetic) Unigine Heaven 4 (medium settings):
    Horribly slow (n/y tested with stock ram, but i don't think that's much better, results anyone?)

    https://docs.google.com/file/d/0B13VWIqxyQpOZ3dxOTdPekNOSTA/edit?usp=sharing
     
  17. cube macrumors G5

    Joined:
    May 10, 2004
  18. adb1973, May 14, 2013
    Last edited: May 15, 2013

    adb1973 thread starter macrumors newbie

    adb1973

    Joined:
    Aug 25, 2008
    Location:
    Amsterdam
    #18
    i did and i tested and retested:
    - 16 GB corsair valueram 1333 c9
    - 4 GB samsung stock 1333 c9
    - 8 GB corsair Vengeance 1600 c9 (9-9-9-24)

    https://docs.google.com/file/d/0B13VWIqxyQpOeE5KR0FUYnRrNUk/edit?usp=sharing
    https://docs.google.com/file/d/0B13VWIqxyQpOOUhPQW1JRXJ6ZjA/edit?usp=sharing

    Basically using 8GB Vengeance gives a rock solid 2,9% improvement on framerate :) duhh
    and about 11,5 % raise over the GeekBench performance (12,4% if one uses the 6741 GeekBench score for the original i7 model, mine has 256 m4 SSD)


    I really hoped to get some better graphic performance but that didn't work out too great...

    Probably Mr. Retrofire has a point here, or hasn't he? I also got the advice to search for memory with CL9 (more specific 9-9-9-24) timings...

    I've read @Intel that the HD3000 has a max memory speed of 1300 MHz so maybe that cause some conflicts indeed.

    I really would love to see results for 1333 faster than CL9, 1600 CL11, 1866 and/or 2133 memory if available
     
  19. cube macrumors G5

    Joined:
    May 10, 2004
    #19
    The latency needs to be increased at higher frequency if the memory is of similar grade.

    You make modules with less latency by using higher quality chips.
     
  20. AppleMacFinder macrumors 6502a

    AppleMacFinder

    Joined:
    Dec 7, 2009
    #20
    You did right: in terms of performance, 1600 MHz CL11 is the same as 1333 MHz CL9. Meanwhile,
    1600 MHz CL9 is faster, and CL9 is the most compatible RAM for the current generation of Macs
    (unfortunately, sometimes people have kernel panics with CL7 or CL11. but it depends on RAM)
     
  21. cube macrumors G5

    Joined:
    May 10, 2004

Share This Page