PC100 CAS2: 256MB - VS - 128MB X 2 ???

Discussion in 'Abit' started by foobar, Jul 5, 2003.

  1. foobar

    foobar Guest

    I have a 'sudden' 'random' reboot issue w/ my ABit ZM6 system (i440ZX
    chipset, Celery OC'd to 550, FSB100).

    It has happened twice in the last 2 days but it's one of those things
    that may be OK for several days or a week or better.

    This has never been a problem for the 1st few years I've had this
    system. But I'm leaning towards a h/w problem. A couple months ago I
    installed a second 128mb stick (both are from Mushkin) and that more
    or less coincides with when this problem started. But the problem can
    be so darn intermittent and it's been quite a few months now so it's
    hard for me to be specific. (And if I have to keep pulling h/w every
    couple week to isolate - it could take months!)

    I read somewhere a while back (cant' find it anymore) that a singel
    256 stick is better than 2 128s and can avoid h/w problems like
    this...?

    Soooo... memory?

    Other possibilities? Power supply? CMOS battery? I dunno where some
    logical places to start might be...?

    Thanks!

    ---

    I know some folks might suggest the OS or virus etc...
    Background: I've done one fresh install of 98SE recently and just last
    week I wiped everything and did 2K Pro. None of this seems to really
    impact it. Between fresh installs from scratch and AV and firwall s/w
    I don't see this as a likely s/w problem. In both cases I completely
    wiped the disk and installed from scratch...
     
    foobar, Jul 5, 2003
    #1
    1. Advertisements

  2. foobar

    TomG Guest

    just as a test, run the system with just the one stick and then the other
    stick of ram and see if it is stable with both by themselves. if it is,
    then it is likely an interaction between the sticks of ram and maybe if you
    talk to Mushkin, they might help you out with the issue.

    --

    Thomas Geery
    Network+ certified

    ftp://geerynet.d2g.com
    ftp://68.98.180.8 Abit Mirror <----- Cable modem IP
    This IP is dynamic so it *could* change!...
    over 113,000 FTP users served!
    ^^^^^^^
     
    TomG, Jul 5, 2003
    #2
    1. Advertisements

  3. Do you have a good, local compute store (one of the medium-sized ones
    with a bunch of components for self-builders)? They should have a memory
    checker and a reasonably-priced stock of replacements. If that proves to
    be the problem.

    Power supply is a possibility (how old is it?). Don't think the CMOS
    battery is the problem (usually interferes with booting, not causing re-
    booting).
     
    Jeff Lindstrom, Jul 6, 2003
    #3
  4. foobar

    foobar Guest

    Wow! All great posts and good keeper info here - thanks very much!

    Yes, I have already pulled 1 stick and am running w/ just 128 right
    now. So far so good. I purchased the original stick in '99 and this
    second stick sometime within the last 6 or 9 months. I'll let it run
    all thru next week and see if it does another 'WHAMO' reboot. I think
    I tried this once before and don't think it ever rebooted before.

    I don't know what Mushkin is likely to do here though if I call
    them...? I'm not even sure I still have the receipts.

    I'll disregard any concerns I had about the battery.

    The PS is the one that it's had since I built everything back in '99
    so it's 4 years old. It's like a 235W or 250W or something. The
    voltages displayed in BIOS are probably within less than a 0.5V or
    spec - I can check and post exact voltages - not sure if that tells us
    anything. What should I be looking for on PS issues?

    It's been so long since I have checked the voltages - I'd have to boot
    and double check but I want to see it's voltage is pretty close to
    'stock', maybe 1.8 or 1.85V?

    I've not noticed any obvious visual problems w/ the caps but I wasn't
    looking for anything either - I just had it opened up and on the table
    and blew out some dust when I went thru things before installing W2K.
    It's back in it's 'corner' but I'll double check w/ a flashlight for
    any signs as suggested.

    I do have memtest86 and have run it in the past including recently
    (since this problem manifested) and it has always been fine. HOWEVER
    just the other day I ran it again and did got some errors right around
    the 1mb and 127-129mb area - maybe 2000 errors specifically in TEST 5
    [Block Move, 64 Moves, cached]. That's strange because I hadn't seen
    that before. I pulled sticks and re tested 1 stick at a time - got the
    same error in similar locations but the # of errors dropped
    considerably to only 10 or 20 errors. I've since adjusted some of the
    CMOS settings and need to retest and verify if the problem was due to
    a setting I had previously fiddled w/ or not. Strange.

    I'm a bit confused about memtest86 - I've got some older systems too
    and sometimes memtest has reported EDO memory errors on those and yet
    they are rock steady just fine...? Is this generally considered a
    pretty reliable tool?

    Any good sites/recommendations that explain all the intricacies and
    such on some of these BIOS settings...?

    Prime95!? Hadn't run that since I built this system. I'll give that a
    shot as well as retest w/ memtest86.

    I do have 1 PC store that could probably test memory - we used to have
    a bunch but they seem to be getting pretty rare around here these days
    - not sure what's up w/ that?

    ---

    Again, thanks for the great info and suggestions! I'll follow up w/ my
    findings and see what happens. But I suspect due to the nature of this
    type of problem - this may be a sloooow process.

    -T
     
    foobar, Jul 6, 2003
    #4
  5. Power-transistors don't last forever.
    Neither do caps, but they could last a few more years.
    Have a look inside the PSU to check the caps and while you're at it,
    clean out the dust - dust buildup could make some of it run so hot, that
    voltages would be somewhat off.

    Another thing - I also have an Abit ZM6 (somewhere) and I remember
    putting thermal paste below the heatsink on the Northbridge.
    There was none when I originally removed it to check.
    A heat problem with the Northbridge could also cause "memory" errors,
    although the memory sticks would be fine.
    If you buy a new one, I would recommend an Antec.
    I currently have an Antec PP-303XF powering a system with a T-Bred
    running at 2000 MHz on an Abit NF7. Very low noise.
    Extremely stable.
    They should be within ±0.2V of spec for +3.3V and +5V, (and within ±0.5V
    for +12V - the +12V rail is not very sensitive to voltage out of spec
    though).
    Particularly, the voltages shouldn't be too low.
    A system might still run seemingly "stable" with voltages out of that
    range, but it would be very sensitive to voltage drops/hops - and would
    either reset, lock up or exhibit strange errors occasionally.
    Well, practically what you are experiencing right now - a not 100%
    stable system.
    Voltage out of spec indicates a defective PSU (Power Supply Unit).
    You can check the voltages with Winbond Hardware Doctor, downloadable
    from Abit.

    ftp://ftp.abit.com.tw/pub/download/fae/zm6.exe

    Check voltages when running at full load - use cpuburn to do that;

    http://users.ev1.net/~redelm/
    Stock voltage(Vcore) for a Celeron 366 is 2.0V, but running at 550 it
    would have to be overvolted to 2.2V (at least).
    Below 2.2V it will probably not be stable, especially since it is now 4
    years old.
    Has it been running 24/7, or just some hours every day?
    A Vcore that is low, compared to the Bios setting, indicates defect caps
    or PSU.
    Good, it could still be the caps.
    And no errors in test 5, non-cached?
    I didn't go any deeper into it, because I would wait and see whether
    memtest86 reported errors.
    Same (test 5, cached) error with both sticks, one at a time?
    No errors in test 5, non-cached?
    (Let's call them stick A and B)
    Which settings?
    Have you also tried running the Prime95 "torture test" on those systems?
    Yes.
    If memtest86 reports errors, it's because there is something wrong.
    (no false-positives)
    Memtest86 stresses the memory in a way that rarely occurs in everyday
    use, so it will even find errors that might only occur once or twice a year.
    Tomshardware's BIOS Guide maybe..?
    Well, now that we know memtest86 reports errors, we don't need it.
    It is usually run to check CPU+Chipset+RAM stability.
    If Prime95 and memtest86 had reported no errors, I'd have said it was a
    software problem.
     
    Jens C. Hansen [Odense], Jul 7, 2003
    #5
  6. Minor correction;

    "although the memory sticks would show no errors in another board".
     
    Jens C. Hansen [Odense], Jul 7, 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.