Are they really throwing more hardware at the problem?
The die size for the whole M1 SoC is comparable to or even smaller than Intel processors, and the vast majority of that SoC is non-CPU related stuff, the CPU cores/cache/etc seem to be at most 20% of that die - though a more dense die because of the 5nm process. This also seems to imply that the 'budget' of number of transistors for that CPU-part of the SoC is also comparable to previous Intel processors, not a significant increase. (Assuming 20% of the 16b transistors in M1 is CPU part, it would be 3-ish billion transistors, and the Intel does not seem to publish transistor counts but I believe it's more than that for the Intel i9 chips in last year's macbookspro)
Perhaps my estimates are wrong, but it seems that they aren't throwing more hardware, but managing to achieve much more with the same "amount of hardware" because it is substantially different.
Well you don't have any of the AVX512 nonsense, but probably the OOO of the M1 uses more transistors than on an Intel chip. And Google tells me that a quad core i7-7700K has 2.16 B.
The die size for the whole M1 SoC is comparable to or even smaller than Intel processors, and the vast majority of that SoC is non-CPU related stuff, the CPU cores/cache/etc seem to be at most 20% of that die - though a more dense die because of the 5nm process. This also seems to imply that the 'budget' of number of transistors for that CPU-part of the SoC is also comparable to previous Intel processors, not a significant increase. (Assuming 20% of the 16b transistors in M1 is CPU part, it would be 3-ish billion transistors, and the Intel does not seem to publish transistor counts but I believe it's more than that for the Intel i9 chips in last year's macbookspro)
Perhaps my estimates are wrong, but it seems that they aren't throwing more hardware, but managing to achieve much more with the same "amount of hardware" because it is substantially different.