I thought its pretty intense for a student to ran all these bench tests. So I came over and post the links (yes, it's plural, and in traditional Chinese, in 3 different parts, reviewer obviously put a bit of man hour into this) Maybe you guys like to stroll over and take a peak?

http://forum.coolaler.com/showthread.php?t=172235
http://forum.coolaler.com/showthread.php?t=172385
http://forum.coolaler.com/showthread.php?t=173136

Hey John, Long time no see. Are you interested in play this too?

Frankly, I haven't done any detail performance comparison tests like this one for a long while now. Got to respect the diligent from the kid and his hard work.

[jmke]

thermal compound testing is hard, Keith spend several months on this review:
http://www.madshrimps.be/gotoartik.php?articID=635

and the outcome doesn't really show a very large difference between worse and best thermal compound. First off you have the "curing" period which is different per past, per application, per CPU, per cooler used. Than there's the application method, the thickness, the constant ambient variable which can lead up to 2°C difference between test runs without noticing (-0.9°C and +0.9°C);

I pretty much gave up on those tests as the material needed in a room able to keep constant temps at all times, is way too costly, and home results are hard to be repeatable, and the end results is so small, that a different thermal paste won't mean the difference between stability and crashing. If your CPU overclock is so "on the limit" that a few degrees higher load temp causes problems, you should consider loosing a few Mhz on the OC in favor of stability, with applications and current gen CPUs a few 100Mhz CPU speed up/down is hardly noticeable, except in pure CPU related benchmarks.

[geoffrey]

The higher the heat output, the higher the thermal conductivity will play a role; for modern state-of-the-art processors you won't notice much difference at all, Arctic Silver only gives me a better feeling