Tyan s2668 dual Xeon 2ghz 512kb L2 2MB L3 - bios problem

Discussion in 'Tyan' started by nSurferx, Feb 27, 2008.

  1. nSurferx

    nSurferx Guest

    By using DrDOS7 bootdisk which does not load any memory manager which is
    good,
    and I've done flashing bios from 1.00 all the way to its latest bios
    1.05 yet after
    each bios flash it says:

    missing microcode update for processors

    I've searched the net with no results.

    What I have on my first self-built Xeon PC:

    Tyan i7505 s2668
    BFG 650w with 24 pins / 8 pins
    4x256mb ECC PC2100 ram (1 gb)
    ATI x800 Pro 256mb agp
    BenQ DVD DL burner

    and the "offending" CPUs? 2 ghz x2 512kb L2 / 2MB L3 / 400fsb.

    I installed Vista Ultimate with no problems except pushing F1 to resumble
    after each reboots.

    Any solution to this baffling problem?
     
    nSurferx, Feb 27, 2008
    #1
    1. Advertising

  2. nSurferx

    Paul Guest

    nSurferx wrote:
    > By using DrDOS7 bootdisk which does not load any memory manager which is
    > good,
    > and I've done flashing bios from 1.00 all the way to its latest bios
    > 1.05 yet after
    > each bios flash it says:
    >
    > missing microcode update for processors
    >
    > I've searched the net with no results.
    >
    > What I have on my first self-built Xeon PC:
    >
    > Tyan i7505 s2668
    > BFG 650w with 24 pins / 8 pins
    > 4x256mb ECC PC2100 ram (1 gb)
    > ATI x800 Pro 256mb agp
    > BenQ DVD DL burner
    >
    > and the "offending" CPUs? 2 ghz x2 512kb L2 / 2MB L3 / 400fsb.
    >
    > I installed Vista Ultimate with no problems except pushing F1 to resumble
    > after each reboots.
    >
    > Any solution to this baffling problem?


    I expect "live with it", is the answer in this case.

    Every processor has bugs or errata, and an average processor might have 100
    known issues associated with it.

    To help deal with it, Intel placed an area of RAM inside the processor, to
    hold microcode patches. Say, for example, the FADD instruction was bad. Then
    perhaps they could patch the instruction and changes its behavior. Or do things
    like disable the TLB, if the TLB was badly broken.

    In the P4 era, the microcode patch was around 2KB max, but the available space
    has grown with later generations.

    There are two opportunities to load any necessary microcode patches.
    When the BIOS starts is one of them. A typical BIOS contains a module with a
    name something like CPUCODE.bin, and it may contain eight different microcode
    patches (can be more or less). They're loaded into the CPU, until one "takes".
    The patches are protected by a checksum of some sort, so the CPU can check them.

    A microcode patch loaded at BIOS time, ensure "correct computing" during the
    boot phase.

    The next opportunity to load a patch, is via a microcode loader in the OS.
    When the desktop appears on your screen, the microcode loader would have
    an opportunity to run and install whatever 2KB patch the OS happens to have
    resident. (The last time that screwed up, was I believe during the deployment
    of WinXP SP2 perhaps.)

    When the only microcode patch, is installed by the OS, it means the boot
    phase of operation, runs without a patch.

    Intel has a processor identification utility, and it gives a family code,
    like F29 and a revision, like 17. The F29 varies with processor family, and
    the F29 part is actually compared to the info in a patch, to decide whether
    it belongs or not. The F29 is a permanent part of the silicon die, ROM if
    you will.

    The "revision" field, is actually keeping track of the microcode patch. If
    no patch ever got loaded by BIOS or OS, the revision = 00. If the BIOS had
    revision 05 and the OS had nothing, you'd see 05. If the OS had a later
    revision, and installed it over top of 05, you might see 17. So the
    revision field, allows a runtime check, of what microcode is running on
    the processor. If you knew for a fact that the BIOS had an 05 revision,
    and you saw 17 while in Windows, you'd assume 17 got loaded by the OS
    microcode loader.

    If you had an Award BIOS, the CTMC package allows "adding" microcode to an
    Award BIOS, in a safe and non-destructive way. I'm not aware of an equivalent
    option for AMI or Phoenix (but there may be a way). And your BIOS appears to
    be Phoenix as near as I can tell.

    For an Award BIOS, the CTMC package has a program called "splitawd" or the like,
    which splits the Award BIOS into separate modules. One of the modules would be
    CPUCODE.bin or some equivalent name. (The name is not important, as long as the
    BIOS knows what it is called.) Once split into modules, the "lha" program
    can be used to decompress each module and show plaintext.

    For an AMI BIOS, the "mmtool" allows BIOS module extraction and decompression.
    So the dissection is a one tool process.

    I don't know what you use for Phoenix.

    If I had an Award BIOS, I'd use splitawd and lha, to convert CPUCODE.bin to
    plaintext. The Intel processor architecture documents, document the header
    format of each microcode patch, showing family code and revision. So in
    fact, it is possible to parse CPUCODE.bin by hand (I've done it a few times
    now), and see what microcode patches have been installed. In the most recent
    Intel processors, where the microcode patches are variable length, it is a bit
    harder to parse them (a nuisance) but it can be done. I've tried my hand at a few.

    Using CTMC, I've actually added a 2KB patch to my old 440BX board with a Tualatin
    processor installed in it, and made the microcode warning message "go away".

    I took a look at the S2668 BIOS with a hex editor, and I don't think I can
    do anything for you. AFAIK, all BIOS are internally compressed, and unless I
    can identify the compression method, it is pretty hard to dissect the BIOS.

    If you have the right BIOS editing tools, it is possible to add your own version
    of CPUCODE.bin to the BIOS. The BIOS is protected by multiple checksums, and
    the BIOS will reject changes that don't checksum correctly. Since BIOS development
    tools are revved occasionally, it means that when the odd tool is "released" to
    the Internet, it may not work forever, so expecting to "DIY repair" the BIOS
    would be expecting a lot. I've never actually modified a BIOS file, because
    the tool I was using, kept reporting checksum problems, and I didn't have the
    guts to try flashing the resulting file into hardware. (CTMC uses a different
    method, which is why I could do it.)

    I wouldn't be overly worried about this. If your motherboard is being used as a
    "critical server" in a corporate environment, then of course you'd "dot the i's
    and cross the t's" and contact Tyan and try to resolve the issue. Because you
    wouldn't want to risk some problem showing up during an automatic reboot at
    night (no matter how infinitesimally small those odds were). But if this is
    just being used as your desktop, and you have a recent OS equipped with
    microcode loader (WinXP and probably Vista), your exposure is a small one.

    Before talking to Tyan, you'd want the processor order code or SSPEC in hand,
    to identify the processor you are using. Telling them "2GHz" is not enough.
    For example, a Q6600 quad desktop processor, has an SSPEC of "SLACR", and
    telling them "SLACR" would identify the unit. You can see an example on
    processorfinder.intel.com, where more details are available.

    "SLACR" is a 6FB family processor. Stepping is G0 in this case.
    http://processorfinder.intel.com/details.aspx?sSpec=SLACR

    To get a "warmer feeling", find the appropriate Intel processor identification
    utility, and view the revision number field for each processor. As long as it
    is >00, you should be "safe".

    I'm not aware of what method AMD uses for their processors, and their documentation
    leaves a bit to be desired anyway. So I'm in no rush to find out :)

    HTH,
    Paul
     
    Paul, Feb 27, 2008
    #2
    1. Advertising

  3. nSurferx

    nSurferx Guest

    Thanks for the reply - that's quite a long one!

    I think I saw CPU ID or code 0080 when I hit F2 to enter the bios.
    It's just my personal PC. I built many PC as my hobby and Xeon is
    the first one just for fun. And yes it's a Phoneix BIOS. I've read
    other people's issues in other forums and he said everything's running fine
    despite that 'missing microcode' at POST so I guess I'm in the same boat.
     
    nSurferx, Feb 27, 2008
    #3
  4. nSurferx

    Paul Guest

    nSurferx wrote:
    > Thanks for the reply - that's quite a long one!
    >
    > I think I saw CPU ID or code 0080 when I hit F2 to enter the bios.
    > It's just my personal PC. I built many PC as my hobby and Xeon is
    > the first one just for fun. And yes it's a Phoneix BIOS. I've read
    > other people's issues in other forums and he said everything's running fine
    > despite that 'missing microcode' at POST so I guess I'm in the same boat.


    Mine was a "learn by doing", so I figured I'd pass what I learned, along.

    Paul
     
    Paul, Feb 27, 2008
    #4
  5. nSurferx

    nSurferx Guest

    I replaced 2 x 2ghz with 2 x 2.8ghz / 533fsb / 512kb L2 cache and that
    solves it! Just boots into Vista with no problems.
    Vista seemed to run quicker since I added 2gb rams (4x512mb).
     
    nSurferx, Mar 17, 2008
    #5
  6. nSurferx

    deep23

    Joined:
    Jun 10, 2008
    Messages:
    3
    Message

    I think I saw CPU ID or code 0080 when I hit F2 to enter the bios.
    It's just my personal PC. I built many PC as my hobby and Xeon is
    the first one just for fun.
     
    deep23, Jun 10, 2008
    #6
    1. Advertising

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

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Walter H.

    TYAN S2668

    Walter H., Dec 1, 2003, in forum: Tyan
    Replies:
    0
    Views:
    992
    Walter H.
    Dec 1, 2003
  2. TDK

    S2668 with one Xeon

    TDK, Apr 7, 2004, in forum: Tyan
    Replies:
    2
    Views:
    404
    Paul Missman
    Apr 14, 2004
  3. nyala5000
    Replies:
    11
    Views:
    1,054
    nyala5000
    Mar 29, 2005
  4. Sammy Smith

    Need help with AMD Barton XP 3200+ 2.2GHZ 512KB 400FSB

    Sammy Smith, Oct 29, 2005, in forum: AMD Overclocking
    Replies:
    3
    Views:
    257
    Ashley
    Nov 1, 2005
  5. mr_v8
    Replies:
    0
    Views:
    1,145
    mr_v8
    Mar 29, 2008
Loading...

Share This Page