THE REAL TEMPS FOR INSTABILITY

[ivanski]

I read somewhere that the real core temp can be up to 10 degrees higher than what most monitors read..(speedfan,Hmonitor.etc)
I downloaded the "core temp" program and it showed a diference of 10 degrees higher than the others.
The rule is "keep it under 55" and since most monitors are 10 degrees off does that really mean keep it under 65.
AMD states instability occurs above 71 for my 4000+ San Diego.

NO I DONT WANT TO PUT A PROBE IN THE CPU,I'm just asking if anyone knows.

 

[maury73]

Core temp depends on thermal resistance between the die and the diode which measures the temp.
It's not integrated in the chip die, but molded in the external package.
You can't actually assume a constant difference in degrees between the die and the package, because the thermal resistance is measured in °C/W, so it depends on the power the chip is dissipating and this value varies between variuos models and revisions of the same package and CPU.
10°C is a good average value for powers in the range 80-100W.
AMD specification are based on the temp read by the diode, not the real die temp (which can be as higher as 100°C), so 71°C is the maximum temp allowed for stability actually read by a monitoring software, without adding the 10°C difference.