On Tue, May 27, 2003 at 02:01:19AM -0400, Mark L. Kahnt wrote: > On Mon, 2003-05-26 at 23:37, Kevin McKinley wrote: > > On 26 May 2003 17:28:51 -0400 > > "Mark L. Kahnt" <kahnt@hosehead.dyndns.org> wrote: > > > > > I run lm-sensors on hosehead here, and while the heatsink is smaller > > > than I've found on most other systems, this box has been keeping the 800 > > > MHz P3 at 112C. I don't feel that is bad, as the screenshot I got of the > > > gkrellm sensor monitor indicated that the machine it was snapped from > > > ran with a CPU at over 200C. BTW, those are Celsius temperatures, not > > > hex values ;) > > > > I don't know much about optimum temperatures, but that's well above the > > boiling point of water. Most other people who post temperatures are running > > at about 80C. > > > > Kevin > > Okay, this got me wondering, because when lm-sensors first reported > 112C, it also indicated anticipated temperature ceilings of 52C, which I > thought sounded a bit low. I just spent half an hour digging around on > the Intel website, though, and found that the "thermal spec" of P IIIs > appears to generally be around 80C, and for P4s, around 72C. The > temperature has stayed steady, and the other reported temperatures have > stayed consistent, so I concluded that it must be reasonable, even if > the chip is not overclocked. My thought is that if the temperature was a > problem, it would trend up, both on the CPU and the other box sensors, > which doesn't happen, so I have believed (self-deluded) that it was a > reasonable temperature. There's got to be a bug or malfunction of some sort somewhere - 200 deg C is instant death for a silicon device, 150 deg C is about the absolute max for any silicon device. > That said, there could be a difference between where the temperature is > measured - I don't know exactly where any of the three reported > temperatures comes from, and my luck is such that it could well be > "innovatively" placed by the maker of my motherboard ;) On my MB, unless an Athlon has a sensor built into the chip, the CPU temperature is measured by a sensor in the big square hole in the middle of the CPU socket. It sort of flops about on the end of a short piece of that metallised-plastic ribbon cable such as is used to make connections to moving print heads, scanner heads etc. I think the cable is supposed to act as a spring to press the sensor against the underside of the CPU. I very much doubt that it works. I have seen several secondhand MBs with a similar arrangement; there is never any trace of thermal transfer compound having been applied to the sensor to try and give it a better chance of making some sort of thermal contact. So I suspect that most people's CPU temp readings are well under. Me, I run with the sides off the case and if I want an estimate of the CPU temp I stick my hand in and feel the heatsink. -- Pigeon Be kind to pigeons Get my GPG key here: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x21C61F7F
Attachment:
pgp7Edv_uCPRV.pgp
Description: PGP signature