Re: P4 overheat *not* confirmed

[Peter Soendergaard <soender@cs.utk.edu>]

Just to add to Clints report, I have also tried running my tests again,
(both SSE and SSE2) and I also cannot get the same slowdown as before,
even after running the machine in full speed for 1.5 hours.
 
cheers,

Peter.


On Mon, 14 May 2001, R Clint Whaley wrote:

> Guys,
> 
> Just a quick note on the P4 thermal throttling I was reporting a couple of
> days ago.  After those timings, where I was causing it to happen at will,
> I took the machine apart to make sure everything was installed correctly.
> In getting the case open, I made several rather crisp impact adjustments.
> Finally having figured out the case, the heat sink checked out OK, but the
> model I have has an air duct that fits over the heat sink's fins, which 
> channels the air to a case fan.  That duct appeared to be askew, in a way
> that would not block air flow, but would fail to channel it optimally.  At
> the time I put down its misplacement to my rather vigorous case opening.
> 
> Since then, however, I have been completely unable to reproduce the overheating
> problem.  Now, it would seem to be a easy conclusion that the problem was
> obviously misinstalled ductwork.  It is not that simple, however.  The P4
> is in my office.  My office does not possess a thermestat, and while supposedly
> a controlled climate, the temperature varies enormously.  When the overheating
> occurred, the times I was in my office, it was as hot as it ever gets; I would
> guess roughly 76-78F.  I corrected the ductwork Friday, and did my first
> retry Saturday.  Everytime I've entered the office this weekend or today, my
> office has been as cold as it ever gets, my guess is 72-74F.
> 
> So, all this is to say that I still do not know if the P4 we have, when
> properly installed, will overheat in a human-habitable office.  I don't have
> plans right now to figure this out further; it's impossible to do at work,
> where only the gods seem to control the office temperature.  To do this
> experiment correctly, I'd need to take the machine home, and use the
> AC & heating units, along with a thermometer to put this question to rest.
> Right now, it doesn't seem like something I need to do (surely others on
> the web will);  I'll let you know if I learn of anything further.
> 
> Cheers,
> Clint
>