The point I dislike is about people who think that IHS removal is the only key to test WBs, but it represents only a few people. I agree with Cathar about differences between bare die and "IHSed" die for WB techniques and their capability to remove high power density, but we need to take the 2 situations in a same time, not only one to give global answer about WB performance for ex. In my comparative, I said Storm is an excellent WB because it's true but under some circumstances it's not necessary to pay $$ because others designs less costly could do a similar job with actual processors conditions. This statement doesn't mean that Storm is useless, not at all. I worked more than one week each day on Storm/APogee to be sure results were valid on my tesbed because difference gave me a bit of surprise, but I obtained the same on each mount (using differents fixing too to apply more pressure and see the changes), what could I do more ? To enlarge the view, another processor will be used and an update will appear very soon to see differences between a 478 and a 775 Intel, the more controled data we get, the better generally...
To be stubborn on bare die testing only is idiotic. We can talk about that because AMD allows the IHS removal till now, but when their IHS will be permanently attached using solder (when?), what Cathar and others will think about bare die ??? That's a good question IMO... As a lot of things, we need evolution in testing methods, bare dies are the past (for typical use). Constructive criticisms are always welcome if people could show a bit of mind opening, and on the contrary, it's lost time.
Did you see the new Joe's die here :
http://www.overclockers.com/articles1284/ . First approach but not agree on the wood used to applied IHS, too soft for me to ensure a good mechanical stability under pressure (clipping force could be a bit askew with a 4 points fix). IHS should be lapped to be flat too (like Intel said in their datasheets) to ensure no bias between WBs.
Thermal aspects about IHS could be controled if we are a bit conscientious. "Mushroom die" as we discussed in the other thread could reduce variability (TIM1 is an offset, not really useful for measurement : a die =! processor). What we want is the thermal spreading caused by IHS to see changes between WBs. Thermal flux are never the same on WBs with an IHS because of the spreading (bell shape power density, more dependant of the convection coeff on whole baseplate), on the contrary of a 1xx mm²bare die (~flat power density). Storm is almost unbeatable when dealing with high power density on small surfaces (and yes Apogee will be beaten easily here), but others simple designs could be more at ease with big surfaces and less power density (at a same power obviously)...
The final choice about WB is the decision of the consumer : the cheaper one? a Storm because of Cathar? the most bling bling? the smaller/lighter? a good performer at reasonable price? etc. For the price of 1 Storm, we could have ~2 Apogee, and at final the typical consumer won't see any difference at all between the 2 models (who is at full load under BurnP6 each time to see 0.5°C difference? Nobody...)