The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
Problems? See this thread at archive.org.

    How to stress test the reliability of memory modules on a laptop

    Discussion in 'Hardware Components and Aftermarket Upgrades' started by sniper_sung, Oct 25, 2009.

  1. sniper_sung

    sniper_sung Notebook Evangelist

    Reputations:
    66
    Messages:
    611
    Likes Received:
    0
    Trophy Points:
    30
    Just tried Memtest86+. Not sure whether it supports more than 4GB. The name "86" reminds me about the 32-bit OS problem: usually it cannot address more than 3GB (both Windows and Ubuntu), and each single process can only utilize up to 2GB (without the /3gb switch).

    It's the first time that I've realized 1066MHz is way too slow for 8GB capacity. It takes about 2 hours to finish a pass. On previous setups (e.g. 2GB 533MHz, or 1GB 333MHz) it takes only about 1 hour to finish a pass.

    I have spent 18 hours of memtest86+ on these 8GB and it looks like it didn't return any error during all the 9 passes. What's the expectation of number of errors the memory will encounter within one week, if I ever plan to use my laptop for a 7x24 computation? (AC adapter plugged + battery loaded, with only one core at 100% load, i.e. 50% load for the P9600, at a temperature of about 65 degrees centigrade)
     
    sniper_sung, Oct 25, 2009
    #1
  2. newsposter

    newsposter Notebook Virtuoso

    Reputations:
    801
    Messages:
    3,881
    Likes Received:
    0
    Trophy Points:
    105
    Did you give a glimmer of thought to the possibility that a 'longer pass' on 8Gb of ram might be might more thorough than a shorter pass on 2 Gb of ram?
     
    newsposter, Oct 25, 2009
    #2
  3. sniper_sung

    sniper_sung Notebook Evangelist

    Reputations:
    66
    Messages:
    611
    Likes Received:
    0
    Trophy Points:
    30
    Do you mean a "longer pass" gives more confidence than a "shorter pass"? I didn't do statistics and I don't have much idea about MTBF... :confused:
     
    sniper_sung, Oct 25, 2009
    #3
  4. moral hazard

    moral hazard Notebook Nobel Laureate

    Reputations:
    2,779
    Messages:
    7,957
    Likes Received:
    87
    Trophy Points:
    216
    I did do stat in uni, but even I dont know what MTBF is :confused:
    Is is mean time before failure?

    I guess I could get out minitab and draw you a few graphs, but you need to give me lots of data from simmilar ram and how many errors it had.
     
    moral hazard, Oct 25, 2009
    #4
  5. Trottel

    Trottel Notebook Virtuoso

    Reputations:
    828
    Messages:
    2,303
    Likes Received:
    0
    Trophy Points:
    0
    MTBF is meaningless to consumers. It is the mean time before the first failure for a large group of the hardware, and the time is the aggregate time of the whole collection.

    The pass takes a lot longer because it is a lot more memory. 1066Mhz is not way too slow for any amount of memory. Quantity and frequency are different things and not interrelated. Obviously with a static amount of bandwidth it is going to take twice the time to transfer 8GB than it will take to transfer 4GB.

    You can't expect anyone on earth to tell you that.
     
    Trottel, Oct 26, 2009
    #5
  6. o8x8

    o8x8 Notebook Enthusiast

    Reputations:
    35
    Messages:
    46
    Likes Received:
    1
    Trophy Points:
    15
    o8x8, Oct 26, 2009
    #6
  7. o8x8

    o8x8 Notebook Enthusiast

    Reputations:
    35
    Messages:
    46
    Likes Received:
    1
    Trophy Points:
    15
    It's possible to compile kernel in live cd:
    sudo su
    apt-get install linux-source build-essential
    cd /usr/src
    tar jxf linux-source*bz2
    cd linux-source*
    cp /boot/config-* .config
    make oldconfig ; make bzImage -j4 ; make modules -j4

    But this filled up my live cd session's ramdisk...
     
    o8x8, Oct 27, 2009
    #7