SUDDEN Reboot traced to memory - calling 'old timers' for tech info...

Discussion in 'Abit' started by foobar, Aug 12, 2003.

  1. foobar

    foobar Guest

    I've had a sudden reboot problem for perhaps a year now. Well I
    finally isolated it to memory (sort of) of all things.

    This is an older ABit ZM6 MB from a few years back. It's the ZM6 w/
    440 chipset and addresses up 256 RAM. I have a Celeron OC'd to 550 w/
    a 100FSB. It's been like this since I built it brand new years ago and
    absolutely rock steady (until a year ago).

    I originally had a single PC100 128mb stick in there. Roughly a year
    ago I installed a 2nd identical CP100 128 stick (from same maker) in
    the second slot and this was about when the random sudden reboot
    situation occured. It took me quite a bit of time to narrow things
    down to memory. Lots of H/W swapping around and an upgrade from 98SE
    to 2K was in there during this process.

    But I have verified beyond doubt that the system can be returned to
    rock steady w/ EITHER of my 128 sticks in ANY of the 3 slots on the
    MB. But I've spent the money on bringing it up to 256 and would like
    to get there w/ stablity.

    if I install BOTH 128 sticks (2X128) together - I incur sudden reboots
    at about 24 to 48 hours of runtime. The memory supplier was kind
    enough to send me a single 256 stick of PC100 brand new and I tried
    that. Same problem (1X256) BUT it will run longer, say 4 or 5 days
    before the sudden reboot occurs from what I can tell.

    1X128: STABLE (can run weeks or months and do pretty much anything
    from games to surf to spreadsheets)
    2X128: 24 to 48 hours MAX before SUDDEN reboot
    1X256: Maybe 4 or 5 days before SUDDEN reboot

    The BIOS settings are currently pretty standard, perhaps even
    conservative - just going for stability: (what would be some MORE
    conservative/stable settings?)

    SDRAM CAS Latency Time: 2
    SDRAM CAS to RAS Latency: 2
    SDRAM RAS Precharge: 2
    SDRAM PreCharge Control: Disabled
    System BIOS Cacheable: Disable
    Vidio BIOS Cacheable: Disable
    Vidio RAM Cacheable: Disable
    8 Bit I/O Recovery Time: NA
    16 Bit I/O Recovery Time: NA
    Memory Hole at 15-16M: Disable
    Passive Release: Disable (If I remember correctly)

    As far as the memory itself goes I don't believe it's bad. Both of the
    128 sticks test fine at the local PC store's memory checker. The 256
    stick is brand new. All three (BY THEMSELVES!) test pretty good under
    MemTest86. Each incurs a few TEST#5 Errors w/ memtest86 but I'm told
    this is fairly common for TEST#5. The only notable problem I've found
    is that when I run MemTest86 with BOTH of the 128 sticks installed
    TOGETHER - I do get a lot of Test#5 errors. (1000s).

    1) It's been suggested to try substituting a pair of 64 sticks for one
    of the 128s (I only have 3 slots available total).

    2) I wonder if the board itself has a problem? On the other hand I can
    stick either 128 into ANY of the 3 slots and it runs seemingly
    indefinately however w/o any sudden reboots.

    3) Would another possibility be the power supply? Is it logical that
    upping the memory would suddenly just push the power supply over it's
    limit somehow from stable to unstable? And why would it be so
    predictible in it's timing for the sudden reboot? It's no great PS,
    probably a little less than 300W (285W?). The voltages shown in BIOS
    seem to be fairly close to where they are supposed to be at 12V or 5V
    or whatever. (+11.79V, +5.04V, -12.03V, -4.97V, 4.94VSB). In the past
    (when stable) I've had every slot filled and was pulling power for 2
    USB devices and a card based scanner and now the system is somewhat
    stripped down so I wouldn't 'think' it would be an issue but thought
    I'd mention it. Some folks seem to put a lot of stock in big wattage
    PSs and others feel this is an overrated strategy.

    Anyway that's essentially where I'm at. Rock stable at 128 mb of RAM
    and sudden reboots at 256 sooner(2X128) or later(1X256). Any clues,
    suggestions, settings, etc is apprecaited,

    T
     
    foobar, Aug 12, 2003
    #1
    1. Advertisements

  2. foobar

    GTD Guest

    All the 'Old Timers' in here are in denial about their age, so you
    might want to re-word the request if you want to get their attention.
    You know how Old Farts are.
     
    GTD, Aug 12, 2003
    #2
    1. Advertisements

  3. foobar

    foobar Guest

    heh heh, OK fair enough - I was being tongue in cheek and mostly
    referring to myself! Anyway I'll repost...
     
    foobar, Aug 12, 2003
    #3
  4. foobar

    Rick Guest

    Have you checked the capacitors on the motherboard? Any signs
    of bulging, leaking at the tops or bottoms etc?

    Also, have you tried running your memory at CAS3 instead of 2?
    Any difference in the number of reboots?

    The increased number of Memtest errors when 256MB is installed
    is almost certain proof that it's a hardware problem, not related to
    Windows or application software.

    Rick
     
    Rick, Aug 12, 2003
    #4
  5. foobar

    Rick Guest

    It's something you should definitely check for, because your board
    dates from the same 1999-2001 timeframe where Abit and other
    manufacturers were using bum caps. The suspect ones are the
    1500uf's immediately behind the cpu.
    Couldn't hurt, although it probably won't help.
    Looks ok. You might want to consider adding 1 to your "8 Bit
    I/O Recovery Time" to avoid flaky problems with your floppy
    drive(s) and other 8 bit devices.
    Stick to slots 1 and 3 for your dimms. In most cases I've seen,
    slot 2 is most problematic on these older BX/ZX-era boards.
    Although not always.
    From the symptoms it looks like either bad caps or a flaky
    motherboard. In either case the dimm capacity won't make
    a difference.

    You gave voltage readings for your +5 and +12 rails on the
    power supply, what do you get for your 3.3v? Try bumping
    up the voltage in the system bios to 3.4 or 3.5, and if it's
    already at 3.5 try reducing it to 3.3. See if it helps.

    Rick
     
    Rick, Aug 12, 2003
    #5
  6. foobar

    foobar Guest

    OK I looked at the voltages again, Not sure what you mean by bumping
    but there is a 3.45V showing (Vio).

    The Vcore (cpu) is at 1.85 and I did bump it to 1.90V for the CPU -
    might run a tad hotter but I don't have any real heat problem.

    I also virtually eliminated all the remaining MemTest86 errors down to
    a single one (test#1 @ 155mb) w/ the following setting:

    CPU Level2 Cache ECC Checking: Enabled

    With the improvements in MemTest86 - I'm hoping this may stabliize the
    reboot problem - I'll be letting this setting run and watch it for the
    remainder of the week and see if that works w/o any reboots.

    thanks!
    -T
     
    foobar, Aug 13, 2003
    #6
    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.