So, compare the power efficency. M1 idles at ~6w + 2-5 watts for the display. The 4800u idles ~10w + 2-4 watts for the display. A simple node bump would close the efficiency gap between the two chips, at least on paper.
Unless you’re comparing perf per watt as a curve, idle wattage is largely not representative of real world use. Similarly for peak performance if you’re excluding the wattage or throttling at those levels.
It's impossible to compare either of these CPUs because one is heterogeneous and the other is big.LITTLE. It's still fair to analyze their power consumption as a black-box machine though.
That’s true though from a user perspective, the PpW of the package is more important than the actual core.
Though of course there are other comparison issues like the ANE and video decoders making certain tasks significantly more efficient, so testing the whole package does become much more complex to represent.