If you find yourself in disbelief, ask yourself: How could a temperature probe on the outside of the CPU (socket temp) measure a HIGHER temperature than reported core temp? The heat SOURCE is the core of the CPU, as heat dissipates from that location, the temperatures steadily decrease. (When thermal energy is SPREAD OUT temperatures are REDUCED). The only possible way, for a socket to report higher temps than the core, is if one of them is WAY OFF from reality. The AMD core temp reporting is nothing more than a relative concept within the confines of an unknown algorithm.
My core temps are always higher than my socket temps after 45C (when the internal diode becomes meaningful) because I have a 92mm fan on the back of the mobo with lots of cool air behind it and a fan blowing on the vrms thats angled also towards the cpu socket. I believe the reason most others have higher socket temps is because of the trapped hot air around the motherboard sensor thats under the cpu. Fortunately, the 1/16th inch fiberglass mother board has a R factor of way less than 1.
But youre correct, neither the intel or amd diodes are calibrated instruments and are relative concepts within the confines of an unknown algorithm of an unknown actual temperature. And we dont know who is being more conservative with their Tjmax numbers, heh.
1,414 °C - Googled it
A better question is the melting point of the epoxy glue used to laminate the fiberglass in the motherboard. Most of the epoxies Ive looked at, that are reasonable in price, dont have a very high temperature threshold. Also keep in mind this is Chinese epoxy glue. 150C (and maybe less) and you may start to get some delamination.
And that just makes for a dandy time while overclocking, heh. Edited by cpmee - 3/27/14 at 7:27am