Before the giddiness gets too out of hand, Intel’s move to TSMC isn’t about producing better chips, it’s about muscling away production from AMD and Nvidia over the next decade or so.
I find it amusing how my nearly a decade old, 22nm CPU stacks up against the newer stuff - at 4.4GHz which is a mild overclock - brings it inline with 2000 series Ryzen and newer Intel platforms:
https://valid.x86.fr/u28wd2
This CPU is good for at least 4.7GHz (which brings it a hair behind the 3600) and probably more but I don't have the cooling currently, with the 3070FE venting hot air in the CPU area
, to sensibly run it above 4.4GHz for 24x7 operation.
You may find benchmarks around the net with the 1650 not doing so well but that is usually the V1 or someone running the V2 on bog standard 1600 or 1866MHz memory which holds them back.
EDIT: Someone's at almost 5GHz
http://valid.x86.fr/361enm
Well, technically it is the very same architecture. Only the marketing gives new names (lakes) and the internal widths are increased (few new instructions - AVX2, FMA3) but the changes end there.
Ivy Bridge EP/EX is more a branch off - it has some disadvantages when you scale beyond 12 cores and even at 10 cores starts to drop off a bit - but up to 12 cores the architecture still has a lot going for it. The top tier silicon is rather remarkable actually - light years away from the consumer variants like my 4820K before - but so it should be when you are talking a family of CPUs which were mostly in the 1000s of dollars on release.