???
I don't know what's leading you to this conclusion, but the temp could be
off by a thousand degrees and still crash when processing heavy loads.
The temp offset could be wrong but the granularity of the change could
still remain constant. In other words, if the temp reports a 10C change,
it might actually be a 10C change, but the true temp at both periods might
be off by (whatever, a few degrees).
The thing is, electricity is converted to heat. CPU temp rising is a sign
that the motherboard and system power supply are also working "harder".
For example, if the voltage were too low, or current variation response
too slow, when the CPU gets warmer it'll be more susceptible to these
types of problems. I'm not implying that this is necessarily your
system's problem, but it might be.
This is curious, have none of these solutions been able to keep the CPU
below 60C? There was no point in trying low-end generic solutions, a big
ole copper-bottomed 'sink with an 80x25mm fan on it is the best solution
for that CPU, not some Zalman gimmick. Even so, the Zalman should be
enough to retain stability unless there's excessive case ambient temp.
Are you certain there's adequate air I/O though the case?
I don't know what you mean by "Pentium approved". Does that mean it's the
Retail P4 'sink, or a 'sink qualified by Intel for your specific P4 speed?
An "award-winning" Zalman 'sink doen't necessarily mean a whole lot,
without knowing what award and how it won... marketing is everything. I
am sure there are no Zalman 'sinks that can compete with the best 'sinks
like a Thermalright SLK-947, but that's a pretty expensive 'sink, since
the p4 has a heat spreader you ought to be able to achieve sub-60C temp
with a moderate priced cooler. Is it possible the heatsink retention
mechanism isn't doing it's job? You didn't respond about what the
heatsink & CPU looked like when you removed the 'sink, if it appead to
have been making a good contact.
If these heatsinks (or the CPU) look to have a poor finish you might
consider lightly lapping them. Mirror-finish is not necessary but
roughness should be reduced to smoothness.
If it's just as simple as always trying junk heatsinks, then read some
heatsink reviews of the ones you've tried, and if they look to do a poor
job relative to other options, then pick a better 'sink. With your case
closed and the cooling suggestions I make in this post, a decent 'sink
will keep the CPU below 60C.
Where on the side panel?
Generally it's best to have front and side fans blowing in, and rear
blowing out. Of course that can vary.
Huh? The rear case fan is sucking in exterior air, blowing it into the
case? That's backwards, rear fans should exhaust out of the case, blowing
the same direction as the power supply.
Get rid of the stamped in metal grill. With only a single rear aux. fan
you'll need to maximize it's flow.
If it also has a stamped-in-metal grill, you should cut that out too...
High end CPU and video draw a lot of power, there's a lot of heat being
created in the power supply.
Make/model of power supply?
"If it's a generic you might as well just replace it now."
I ought to put that sentence in a sig. since a lot of these generics are
failing as systems continue to consume more and more power. It seems to be
one of, if not THE most common problem(s).
Ok, from this I can realize that it's not case airflow, but I'd still
review/alter the chassis cooling.
Did the CPU temp not change at all with the case cover off + external fan?
It still remained at 60C? Can we assume your room air temp is moderate,
you're not in the middle of a summer heat wave?
You previously wrote that it's less prone to crashing with the cover off.
Is it possibly not even the CPU but the northbridge or memory?
You might visit a motherboard forum like the one at
http://forums.amdmb.com and see if anyone else is having temp issues with
your model of board, and if it's temp report is accurate. If you haven't
updated it's bios you might consider doing that too (just make sure it's
as cool as possible, trying to make sure it doesn't crash while updating
the bios).
After all I've written, two thoughts come to mind. MSI boards seem more
picky about memory to me. You might put a hairdryer pointed at the
memory, not getting it REALLY hot but just a little warmer and rerun
memtest86. Unfortunatley it's not realistic to do that for several hours,
but if all else fails a little while is better than nothing.
Second thought- Again, if it's a generic power supply, just replace it.
