Drunkenmaster, die size may dictate manufacturing cost, but performance dictates pricing. If the new cores are say 20% faster clock-for-clock than Ivy, then theres not much reason for Intel to up their mainstream core count (well unless Piledriver turns out to be something special). Yes, i'd love an eight core Intel chip for under £300, but TBH I can't see it happening any earlier than 2014 or even 2015.
Did the 2600k cost roughly 30% more than the previous top quad core i7, no. Is the 2500k cheaper than many of the previous gen quad cores, yet faster, yes.
Was the 6970 more expensive than a 5870, was a 480gtx more expensive than a 285gtx, was a x6 more expensive than the previous x4's. no, no, no, no and no is the answer.
Performance does not, and has NEVER dictated price.
Intel have had chips at roughly £250 and £150, for a decade, these are price points they can sell chips at. What those chips can be is determined by die size, they won't sell a 500mm2 core at £150, and they won't see a 80mm2 for £300.
Octo cores would be unviable with Intel's target margins till it hits a certain size.
As I said, the general thinking was that Haswell would be a relatively similar core design to now but move to an octo core, even with a slight bump per core.
Remember a dual core was a MASSIVE leap over single core performance, and a quad core was a massive leap over dual core, yet one of their first quad cores, the Q6600, was the same price point as the current 2500k for most of the time it was available.
Core size increases, cache is a big problem(roughly half of the 8 core bulldozer is the 16mb cache) as you move up cores as the only real way to solve keeping 8 cores full is better AND more cache(something Bulldozer only got half right).
Anyway, octo core, 4 "fat cores", 8 "narrow" cores, 16 ultra narrow cores or 2 mega stupid wide cores, I have no interest in PR, but how it works. The 2 generation shrinking of die sizes of quad cores should be leading to fitting a huge increase in cpu power soon, be that 4 fat cores or 8 narrow cores.
If a die is 250mm2, then changing the 4 cores to 4 fat cores that maybe adds 30-40% to the die size, would make a chip that is too big for Intel to get the margins they want. They could sell bigger chips at higher costs but Intel wants a £150 and a £250 chip because they'll sell 1000 x more £150 chips than they can sell £350 chips. Haswell was always touted at a very big leap in performance.
As I said, being quad core doesn't prevent that, but 8 Ivy cores seems a pretty natural move over 4 much more complex cores. If as said we end up with another quad core with a minor bump in IPC then, meh.
By your logic of, if they can improve single core IPC, well, there would never have been a dual core, or a quad core, or a hexcore. Single thread IPC now is massively ahead of p4 single thread, or Conroe single thread, yet Intel still went to quad cores, and with higher IPC and gave us pretty much a doubling in performance twice before(well 3 or 4 times before now).