On 19 Dec 2006 13:10:29 -0800, "Craig"
<caheaton@netzero.net> wrote:

>I thought about trying a memory tester, but I suspect it will simply
>blue screen on me during the test (it's consistent about blue screening
>as soon as it has warmed up). Good tips about the PS rails. I'll try
>reseating everything when I start to remove hardware.
>
>The computer has been overclocked in the past (video, cpu and bus), but
>at the moment it is running at stock speeds and the memory (and the
>video card) are actually running a bit underclocked. (Card came that
>way from the manufacturer that way and as for the memory, I purchased
>DIMM's to use on a Athlon 3200 but ended up building it with a 2500.
>(With the intention of overclocking to 3200 speed, but it was never
>entirely stable at that speed, it would blue screen on me at bad
>times...like when I was winning during a game! A rarity for this old
>man. :-) ) Besides, it had been running rock stable until just
>recently. The problems happened suddenly, as though someone had
>"thrown a switch" and turned on the blue screens. This is why I'm sure
>it is a hardware failure of some sort.


Actually sudden changes are more often software, hardware
more gradually degrades if it's still working as much as it
is/was.

Take the cover off and check all fans, and clean out dust if
warranted. Leave cover off and point a desk fan into the
system and see if it helps. Inspect capacitors for bulging
or venting.

Memory can easily become less stable if it overheats, but
it's doubtful the system temp is changing THAT much unless
you had a serious cooling problem. The PSU could be having
trouble, but more significant at this point might be
examining what those Blue Screens state... what's the error
message and stop code? It could be a great hint, in
addition to checking Event Viewer.

Don't think about "but it might blue screen" while checking
memory, just do it... besides, we mean memtest86+, not a
windows based memory tester that can't test nearly as much
memory.

If you have a multimeter, check and monitor voltage that way
and see if you can cause the fault, perhaps running a gaming
benchmark like 3DMark (whichever version is appropriate for
your particular video card so as to not be bottlenecked too
much, allowing a good load on both video and CPU.

Is it possible you have AC electrical problems, perhaps a
heater is coming on and your PSU is too anemic to cope with
this? Perhaps the system is on the floor next to a heater
duct that came on and elevated temp a lot?

I doubt your chipset is overheating if the rest of the
chassis is reasonably ventilated, especially with the CPU
fan blowing over towards it. However, some software monitor
programs may be able to show it's temp, it would be good to
more closely monitor voltages and temps, and while the cover
is off as mentioned above, recheck all mechanical
connections such as cards, cables, memory, etc.

"Rarely" it's possible for an open core CPU like an Athlon
XP era to dry out it's thermal compound and appear cool
enough, but to have an island of compound over the area with
the thermal diode embedded but no 'sinking of heat over
other areas of the CPU... so the temp looks ok but some
parts of the CPU are hotter than others. What I'd just
described I would normally consider so unlikely and remote
to be unworthy of consideration- but I had just this
situation occur a few months ago, and after pulling off the
heatsink, putting fresh compound on it was fine thereafter.