don't think many people care too much about power consumption regardless of how much of a big 'against' point it apparently is at the moment, considering half the people on here are running monster systems with Crossfire and SLI, with 2500K/2600K clocked to their maximum with triple monitor set-ups and what not.
but indeed, the performance increase I wouldn't expect to be above ~10% in some situations, such as gaming where hopefully the operating system will dedicate modules to these power hungry gaming threads rather than spread them across a module, that alone can be worth 10 - 15% depending on the coding, which isn't to be sniffed at to be honest. in other situations it'll probably make no difference in the slightest and in others it could have a slightly negative impact on performance though wouldn't expect that to be notable difference.
things like that though, potentially 10% boost in performance in some applications where the processor is lacking, essentially for nothing is the sort of thing that goes to show that at least some of Bulldozers problems are entirely software related, granted there are indeed a number of hardware problems like the painfully slow L2 and L3 cache, fix that and Bulldozer would likely be a proper contender when you take into account that ALU performance (measured in Dhrystone) has improved over K10.5 and FPU performance (Whetstone) has also improved, has to be a bottleneck somewhere in the processor, cache could be the likely culprit.
Edit: upon reading some word on the web, apparently gains are there. smoother gameplay and such being the most apparent ones, another interesting thing (which backs up my argument from day one of it
not being an eight-core processor) is the fact it now shows up as 4C/8T like a 2600K in Windows. so there we go free performance boost from a simple scheduler update, as a certain supermarket say 'every little helps...'
