[nikhsub1]

On the 4 or 5 P4's (Northwoods) that I removed the IHS i lost about 5C on average each time. The interesting fact, is that I had a board (Abit IC7) which read the diode of the CPU and did a fairly good job. It was always claimed the Abit's read high, whatever. Anyway, I had in my possesion a WW, Cascade, Cascade SS and a modded TC-4 to fit 478. With the IHS on on one particular CPU, I mounted each and every one of said blocks. Not just once or twice, but at least 3 mounts of each. Guess what? Every single block registered the same exact temp, it made no difference whatsoever. The flip side to this excercise was the removal of the IHS and remounting each and every block again. Now, without the IHS distinct differences could be accounted for. Without the IHS the TC-4 was somewhere around 4-5C worse than the WW. The cascade better than the WW etc, etc. How could this be? So, if one performed a test with the IHS on, well sure, all blocks will perform the same because the IHS > TIM etc numb the cooling effect considerably. Does this mean the TC-4 is just as good as the Cascade? Hell no.

This is the key to the huge disagreements around here. Some say, test as a real CPU with crap IHS. I say doing so will get us no where. Just look at Lee's test, it proves the point 100%, I don't see how everyone can't see that. I will say that Intel's IHS > TIM is far superior now to AMD's. How much so I can't say because I can't get the damn thing off, that is how well it is affixed. Most of you know that taking it off means ripping the core in two. To those that think testing on sims is a waste of time, I would suggest that testing with an IHS to judge WB performance is just plain silly. I think it is fair to say that if you are using an AMD IHS then just about ANY block will perform the same, from a Maze 3 or 4 to the Apogee to the Storm. Test directly to the heat source, and it creates a clear picture on how the block actually cools.