this obviously creates problem as now you can get thermal paste on the transistors...
Psht, not like that problem wasn't there before.
A couple interesting things to consider about the supposed 'fragileness' of the cores. As one dude pointed out in another forum I saw, the old socket A mobos had that gap in the middle right below the core. That made it fairly easy to crush the core. This time it's a smaller CPU and it's completely backed by the socket so in theory it shouldn't be as hard to crush. Beyond that it wouldn't make sense for AMD to just all of a sudden make their cores more fragile simply because it has a heat spreader on it.
One thing that's true about the heatsinks though, is that you can only use the bolt through style heatsinks like how water blocks are. Any that use retention clips can't be used.
Gaara - Mine is doing 305x9 right now with 1.4 + 113% but I'm pretty sure it doesn't need that much. For me it's reading as 1.52 in CPU-Z and ITESmart Guardian, but it reads like 1.54 in the BIOS. For 301x9 (for whatever reason my computer HATES 300x9, but 301x9 is ok...I saw this happen to another dude as well)
Normally for 301x9 I'm using 1.375 + 113% which translates to about 1.504v in CPU-Z and that is 14 hours stable in prime95.
I want to remove that IHS and see if I can get 2.8GHz goin on. Might happen, but like you said, I just need to play a game like oblivion for a while and if that doesn't crash I assume I could consider it stable.
However, you will notice a difference in SuperPI times based on stability. For example I'd get like 31 seconds at 2.7GHz, but if it was unstable it'd show like 31.297 seconds and of course in this program that .297 seconds matters.
My RAM is always running at 2-3-3-6 though, I've told you about that BS thing with my RAM and not wanting to work correctly at CL2.5
But yeah......remove IHS, put some electrical tape over the transistors. That's what a lot of people seem to be doing to protect them from sparking against a heatsink. Should also protect from TIM spilling over.