On Thursday 10 July 2003 09:51 am, Tom Brinkman wrote: > On Thursday July 10 2003 09:19 am, Jack Coates wrote: > > Is it just me, or does 205 to 230 degrees Fahrenheit seem a bit > > excessive for the maximum temperature of a desktop? Yikes! I'm > > nervous enough about the operating temperature of 122 to 158 > > degrees Fahrenheit. > > > > > The AMD docs I read said 90 to 95C internal core is the > > > failure limit. They also said to add 10 to 20C to reported > > > probe (thermistor) cpu temps, to approximate the internal core > > > temp. > > No, I agree. 230F = 110C is a temp I've never seen in AMD docs. > Even 203F = 95C seems unbelieveable. But that is the max temp AMD > specs for processor failure. I think they mean permanently fried ;) > IMO, AMD's should be kept under 60C core (< 50C from a probe) at > extreme load, or you're gonna see heat related problems. Lettin 'em > run hot causes a gradual degradation of the core.
I had a system that was experiencing periodic lockups recently. Turns out the cpu temps were creeping up to around 78C. Improving the case cooling brought the temp down to around 63C under load. Still not ideal, but the system is in a safe and has limited cooling. I may have to try a different heatsink fan combo. This is a Tyan dual cpu mb with 2100+MPs. Obviously reliability fails well below 110C, at least for this system. > > I only wonder why recently AMD began internal diode temp sensing > for the core. Something even the first Pentiums have had all along, > including good motherboard support for it. From what I've read, the > few recent AMD boards that can read the new AMD diode, don't do it > very well. I believe that's why Intels are the preferred cpu's for > servers. Anyone know which mbs have the core temp sensors? -- Thomas K. Gamble Los Alamos National Laboratory Advanced Diagnostics & Instrumentation (C-ADI) p:505-665-4323 f:505-665-4267 MS-E543
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
