rge over at XS has done some thorough testing. Intel says that the E8x00 Dual Cores have a TjMax=100C. When the DTS sensor reports zero, my IR thermometer shows 95C. That implies that there must be a 5C gradient between what I'm measuring on top of the IHS and the actual core temperature.
It makes sense that the core is slightly hotter than the temperature I'm able to measure on top of the IHS so I've adjusted TjMax to 100C for the E8x00 CPUs.
Core Temp originally guessed that it was 105C and RealTemp guessed that it was 95C. The truth, according to Intel, is half way between those numbers.
At the last IDF conference, Intel released TJMax specs for the 65nm CPUs. In my opinion, those numbers were WAY out there so they've contacted the programmer of Core Temp with some updated numbers. I don't agree with this newer version of the truth either.
For my E6400 B2, Intel originally stated that TJMax=70C. Now they have bumped that up to 80C. My opinion, based on testing and the fact that a 5C gradient exists between IHS and core, is that the truth is probably closer to 90C and that's what RealTemp is presently using.
When DTS=0 for this CPU, the IHS is at 85C and to me that implies that the cores are at 90C. Heat transfers very quickly from the cores to the IHS. If I were to use Intel's latest TJMax numbers then my core would be reported at 80C even though the IR gun on top of the IHS is showing 85C. That's just not possible.