Re: Fan speed errors and startup hang time

In article <1121319310.916301.224890@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
CrazyDrummer@xxxxxxxxx wrote:

> Hey guys, I have a P4B266LM and had a P4 1.8 installed. I recently
> upgraded to a P4 northwood 2.8, and it ran normally for 2 boots, but
> after that my PSU fan wouldn't slow down to idle RPM like it usually
> did. It's weird because the northwood core requires less voltage than
> my old willamette core, and runs cooler because I installed a copper
> heatsink with artic silver.
> Another problem, after trying to remove my heatsink (which I couldn't
> do because it was on too tightly) to reinstall my old processor to see
> if I could possibly fix the fan speed issue my system hangs much longer
> than normal at the boot screen past where you get into BIOS, but before
> windows starts loading up. It sits there for a while (about 10x longer
> than it used to before I tried to remove the heatsink). It also gives
> me a message something like bad BIOS information. I'm considering
> removing some expansion cards to see if they're the problem, but i
> doubt it, because it booted normally after I installed the new
> processor.
>
> Any help is really appreciated.

If your PSU fan won't slow down, it could mean two things. It
could be the PSU is defective, and has just decided now to
misbehave. Or, it could be an indication that the motherboard
is drawing a lot of current from one of the rails - that could
be a short somewhere. (You could try the cardboard test. That
entails removing the motherboard from the case, and assembling
it on top of a non-conductor, like a thick telephone book. You
can remove the four screws from the PSU and put it next to the
motherboard etc. See if the symptoms are still there or not. It
could even be something innocuous, like if the +5V feeding the
speaker that beeps at POST, is getting pinched and shorted to
ground.) If you had a non-contact ammeter, testing for this would
only take seconds (I've got one and it is a great toy), but
without equipment, debugging is much more difficult.

In terms of power, in fact the two processors are pretty close
to one another. Here are examples of which processors I think
you are using:

2.8GHz/FSB400/512KB cache CPUID 0F29 ~1.5V 68.4W
http://processorfinder.intel.com/scripts/details.asp?sSpec=SL7EY

1.8GHz/FSB400/256KB cache CPUID 0F12 1.75V 66.1W
http://processorfinder.intel.com/scripts/details.asp?sSpec=SL5VJ

The new processor uses a lower voltage, but more current.

If could be that the message you saw, was talking about "microcode".
Your BIOS probably doesn't have microcode for a 0F29 processor.
If you flash upgrade to the latest BIOS, like 1008.002, it has
microcode for 0F0A, 0F12, 0F13, 0F21, 0F23, 0F24, 0F27, 0F29. The
1006 has 0F0A, 0F12, 0F13, 0F21, 0F23, 0F24, 0F27 and is missing
0F29 (your new processor).

Microcode is a 2KB file, that is loaded into a special RAM inside
the processor. The code in the file is used to correct bugs in
the processor, and by having the BIOS install the code, that
prevents processor bugs from stopping your system from booting.
I think WinXP has a microcode loader, and it can take over once
the desktop appears. (IOW missing microcode is not the end of the
world.)

The Microcode, once a new processor is recognized, is written into
a segment of the flash chip, during POST. (This is like a cache, only
using the flash chip as a non-volatile store.) It could be this
operation, or some DMI/ESCD activity, that is slowing down the POST.

There is another recipe for installing microcode, that uses CTMC,
and requires extraction of a 2KB microcode file from another
BIOS. I've done this on a Tualatin, and it works no problem. CTMC
uses a hook in the Award BIOS, and there is no similar hook in
an AMI BIOS. So, while CTMC makes the installation of microcode
less risky (as a failure won't stop the system from POSTing), the
program does rely on a certain feature being available in the BIOS
code.

Once you've improved the BIOS microcode situation, you might
notice the delay goes away. I cannot promise anything, as I
didn't see a delay on my system, when the microcode was
missing.

Paul
.

Relevant Pages

  • Re: New CPU installed! error
    ... install a more recent beta version, ... There should be an Information tab, and a DMI Explorer option. ... In any case, what we are checking here, is what the BIOS has ... The microcode is the greater of the revisions found in the ...
    (alt.comp.periphs.mainboard.asus)
  • Re: How can I reinstall "microcode update device" (in device manager)?
    ... the "microcode update device" icon disappeared. ... Normally, your BIOS will look through its collection of microcode files, ... If you install that program and run it, then go to the "CPUID" tab, it gives ...
    (microsoft.public.windowsxp.hardware)
  • Re: How can I reinstall "microcode update device" (in device manag
    ... How can I reinstall the "microcode update device"? ... Normally, your BIOS will look through its collection of microcode files, ... If you install that program and run it, then go to the "CPUID" tab, it gives ...
    (microsoft.public.windowsxp.hardware)
  • Re: xp sp2 download, computer wont boot up now
    ... > It hard-hangs if Prescott is insufficiently rev'd (by BIOS). ... Both might do microcode updates. ... Now can anyone report that update.sys ever changes the 'CPU revision ...
    (microsoft.public.windowsxp.perform_maintain)
  • Re: xp sp2 download, computer wont boot up now
    ... >> tests to see if the required microcode revision level is present. ... It hard-hangs if Prescott is insufficiently rev'd (by BIOS). ... really "needs newer CPU bugfixes". ...
    (microsoft.public.windowsxp.perform_maintain)