Consider that 70% of 4770k samples can hit 4.5ghz, and a 4.5ghz Haswell easily matches an IVB hundreds of mhz higher. I've seen benchmarks of a 4.4ghz Haswell exceeding 10 in cinebench 11.5, which oddly enough is higher than what my 3770k gets at 4.7ghz.
I just don't get some folk's mindset on this. Haswell is not different than IVB, they are both temperature limited - I really feel that some have rose tinted glasses or skewed memories of what overclocking on IVB is *really* like. People are so focused on trash talking Haswell that they forget that IVB really was not THAT much different. It's not all rosy like some are suggesting here, it's not like every 3770k on the planet is going to hit 5ghz with ease. Or even 4.6-4.7ghz. Give me a break. IVB gets hot as well, let's not kid ourselves.
Aside from this, there is the fact that Prime95 increases Haswell's offset voltage by .1V. The temps you see in Prime95/IBT will *not* be replicated anywhere else, you can be certain of that. I think a lot of folks are approaching 4770k stress testing wrong, because prime95 and IBT have not been updated for Haswell yet. Haswell, unlike IVB, increases the offset voltage by a whopping .1V with certain AVX type instructions which only occurs in prime95 and IBT. Even if you get ridiculously high temps in prime95, you will definitely NOT get those same temps anywhere else.