Forms of ILP have been around much longer than 2006 e.g. branch prediction. The main industry "event" that concluded the "frequency wars" was the commercial flop of the Pentium 4. Modern high-performance CPUs are not a commodity industry - due to the requirements for a high IPC, only Intel, Apple, and AMD can really claim to have "industry leading" products.
> "Forms of ILP have been around much longer than 2006"
You're absolutely right, but what I'm trying to get across is that the CPU design problem has largely shifted from being capacity-based to being efficiency-based, and this happened pretty silently in 2006. Sure there is still innovation capacity in the CPU industry, but it's qualitatively not the same kind of approach which evolved the Apple 2 into the iPhone.
Personally I look at the energy efficiency of the human brain and think there's a lot of performance improvement to be done for synthetic processors, I just don't think major developments will be as easy, fast, or predictable as they were before individual CPUs kind of stopped getting faster.
