Microsoft analyzes over a million PC failures, OC stability results

Discussion in 'Overclocking & Hardware' started by boron boy, Jun 28, 2012.

  1. boron boy

    boron boy Member

    Joined:
    Nov 30, 2004
    Messages:
    303
    Location:
    Melbourne
    http://www.extremetech.com/gaming/1...-pc-failures-results-shatter-enthusiast-myths

    Discuss.

    Edit: Ooops jumped straight to the forums and posted this before checking the main ocau site, apologies if I cut anyone's lunch.
     
    Last edited: Jun 28, 2012
  2. .Radiant

    .Radiant Member

    Joined:
    Nov 25, 2009
    Messages:
    4,737
    Location:
    brizbun
    AMD overclocking overall is a lot more fiddly than intel. You can get stable speeds at different configurations e.g something might be stable at x16@250 but not x20@200, even though they give the same freq they're fairly different. One might also requires less volts than the other, not to mention NB overclocking. Vendor A is more likely to be AMD because of that IMO. Depends on when the data was taken though.

    It'd be much more conclusive if ms gave more info about the study, at least the numbers were taken from machines with at least a 5% overclock.
     
  3. callan

    callan Member

    Joined:
    Aug 16, 2001
    Messages:
    4,664
    Location:
    melbourne
    Frankly, I don't see the point in anything less than a %10 overclock - %5 was a conservative threshold.
    I'm ashamed to say that I've not overclocked anything for quite a few years - I see little point. I DID have one of those lovely Celeron slot1 233Mhz processors that could run quite stably at 350Mhz (but BOY did it run HOT!).

    Underclocking - well I did that once on a 3Gig Prescott. By pulling it back to 2.9, I could significantly undervolt it. This kept it only scalding, as opposed to searing hot, and stopped the QBIC breadbin sounding like a harrier Jump Jet performing a vertical takeoff (Bloody hell those Prescotts pushed out heat - it's not a myth).

    Interesting that they rated Laptops as significantly MORE stable than PC's.
    I have a sneaking suspicion that may have to do with the proliferation of absolute crap power supplies (particularly in yum-cha whiteboxes), and poor or non-existent airflow (roominess in cases does NOT mean good airflow).

    The question about ECC memory, also, is interesting. Working as much as I did with server gear, I'd use ECC memory in any system I built (particularly with the large amount of RAM in todays machines.) It would probably push me into essentially building a "server-as-PC" - but that's not a bad thing.

    Callan
     
  4. PAiZley

    PAiZley Member

    Joined:
    Nov 18, 2010
    Messages:
    191
    Location:
    Melbourne
    when your finding your perfect overclock, you crash all the time, but i dont think these results are true for when your all done and stable.
    ive had my cpu overclocked for nearly 2 years now, and it never crashes.

    alot of the crash reports are probly when you first start overclocking and have not too much of an idea what your doing... your machine is crashing every half hour.

    Disk reliability is same on laptops ???? are these laptops that never get moved around while they are turned on ? hard disks are generally the first to go in laptops imo.
     

Share This Page