Yes, it is quite impressive how Apple leads with instructions per clock. Any reason why Apple would not be able to apply this to a N1 design? For a Mac Pro where liquid cooling would allow higher clock rates, the raw performance of a 80 core 5ghz should still brutally outperform the A14x in most workloads.Because Ampere IPC isn't even matching Apple's own ARM implementation. They are basically ARM N1, or you could look at Amazon's Graviton 2 and see how it performs. Decent, but not enough for Apple yet.
A 5W A14 only run max clock on one core for 10-20 seconds. Bumping up Clock speed and all core you get much higher Power usage.
I don't know how well the A14x can be scaled on TDP or clock speed. Should not be that simple to "just" increase the TDP to 85w, and have it run at 5ghz.
I'm a software guy, not silicon. So, honestly interested in understanding the options a bit better.