This is not advised. But I took a razor and cut the heat spreader off the cpu. On the P4 the heat spreader is held on by silicone ( or something like that ). The heat spreader is the 1.5in x 1.5in metal you see on the top of the P4. Underneath that is the CPU heatsink ( .5in x .5in ). There is a small amount of thermal paste between the cpu heatsink and the heat spreader. The rest is open air space, so there is a gap around the cpu heatsink between the heat spreader ( except for the .5in section ). I epoxied the sensor to the actual cpu die, next to the ( .5 x .5 ) cpu heatsink. Then after it was dry, I filled the heat spreader with epoxy and pressed ( under pressure ) the 1.5in heat spreader back on to the cpu die. So now I have a sensor permanently imbedded "in" the cpu as close as possible to the source of the heat. I know that the layer of AS epoxy ( around the cpu heatsink ) is rather think, but I figure it has to be better than dead air.
The temp change was minimal ( maybe -1*c ), but now I have a true CPU temp. NO secound guessing. When I open a program, say IE, it raises ( around 8*c ) for about one secound, then it drops back down. I mean instantly!
All of this has nothing to do with the heatsink attached to the motherboard. I think Im wrong on the 1.5in stuff. I think its actully 1.25in. I dont have a tape measure in front of me though.
I am pretty sure my BIOS temps are way off. It kind of sounds like yours are too, so I wouldnt trust the BIOS temps in the versions that are out so far ( ive tried both ). Better safe then sorry.
I have gotten better test results ( in cpu performance ) by changing the memory timings than just overclocking. Now my tests are much better, at a much lower FSB. Memory performance is a big part of CPU performance.
"This is my opinion and my not be expressed by others"