Properly applied, most thermal compounds "should" produce close (not the same though) results, but with an on-die thermal probe reading a difference can be seen. Its not as significant a change as say a delta over a quiet heatsink, but it is there. An external thermal probe or socket thermistor can only measure the change that happens to the body of the chip, not the die. So saying on an AMD all the pastes perform the same doesn't really mean much, when you have that much variance between boards from the same manufacturer with the same parts. And that doesn't take into account the difference between how different motherboard manufacturers measure and how they compress the data.
I wish some review place would do a test for longevity effects of all the different thermal compounds. Lets say several identical Pentium III machines, with the exact same components running as close to 100% cpu usage as possible, over say 2 weeks, a month. Have them run the same exact copy of the harddrive with MBM on checking for variance in the temps in order to see if at a preset time later. That would testing "Quality" of the different compounds and what I would consider real performance. To be fair you would have to blindly rotate all the compounds to different machines to measure the variance of the compound and not the machine.
Major downside of this experiment would be a halving of the actual power used in an AMD machine and the increased potential for thermal degradation that all the heat of a high-end overclocked cpu can produce, It seems a whole bunch more people are doing voltage mods and are using the built in capability of the Epox board. I mean, yes thermal compounds are important to the stock processor as much as the overclocked one, they just aren't as important. My advice (lengthy-ranty *apologizes*) is to use high-end paste, whatever it may be (I personally recommend Arctic Silver II and will be until I see a legit test saying that any other compound makes more of a difference than Arctic Silver II, its just not that expensive or dangerous(properly applied according to the instructions) to justify not using it).