The System has experienced boot failures because of overclocking

Discussion in 'Gigabyte' started by red floyd, Jun 1, 2011.

  1. red floyd

    red floyd Guest

    Brand new mobo: GA880-GA-UD3H
    CPU: Athlon II x3 455
    Memory: 8GB (4GBx2) Patriot Sector 5 G 1066/1333EL (PGV34G1333ELK)

    When I boot, I get the error in the title: "The System has experienced
    boot failures because of overclocking"

    Except... I'm not overclocking, all memory timings are set to "Auto"
    and/or "SPD".

    How can I get rid of this? The memory passes the Windows 7 memory test.

    If it helps, I'm using the on-board Radeon 4250HD.
    red floyd, Jun 1, 2011
    1. Advertisements

  2. red floyd

    Paul Guest

    GA-880GA-UD3H memory list (not really useful, info needs better formatting)

    Your processor. No important details here, either.

    PGV34G1333ELK 2x2GB kit, possibly two kits totaling 4 DIMMs.

    "...and DDR3 at up to PC3-10600 (DDR3-1333) speeds. There is a catch
    though - if you're running four DIMMs (to use, say, 8GB of RAM),
    the supported speeds drop to PC3-8500 (DDR3-1066)"

    That is a statement about the impact of bus loading on an AM# processor.
    Bus loading affects the ability to run full speed. You can experiment
    with this manually, to try to improve the situation. Normally, the
    BIOS would make conservative choices to guarantee startup.


    Check the BIOS. You can try selecting Auto for stuff, except for
    the memory speed. With four DIMMs, set the memory speed to
    DDR3-1066 and let the BIOS figure out what CAS to run and so on.
    Then attempt to do your memory test, later booting into Windows.
    It could be, the BIOS isn't setting DDR3-1066 when four
    DIMMs are present.

    If you want an alternative suggestion, drop down to just two
    sticks, set everything to Auto, boot into Windows, run CPUZ
    and verify how the BIOS set up the memory. Of course,
    this won't prove the board really needs DDR3-1066 setting
    when four DIMMs are present, but will at least prove you're
    making progress with two sticks at DDR3-1333 (one stick per channel).
    You can run two separate memory test cases then as well, testing the
    first two sticks by themselves, and when they test good, testing
    the second two sticks.

    See the first or second line of the "Dual Channel Memory Configurations
    Table" in the manual, for how to install the two DIMMs.

    You may be able to verify in the BIOS, what speed is being used
    when either two or four DIMMs is present. On my current motherboard,
    it reports during the POST, whether dual channel is selected,
    the speed, the memory quantity. You can use the Pause key on
    the keyboard, to make that stand still, if you wish to read it
    at your leisure.

    I don't like booting Windows, until the memory has been tested.
    I use memtest86+ from to do the testing. (Scroll half
    way down the web page, to find the downloads.) Once the
    memory passes and no stuck-at faults are found, then I try
    booting into Windows. Passing memtest86+ is not an acceptance
    test. You should use a stress tester such as Prime95 for that.
    A stress tester does a better job of uncovering soft or
    transient faults. Soft faults may not repeat at the same
    address location, each time you test.

    Paul, Jun 1, 2011
    1. Advertisements

  3. red floyd

    red floyd Guest

    Thanks for the help, it's really appreciated.

    Oops. Typo on the part number. It was a 2x4GB kit. It's
    the PGV38G1333ELK.

    I do have everything set to Auto. And it does default to 1066.

    Windows 7 "Recovery Console" has a fullbore memory test, similar to

    Some sites have recommended clearing the CMOS. I've already reflashed
    with the FF BIOS. If using the CMOS clear jumper doesn't work, I'll
    probably return the memory and get some AMD recommended (though that
    PDF doesn't list too many 4GB/1333 sticks).

    Kingston has some that they say will work properly with this mobo.
    red floyd, Jun 1, 2011
  4. red floyd

    Paul Guest

    If you have flashed the BIOS, then clearing the CMOS is a logical step.

    Some BIOS flashing tools, have a software option for clearing CMOS,
    while the flash is progressing. The flashing tool may have
    access to the 256 bytes of RAM in the Southbridge.

    If that doesn't seem to be the case, then the manual method (Clear RTC
    or Clear CMOS jumper) would be the next step.

    The main warning about Clear CMOS, is some implementations have a
    power dependency. The ATX power supply should be turned off, when
    following the instructions in the manual concerning clearing CMOS.
    On my machines, I actually unplug them, just to be safe.

    A number of motherboards, they "short to ground" using the CMOS
    jumper. And there happens to be a path from +5VSB, through a
    regulator, to the node in question. If the power supply is
    left running, installing the CMOS jumper to "short to ground",
    shorts the +5VSB from the power supply, through a tiny dual
    ORing diode package. One of the diodes gets burned. I helped
    one poster repair his board, and he unsoldered the burned one
    and installed two 1N914/1N4148 replacements. (You can replace a
    dual diode, with two single diodes. His choice of a 1N4148
    type was not the best, but he was pretty happy the motherboard
    still worked.)

    As a consequence of the possibility of damage, make sure the
    power is off, before using the CMOS clear jumper. Not all
    circuits are designed in the destructive way, but there are
    enough of them, that it's just easier to turn off the power
    for all of them. I've even seen cases, where Intel provides
    a "safe" implementation, and the manufacturer ignores it
    and installs the "burn if power on" method instead.


    With respect to RAM types, there was a situation in the past,
    where there was a difference between "tuning" for 1GB and
    2GB sticks. It took some BIOS updates, to get good Auto behavior
    with 2GB sticks. Yet, if you look at the electrical specifications
    for the two chip types used, there was nothing hinting at such a
    difference. The electrical interface looked identical. And
    yet, some tweak was required several years ago, to reduce the
    error rate on the 2GB modules.

    Perhaps your 4GB module is a similar situation. They've only
    become popular (and cheap) relatively recently.

    Paul, Jun 1, 2011
  5. red floyd

    red floyd Guest

    Clearing the BIOS worked. I'm going to leave it at Auto/SPD for memory
    timings. Just one question, should I set to Ganged or Unganged? What's
    the difference? Also, what effect does the Southbridge Spread Spectrum
    setting have?
    red floyd, Jun 2, 2011
  6. red floyd

    Paul Guest

    "It’s been reported that the ganged mode usually works better for single-core
    performance, but unganged works better for multi-core."

    Ganged runs the dual channels as a 128 bit wide memory.

    Unganged makes the two channels independent (two 64 bit wide data channels).
    If a channel is available, then a core needing to do a cycle on that set of
    memory addresses, can start its access.

    What's interesting, is their benchmark results on that site, the two
    modes made hardly any difference at all. The difference may be more
    apparent if two programs were running at the same time, and the
    completion time of both programs was measured.

    The sensitivity to the setting may also be influenced a bit, by the
    size of L3 on the processor. Some cheap AMD processors have no L3,
    so they're a bit more dependent on how snappy the memory subsystem is.

    You have the option to test both, and see if it makes any difference
    at all. As far as I know, the setting likely defaults to unganged (independent)


    Spread Spectrum is a technique for clock modulation. It helps the
    manufacturer pass FCC part 15 emissions testing. It has virtually
    nothing to do with the end user, and the normal function of the

    I turn mine off. I would only consider turning it on, if I saw signs
    of severe interference on the screen of a TV set.

    It would really be wiser, if that function wasn't in the BIOS screen.
    If the manufacturer thinks it is wonderful, then enable it, and hide
    it. Exposing the control for it, implies it isn't really needed.

    Paul, Jun 2, 2011
  7. red floyd

    red floyd Guest

    Thanks for the info, Paul! Really appreciated it!
    red floyd, Jun 2, 2011
  8. red floyd

    arnieb Guest

    I've just gone through this myself. I'm building a 6-core Phenom with
    16 GB of 1333 MHz memory in a 4x 4GB configuration. It wouldn't fly at
    that clock speed and I had to back it off to 1066 MHz. I was thinking
    that it might be the drive capability issue. There may be a way in the
    BIOS to increase the drive current and voltage to the memory array to
    compensate for the increased capacitance, but I'm not sure what that
    is or how to do it.

    arnieb, Jun 23, 2011
  9. red floyd

    red floyd Guest

    It's in the M.I.T. menu in the BIOS.
    red floyd, Jun 23, 2011
    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.