1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

P-M Dothan undeclocking and undervolting

Discussion in 'Overclocking' started by giorgio.tani, Oct 20, 2005.

  1. giorgio.tani

    giorgio.tani Guest

    Hi, I'm running since some weeks my Centrino notebook (the CPU is a
    1,7GHz P-M Dothan) at 1GHz with the voltage of 0,988 hinted for the
    600MHz clock for this CPU.
    I've simply undeclocked and undervolted it via software with Notebook
    Harware Control (former "Centrino Harware Control") without customizing
    anything on the hardware. It passed the short and the extensive CPU
    test of the NHC software and run without problem for further hours with
    CPU at 100% of load running some of my softwares. I tested it with the
    NHC extensive CPU test up to 1,4 GHz without any problems with 0,988V!
    The system freezed only when I tryied to run @1,7GHz, full speed, with
    0,988V.
    I think it's quite impressive that the (not customized) system passed
    extensive test at 1,4 GHz with the voltage intended for operating at
    600MHz (I was impressed also when firstly I was able to use that
    voltage @1GHz!), reducing significantly the power usage and thermal
    dissipation, it's that a common situation or is my CPU very lucky?
     
    giorgio.tani, Oct 20, 2005
    #1
    1. Advertisements

  2. One can almost always fine tune a specific system, under a specific set of
    condition, to exceed manufacturer's specifications.

    The reason is, the manufacturer has to specify the processor under all
    operating conditions and over all possible manufacturing tolerances, plus
    life span.

    For example, try running your test at maximum ambient temperature with the
    processor core at maximum specified core temp with all voltages (I.E. 5V,
    3.3V, 12V) at each combination of min/max and with the maximum allowable
    ripple/noise component on the power rails. (There's more, such as clock
    accuracy/jitter, northbridge timing skews, etc. but that's a start to
    illustrate the point)

    Then consider that your particular processor might be skewed to one or the
    other end of the manufacturing tolerances, that your motherboard might be
    skewed as well (better/worse noise, voltage tolerance, etc.), and then
    consider one usually allows a few percent, or so, of 'extra' as a safety
    margin.

    People don't generally bother with such considerations and are then
    surprised that at 25C it runs 'faster'. Then they're often surprised again
    when it locks up during the summer, or some other condition different than
    when they 'tested' it.

    It can be done, and be quite workable, but it usually means you've made one
    or more tradeoffs on the operating conditions whether you know it or not
    and that's one reason why overclockers are always looking for a better
    heatsink to keep temperature lower: the allowable operating temp is lower
    at the higher clock speeds. In your case the maximum ambient it'll still
    work in is probably lower but the throttling mechanism to keep it running
    if it gets hot may be miscalibrated now and not work right since it'll
    likely lock up at a lower temp before it activates.

    Still, if you don't run heavy duty apps for extended periods and are not
    generally out of an air-conditioned environment then the extra risk might
    be worth the benefit of longer battery time. Just be aware that the
    'bought' specifications are no longer your modified operating specifications.
     
    David Maynard, Oct 20, 2005
    #2
    1. Advertisements

  3. giorgio.tani

    Spajky Guest

    IMHO it is quite common, but you are also a bit lucky too ...
     
    Spajky, Oct 20, 2005
    #3
  4. giorgio.tani

    giorgio.tani Guest

    Should I have some long terms problems (or benefits?) keeping the CPU
    running with lower voltage (i.e. at 1 or 1,4 GHz with the voltage
    hinted for 600MHz)?
    Apparently it's working well, it's stable and worked as well some week
    ago with quite high ambiental temperature and now that it's many degree
    colder.
    With NHC I see reasonable temperatures for CPU and HD, with the machine
    that tends to get hotter before fans turns on when I run it at slower
    speed, and the fans that turns off at lower temperature and keep the
    machine colder when I operate at higher CPU speed.
    With higher CPU speed and lowest voltage I have a mixed situation, as
    expected (so I think the mainboard logic is handling the situation in
    the right way), since the machine tends to get hot slowly that at 1 or
    1,4 GHz with original voltage (since at lower voltage the power
    dissipation is lower) but the fans tends to start at the same
    temperature they will start at 1 or 1,4 GHz (so this seem controlled by
    the CPU clock and sensors of temperature rather than from voltage).
     
    giorgio.tani, Oct 21, 2005
    #4
  5. None that you'd ever notice, unless its more sensitive to lockup.
    Colder works in your favor as it's heat that's the enemy of speed.
    How did you determine what's 'reasonable' for a processor operating at a
    much lower than specified voltage?
    This is one of the problems when trying to 're-engineer' something: trying
    to figure out why it does what its doing.

    Odds are they're trying to reduce power consumption in the low power mode,
    since that's it's purpose, and throttle the fans near the cooling edge to
    keep them off as much as possible. However, at 'full speed', and voltage,
    it probably can't keep it cool under full load over all operating temps no
    matter how much the fans are run (a rather normal situation for notebooks)
    so it's kicking them in earlier to give as much 'full load' time as
    possible before processor throttling kicks in.

    Or, it could be something so simple as them monitoring die temp for fan
    control and your temperature display showing heatsink or case temp. In
    which case the die temp goes up instantly but it takes a while for the
    heatsink to warm so it 'looks' to you like it kicked in at a lower temperature.

    That is precisely my point. The processor will likely be more sensitive to
    elevated temperatures when trying to run full speed at a lower than
    specified voltage but the fans still operate as if it was not as sensitive
    because they were designed with the proper voltage at the design
    temperature in mind and not to keep it at a lower temperature that might be
    necessary when operating 'full speed' at lower voltages.
     
    David Maynard, Oct 22, 2005
    #5
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.