Craig <caheaton@netzero.net> wrote

> My computer (a three year old homebuilt Athlon based PC) just
> recently starting giving me random blue screens. These only seem to
> occur when the pc is warm (problem begins about half an hour after
> turning on the pc, with pc undergoing light usage (office apps, email,
> etc.). The problem first exhibited itself, however, during a game of
> Doom 3. After warming up, the computer is practically unusable, with
> blue screens occuring immediately after logging in (and sometime
> before). If I let the computer sit for a couple of hours, all is fine
> until it has warmed up again.

> I've checked my CPU temps and they are same as always....around
> 49 degrees C (around 53 C after processor intensive activities). The
> voltages on the power supply all appear okay according to a utility
> that came with the mother board. I've added no new hardware nor have
> I downloaded any updates recently. The problem persists even with
> the pc booted into safe mode.

> I'm beginning to suspect that one of my DIMMs may have gone bad,

Unlikely, but easy to test that possibility with memtest86 run overnight.

> there is a problem with the power supply,
> or with the chipset on the motherboard.

Or bad caps on the motherboard.

> (Although I do have it undergoing additional cooling by
> offsetting the oversize fan on my cpu such that it blows
> onto the northbridge chip's heatsink. But if it was the RAM,
> shouldn't the problem also occur when the pc is cold?

Not necessarily. RAM doesnt usually go bad, the
problem is that the timing used isnt suitable for the
RAM and that stuff can vary with the temperature.

> The fact that it only occurs when warm makes me wonder
> about the power supply (despite the nominal voltage reports).

Yes, glitches on the rails wont usually be visible in the voltage reports.

> The chipset would be scary...no way to test this without
> purchasing a new board and if I did this I may was well
> scrap the pc alltogether and build a new one!

No reason to indicate that thats a problem.

> Anyway, this is a tough nut to crack and am wondering
> if anybody else cares to shed insight into this problem.

You've just got to try the possibilitys, obviously with the easiest to try first.