Charlie,

When I said set the BIOS to defaults, I didn't want the CPU underclocked!
Sometimes the BIOS will do this, but that's an error. I merely want the
*other* BIOS options set to defaults, I don't want to see any of the other
kinds of optimizations people routinely make, I want a very conservative,
safe, dependable BIOS configuration. But sometimes, in the process of doing
that, the BIOS incorrectly configures the CPU. You should correct that
manually (or set it back to SPD). So if it was running 1GHz before, I want
it running 1GHz now!

Later on, in my suggestions, I then wanted you to slowly back off the
processor in 1MHz increments (if your BIOS permits it) so that *slowly* we
could see if perhaps by relieving the processor's burden, we might see some
stabilization. I had an Athlon XP 2000+ and Abit KT7A-RAID motherboard a
few months ago that simply wouldn't run at stock speed (1.67GHz, 133 x
12.5). The system would eventually hang under stress. I then backed it off
2MHz (131MHz) and all was fine. There was either something w/ the
processor, or my motherboard, that couldn't handle the higher speed, even
though it was "to spec"!

Further comments below.

Jim


"Charlie" wrote in message
...
On Wed, 21 Apr 2004 08:18:37 -0700, "Jim" wrote:

There are an awful lot of possibilities, including a bad driver,
overheating, bad memory module, OC'd CPU, etc. When something like this
starts occuring, it's best to return to "ground zero", i.e., make sure
the
BIOS is set to BIOS defaults, no OC'ing!, etc. You don't want to
complicate
matters by trying to diagnose the problem while the system is over spec
(in
fact, that could *be* the problem!), you want as conservative a
configuration as possible.


The board was not overclocked, but when I reset BIOS to the default
values, it reset my processor speed to 750 MHz. I have seen "fewer"
problems (actually only one lockup) since them, but I'm not ready to
say it's because of the lower speed, because I haven't really tested
it.
Once "at spec", do the following.

1) I suggest taking a *very* close look at the motherboard and look for
leaking capacitors (
http://www.pcstats.com/articleview.cfm?articleID=195 ).
Whenever a perfectly good system suddenly starts acting up, esp. if it's
around 18 months old, I start looking for leaking capacitors. I've had
several people over the past year discover this problem, much to their
surprise. The KT7A seems particularly prone. Not likely, but worth
checking. This is easily missed, even by an expert! You often need to be
*looking* for it to even notice it.


Well, as you say, it probably will take an expert to see any bad
capacitors. I looked carefully, and there is nothing obvious, like
leaking or a split can. One does look a bit different from the rest,
though.


I meant that experts will sometimes overlook to *consider* this (the eyes
tend to assume all is normal, despite stairing straight at it), but if bad
enough, it does become rather obvious. At some point, a lot of brown gook
dripping over the motherboard traces won't look normal to anyone! But if
you don't see even slight indications, like bulging cans, then it's probably
not the issue. But it's best to get this possibility off the table
initially, saves a lot of time.

2) Download memtest-86 ( http://www.memtest86.com/ ) and run it through
several passes, overnight even better. Memtest-86 is very good at
finding a
failed memory module. Or sometimes if you OC too much, it will fail too.


This is on my schedule to do, probably overnight tonight.

3) Download Prime95 ( http://www.mersenne.org/freesoft.htm ) and run the
torture test. For my money, any system that can't run 24 hrs. without
errors is NOT stable.

4) Reseat the CPU and HSF. I had a guy install the HSF *backwards* one
time
and had nothing but trouble.


By "backwards" I assume you mean that air was blowing down, not
upwards from the cpu heat sink? Mine seems OK but I haven't removed
and reseated. I did blow out all the dust that had accumulated.... but
I did that over a month ago.

On AMD systems, the HSF usually has an orientation, a "lip" that has to be
oriented correctly to the lip of the ZIF socket, or the heatsink will be
"tipped" slightly when installed, and thus not making full contact w/ the
CPU. The lip is not very large, but large enough that if installed
incorrectly, it will cause problems. Looking straight down from above, you
might not notice this was the case, it *could* look correct when in fact it
was not. It's more obvious when viewed from the side. Granted, not a
likely problem, but I have seen this happen once or twice.

Later...

Jim

If all appears OK, return to normal usage of your PC ("at spec") and see
if
the problems continue. If so, then go into the BIOS and *slightly*
underclock the FSB, in 1MHz increments, maybe down as far as 4-5MHz max
and
see if stability returns.

If all these measures fail, you probable have some driver issue.

I'm wondering if somehow this is the problem. Although I've been using
the same drivers for a couple of years now, many were recently
re-installed when I reloaded Win98. If I don't find anything else,
I'll completely reinstall Windows and see it that makes any
difference. I can just swap the drives around and compare a new
installation to the existing one fairly easily.

HTH

Jim

Thanks for your suggestions...

Charlie Hoffpauir
http://freepages.genealogy.rootsweb.com/~charlieh/