Wait, so they are going to make Skylake-e(14nm), then cannonlake (10nm), and then go back and do kabylake 14nm 6 core that would have identical performance as the Skylake-e!? Why?
There is no 6 core KBL. Only 6 core CFL. Anyway, the timeline I stated are based on the roadmap, and it ignores server/HEDT parts. You'll see Kabylake soon on U and mainstream desktop chips/enthusiast. Then Cannonlake comes to replace KBL on U chips. Little after that we'll see Coffelake replace Kabylake on mainstream desktop/enthusiast(again, not HEDT).
What happens on HEDT you say? I don't know. I am just extrapolating based on the roadmap shown earlier on this thread. Nothing about HEDT.
It's very messed up. Because "Moore's Law" is dying you see. I guess its more correct to say, "its way too freaking hard to be worth advancing much further". It was known for a time that KBL gets skipped on EP server chips and they go from SKL to CNL. But there's CFL. It suggests Intel's struggling with either 14nm, or 10nm, or even both! Does it suggest that they'll go with CFL EP, or still release it on CNL just later than expected? No one knows. I doubt even Intel, because that's how messed up it is.
And its not bad to suggest we'll see 6 core LGA115x and 6 core HEDT. Why? Because 6 core on HEDT will be entry. That's how it was up to Ivy Bridge. We had 4 core Ivy Bridge HEDT and 4 core Ivy Bridge LGA 115x.