RAM detection problems on P2B series boards

Discussion in 'Asus' started by P2B, May 13, 2005.

  1. P2B

    P2B Guest

    Anyone know what causes erratic RAM detection on P2B series
    motherboards, particularly at 133Mhz FSB?

    I've never had any problems running 768MB (3x256MB) on my P2B-DS boards,
    and one is fully loaded with a gig.

    I recently purchased a bunch of used 256MB sticks, mostly Micron but a
    few are other brands, with the intent of upgrading our single processor
    P2B systems from 256 to 512 or 768MB.

    I tested it all in matched pairs on a modified P2B-AE running a P3-S
    1.4Ghz CPU (133Mhz FSB) - configured the BIOS for 222 timings regardless
    of SPD, and ran memtest86 3.0 in all tests mode for at least 5 full
    passes on each pair.

    All testing completed without a single problem or error, but when I
    install in the target systems I encounter nothing but grief :-(

    On one P2B-AE system, the BIOS only detects either 64MB or 128MB from
    cold start when 512MB is installed, and it won't POST on reset - has to
    be powered off to try again. I tried all the tested pairs, and also
    mixed brands, but couldn't get it to detect more than 128MB when both
    RAM slots were filled - not even at 100Mhz FSB and 333 timings - but
    this one has the same board revision and BIOS as the test system that
    detected 512MB every time!

    Similar story on a P2B-S machine, although I eventually found a
    mixed-brand pair that works properly in this system - but no way will it
    detect three sticks.

    RAM voltage on all systems is within 100mV of 3.4V, the only significant
    difference is Tualeron processors in the target systems vs. a P3-S in
    the test system - but that seems unlikely to affect RAM detection.

    All very mysterious and annoying - comments or suggestions appreciated.

    P2B, May 13, 2005
    1. Advertisements

  2. P2B

    BigJim Guest

    no really the p2b needs low density ram
    BigJim, May 13, 2005
    1. Advertisements

  3. P2B

    Rick Guest

    Assuming your modules are correct (low-density, organized
    into two logical banks of 128MB per dimm instead of one
    256MB bank), the problem is likely related to voltage. Most
    BX board designs, even those with three dimm slots instead
    of four, have anemic 3.3v power supplies to these slots.
    Many share power between slots, so e.g. you might get one
    running with slots 1 and 3 populated but not 1 and 2. Or
    with 1 and 2 populated, but not all three. Etc.

    Not much you can do other than trial and error.
    Rick, May 13, 2005
  4. My P3B-F runs perfectly with 3x256MB PC133 modules from Crucial
    at 100MHz FSB, but at 133MHz FSB (the CPU is a 1.26GHz Tualatin P3)
    there are intermittent read errors detected both by memtest86 and
    by running heavy application loads. 2x256MB works reliably at 133MHz,
    but unfortunately I need all the memory I can get.

    I guess I should have bought registered modules :-(

    Mikael Pettersson, May 13, 2005
  5. You don´t know the i440BX.

    Well P2B, I would suggest you


    and http://p196.ezboard.com/bcubx
    this is a CuBX Forum but very interesting for P2B/P3B users, too.

    Don´t forget to use very well RAM for the BX. 133MHz Cl.2 RAM do not have to
    work in a BX with 133 2-2-2 setting, but in most cases it does.
    You will also find in the Internet some comments about the stability on P2B
    derivates and CuBX. So far as I understood, the CuBX and the P3B-F are the
    best for overclocking, operating at 150MHz without problems. My P2B-F for
    example is finishing at <140MHz, even you can clock it up to 150. But I have
    never gone beyond 105MHz, and staying now at 100MHz. For what should I need
    more MHz????

    Maybe your P2B cannot operate the Tualatin? Voltage regulator?

    Best Regards,

    Daniel Mandic

    P.S.: For BX enthusiasts like me: I have found an article (Japan) about a BX
    with <200MHz FSB givin 2100 MHz Tualatin. You would need a Intel extreme
    edition to outdistance such a system :)
    I am sure the formerly BX-Designers could make such FSB work
    (without overclocking),some years after the official BX-Release, which was
    Daniel Mandic, May 13, 2005
  6. P2B

    P2B Guest

    IME Micron PC133 CL3 always tests clean and runs stable at 222 on P2B
    boards. Other brands frequently show a few errors per pass in memtest86
    and must be run at 333 for stability.
    I have two P2B-S systems running Tualatin processors @ 150Mhz FSB with
    512MB (2x256) of Corsair PC150 CL2, long term stable. I'd get more of
    the same RAM if I could find it...
    The systems have all been running Tualatins @ 133Mhz FSB with 256MB for
    1-2 years.

    What I don't understand is why they won't detect 512MB - but my test
    system with the same board and BIOS revisions, and same RAM voltage
    (within 100mV) does so every time.
    I find that hard to believe - even the best Tualatin cores top out
    around 1700Mhz, and what would you use for RAM?

    P2B, May 13, 2005
  7. Lucky one..
    Very expensive...
    The BX cannot fully recognise 512MB Dimms. If you mean 2x256MB then try to
    insert one in slot0 and the other in slot2, leaving the second slot, slot1
    Who knows what kind of RAM you can buy in Japan. As you know, the most
    companies, like NEC are from Japan. Maybe a PC180 Cl.2!? With special
    cooling you can go to unbelievable MHz. But this is all uninteresting. A
    fine PC should have a consumption of 50 to maximal 200 Watts. With more I
    should sit on it and drive and not siiting behind it ;-)

    Best Regards,

    Daniel Mandic
    Daniel Mandic, May 13, 2005
  8. btw, you'd rather should read
    http://homepage.hispeed.ch/rscheidegger/ram_bx_faq.html which is the
    original source. I don't like others to copy whole pages 1:1 instead of
    linking but don't care enough to really do something about it...
    Those are probably the same guys which overclock P4 to over 5Ghz and
    other similar crazy things. Liquid nitrogen will help a lot to make that
    1700Mhz-limited tualatin run faster :)

    Roland Scheidegger, May 15, 2005
  9. Hi Roland, nice to see you here.

    I have your link in my Favorites, too. The homepage of the other link
    contains really a copy of yours. But the guy was so kind to leave your name,
    at least.

    High density. I did not know that the 32Mx4 architecture was called so. But
    when you count the RAM blocks in a 16Mx8 it would be more likely a high
    density. 8 is more than 4 ;-)

    Best Regards,

    Daniel Mandic
    Daniel Mandic, May 15, 2005
  10. P2B

    P2B Guest

    Update - the problem appears to be BIOS/CMOS related.

    After much trial and error, I have determined the following RAM upgrade
    procedure yields successful results when upgrading P2B series boards
    running at 133Mhz FSB or higher from 1 x 256MB to 2 or 3 x 256MB:

    1. Clear CMOS
    2. Set FSB to 100Mhz
    3. Install desired RAM configuration
    4. Power on, wait for ESCD and DMI update message display
    5. Press reset, enter BIOS and restore previous settings, save & exit
    6. Wait for ESCD and DMI update message display
    7. Power off, restore previous FSB setting
    8. Power on, system should detect all RAM and boot normally


    - Doesn't appear to make any difference which RAM slots are populated
    - Steps 2 & 7 may not be required on some boards, but initial detection
    of new RAM appears to be more reliable at 100Mhz FSB
    - Step 4 may not be required on some boards, but allowing BIOS to update
    DMI/ESCD with values from SPD prior to manually setting RAM timings
    appears to be more reliable

    Clearing CMOS is the key step - a couple of my systems wouldn't even
    POST after I powered down and added another stick of RAM, but worked
    fine after clearing CMOS.

    P2B, May 16, 2005
  11. P2B

    timmy Guest

    I have the PB3-F, fully loaded pci and 3x256mb dimms (DS)(cl222). I
    did not notice this same problem until I swapped cpu's to a 133mhz
    coppermine from a 100 mhz coppermine. Now the ram registers as cas333
    10 instead of 222 8. Setting this manually is no problem and the
    system runs fine. But there is an intermintant problem where the
    memory may or may not be all counted "AFTER" a cmos change and save.
    It's wierd and I will be testing the dimms individually, even though
    I'm sure it is something else, possibly something along the lines of
    what you have described...
    timmy, May 16, 2005
    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.