On 2005-07-19 wolfgang+gnus20050719@dailyplanet.dontspam.wsrcc.com wrote:
> Patrick Welche <prlw1@newn.cam.ac.uk> writes:
> > I would like to just run the chip at 1800MHz under NetBSD, watching
> > viaenv(4) and hope to say "it's fine, it's just an overzealous
> > windows probe programme", however, it could be that it really is
> > getting hot, so do I risk frying the CPU?

  Windows idles hotter than NetBSD.  In my tests a long gcc compile
increases the CPU temperature more than computational heat generating
utilites; IIRC, this can have up to 15 C effect.

> What does a real thermometer say when you measure the heatsink?
>
> Athlon32's don't seem to have a very reliable thermal sensing system.
> My old athlon system says its 50 C when the heatsink (and cpu!) are
> cool to the touch.  I've just learned to mentally subtract 15 C.  Your
> sensor may be more whacked out than mine, so it might be good to get
> an independent reading with a known-good thermometer.

  If your sensors are are connected via lm, see port-i386/18205.  With the
patch, the reported values are usually less than 5 C above heatsink
measurement.  In my system, the chipset always stays a bit hotter than the
CPU.

Matthew Orgass
darkstar@city-net.com