I possess the rare ability of being able to tell when people are being purposefully obtuse over the internet. The OP isn't looking for an actual answer, he is just trying to get people to believe in the ideas that he himself wants to believe for whatever reason.
It's really no matter. We all know that while Haswell isn't the speed king we were all hoping for, it does edge out Sandy Bridge in pretty much all but a few benchmarks when both are overclocked.
I think a lot of people forgot how few Sandy Bridge chips actually hit 5Ghz. Would you guys agree that the number is around 10%? The number Asus gives us of HW chips that are able to do 4.8Ghz is around 10%.
Now this is true enough. There are mostly minor but real gaps when both are OC, this is particularly true in synthetics and in the apps that take advantage of the improvements in IPC that Haswell offers. For pure gaming, all three are a toss-up.
Indeed 5Ghz is not common with SB, just like 4.7 isn't common with Ivy and 4.5 not common with Haswell. However, they are all typical top-end numbers with safe volts and good air cooling. I had two 5Ghz SBs, a 2500k and a 2700k, my 2600k topped at 4.9Ghz, needed too many volts to hit 5Ghz in a safe manner for my purposes. All IBT/Prime stable, first I shot for 24hr, then realized that was insane, and went for 8hr (mainly because if it's stable to 8hr it will be stable to 24hr just as easily). For a while I would crash randomly between 1hr-8hr, and finally found it wasn't the CPUs, it was a ram stick that had gone bad on me
Silicon lottery makes things less clear. If you get a poor SB, then IB/HW look like better leaps. If one had a great SB, but went to an unlucky Ivy, then it is tempting to go back (this happened to be on three successive Ivy 3770ks, 1 DOA, two mediocre ones).
Probably watercooling + very good lapping helps lessen the lottery frustration, though then you run into yet another barrier of the TIM lottery. As we've seen from great work from others, the TIM can really vary in luck as well.
I don't want to be seen as denying the actual FACT that Ivy/HW have real, measurable, and undeniable IPC improvements. Nor that Haswell for apps that match well, beats even those IPC numbers (along with the rare app that doesn't really move the needle). Nor is it deniable that stock v stock (eg; 99% of the actual desktop world in it's entirety) IB/HW were jumps forward in performance, even if not huge for the most part. And it's particularly not even possible to debate the fact that the bare-die TIM/IHS-less mobile Ivy/Hw parts are
superb improvements from the 32nm SB.
In the pure and limited context of OC v OC gamer CPUs, however. SB/IB/HW = meh. Pick one. They're all good. Depending on luck, any one of the three could end up the best, by margins almost always less than 10% due to the mix of IPC vs. max safe clock ceiling. If my 2700k was a 4.6'er instead of a 5 club, a 4.6 Ivy would look really nice, ditto a 4.4 Haswell. On the same note, get a bum 4.3 Ivy or 4.2 Haswell, and you're faced with the dilemma of whether to try again. Reasonably speaking, 4.3 Ivy and 4.2 Haswell are low range typical OCs, but STILL crazy fast in the big picture. It's not like any of the three will fall flat in gaming whatsoever once in the 4ghz range.