Please keep in mind that if you are measuring CPU temperatures on an AMD processor using an external thermistor, either in the socket or placed next to the core, then your measurements are subject to compression and will not reflect the actual core temperature change effected by a change in the thermal resistance of the primary heat path. This temperature change compression is a natural phenomenon of secondary flow path when there is "both" heat flow and thermal resistance from chip to ceramic surface.
This is the normal/natural case and cannot be avoided.
Here at Overclockers.com, Joe has gone to great lengths to develop measurement techniques that will accurately reflect actual heatsink performance rather than relying on the motherboard's in-socket thermistor.
The following links detail the science and math behind the compression...
http://members.nbci.com/candjac/TtoVanalog.htm
http://mikewarrior.freeservers.com/problempage_4.html
http://www.arcticsilver.com/measurement1.htm
Not only are in-socket thermistors in a secondary heat path, they also only measure an average of the CPU rear temperature (minor contact with thermistor surface) and the temperature of the air in the socket (major contact with thermistor surface) which is itself an average of the CPU rear temperature and the motherboard temperature.
With both of these factors contributing to temperature change compression, the differences between different cooling solutions, be it heatsinks or thermal compounds is minimized. In most situations, this compression can be as severe as 4 to 1 so a 4 degree difference in actual core temperature is only measured as a 1 degree difference.
Then there is the problem of heatsink contamination when one compound is switched for another.
Isopropyl alcohol and acetone will not begin to remove all of the existing compound on a heatsink. The fibers on a cloth or paper towel are far too large to fit into the microscopic valleys that the thermal compound fills so once a compound is applied or a thermal pad is melted, some of that compound or pad will be on that heatsink forever. Any subsequent compounds will be --at least partially -- applied over the earlier thermal material.
Please see this link:
So if you have had compound A on the heatsink and you switch to compound B, you are not really comparing A to B, you are comparing A to a mixture of A and B. If B is better, you may see some improvement, but probably not as much as if you had put B on in the first place.
Heatsink engineers understand this. Several manufacturers of high-end heatsinks apply Arctic Silver to the mating surface of all the heatsinks they send out for review then wipe it off. They know that if the heatsink is tested later with a lesser compound, it will have a bit of an advantage because the AS will still be filling in most of the microscopic valleys.
Nevin House
Arctic Silver LLC
