Although there has been some debate, this is how I think it shakes out.
Temp one = cpu socket temp (cpu temp in AsusProbe)
Temp two = motherboard temp in AsusProbe (often mistaken for chipset temp)
Temp three = cpu diode temp
I'm sure temp three is reading the on chip diode from the cpu since that temp rises and falls very quickly, based on cpu load. At first I wasn't sure why it reported lower than temp one, which Asus says is the cpu temp, but I think I've got that figured out now.
I noticed that the cpu temp reported by the DFI board is lower than cpu temp measured with an external probe, and so is temp three on the ASUS, even though both respond like you would expect from a diode on the chip. By this, I mean that the temp rises very quickly when cpu load increases and falls very quickly when the load is decreased.
I think the reason the diode temp seems too low is that it's not taking into account the thermal offset for the cpu. If you check in Central Brain Identifier,
cbid750.zip , click the Mobility button followed by the P_States drop down button, it shows the thermal offset for your cpu. The thermal offset for each cpu is different and mine is 13c. If I add that 13 degrees to the temp reported by temp three in speed fan it seems to be a very accurate measurement of cpu temp. If I leave the cpu at 100% load long enough temp one will be just about equal to temp three + 13c offset.
Since I believe it to be more accurate and definately quicker to respond, I have setup temp three in Motherboard Monitor for my cpu temp monitoring. Motherboard Monitor has a temp sensor compensation option to add the thermal offset and display it accordingly.