bumpity bump, for those interested
I found this in the overclocking guide on Anandtech
Originally Posted by Anandtech Guide on overclocking a P5E3
ASUS decided to surround the LGA775 socket on all four sides with a generous portion of copper heatpipe-pierced heatsinks, in effect thermally adjoining the memory control hub (MCH) and both sets of power MOSFET banks. The portion of the cooler located between the CPU socket and memory DIMMs is "floating" above the board with a thin layer of electrically insulating material isolating it from the components below. As is usually the case, the stock mounting solution does not provide enough clamping pressure, resulting in increased thermal resistance that usually leads to higher component temperatures than would normally be experienced. We noticed this same problem on the abit IP35-Pro, although the ASUS design did provide better overall cooling.
The solution to correcting the lackluster mounting pressure was almost effortless (making us wonder why ASUS didn't make the extra effort and save us the trouble). We simply removed any hardware holding the heatpipe assembly to the board, cleaned all thermal surfaces with rubbing alcohol (making sure we allowed ample time for the board to completely dry), and then remounted the unit with a fresh coating of our favorite thermal paste. This time we used small machine screws and threaded nuts in place of the plastic pushpins. This allowed us to control the loading pressure, ensuring a secure bond between the copper and the heat sources. Looking at the picture above, please note the black replacement screws used to hold everything together. In the end, the effort we invested in improving our board's cooling efficiency resulted in more than an 8C drop in measured MCH load temperatures.
Now I found this quite interesting as both this board and my previous one (P5K) had what I consider to be high temps, usually around 30-32°C. (In fact the day I installed my P5K was not exceptionally warm ambient air temp wise, but on first boot the board reached a pants crapping temp in the low 40°C.
Having read this from the guide and inspecting my board it does indeed appear that the thermal tranfer is hindered by the method of fixing (more push pins
the bane of modern pc's at present imo)
Now an ~8°C drop would be awesome, hell I'd be happy with 5°C lower so here is something to consider doing next time I get the board out of the case (which shouldn't be far off, lol)
In other breaking news, I have sucessfully run my cpu with 1.08125 vCore, this has reduced temps by ~1°C, 100% Prime 95 small FFT stable (4 hours)
Memory will run at 6-6-6-18 (I had the voltage on auto this time as it was a last minute "try it and see thing"
I'll do manual next time) I've also had the memory at 1333MHz firstname.lastname@example.org so all is good.