One problem the AMD/Microsoft patch/driver is suppost to help with is, windows has no idea what a module is. It just looks like a 8 core. If something has 2 or 4 threads it will double it up on a module. Better performance would be to spread it out with one thread on each module. But from what I heard it only jumps performance to around 4-10% on multi threaded. Thats ok, but not a game changer.
AMD should have built it to where a single thread could be run on both sets of pipelines on a single module. Would have really (in my opinion) bumped up single threaded performance. They could have had 4 threads being run on 8 sets of pipelines, damn. Windows thread scheduler would have to shift threads around if you started hitting more then 4 threads. But considering AMD have been designing this thing for 4 years, I'm sure they could have worked it out with microsoft.
But I dont know, since they are already having a problem with the scheduler. What were they thinking at AMD, I mean in 4 years nobody in AMD though, you guess we might ought to contact microsoft about a new processor driver. Your fired, your fired and your fired.