Earlier quoted context omitted.
>Toe to toe? It starches every Ryzen chip in single core performance It most certainly does not. It trades blows, winning on some benchmarks and losing others. It seems a bit ahead on floating-point workloads and a bit behind on integer workloads. https://www.anandtech.com/show/16252/mac-mini-apple-m1-teste... It does simply outperform every Intel design. That's great, but not enough to make it the fastest core in th…
M1 is either first or very close to first in all the Anandtech benchmarks against CPUs using up to seven times more power.
>using up to seven times more power.
Power efficiency is where the M1 definitely wins. It is among the best on the planet for single-threaded performance and definitely the best for efficiency.
Your 7x figure, however, is not accurate. You are comparing TDP. That is the wrong number. TDP is calculated differently by each manufacturer. It relates to how big a CPU cooler you will need to ensure the CPU never throttles to avoid overheating, not to how long your battery will last.
Take a look at the following, in particular the big black-and-green-and-red charts in the middle of the page:
https://www.anandtech.com/show/16214/amd-zen-3-ryzen-deep-di...
(Those are desktop parts. I'd obviously like to compare mobile ones, but I am not familiar with AMD's mobile lineup and there are no Zen 3 mobile parts. They'll still help make my point.)
Note that the power draw at idle (where the CPU will spend most of its time) is 10-20 watts for all these CPUs. This would be lower for mobile Ryzen parts, of course. This is the big advantage of the M1: with its big.LITTLE architecture, it will be able to shut down the big cores entirely and use vastly less power than any x86 part at idle.
Note that the power draw at 100% usage is higher than the stated TDP. The CPU can draw more power than it says on the box if the system detects it has adequate thermal headroom. It will also draw less power, even at 100% usage, if it starts getting too hot or if the user chooses to lower the TDP.
Note that the power consumption varies with the number of cores that are active, but that this relationship is nothing close to linear. A CPU can hit its peak power at a small fraction of maximum load, but performance will continue to scale all the way to maximum load. The big chips are more efficient than the small ones when both are fully loaded.
You can't just say "a chip with N times the TDP draws N times more power". The two parts will draw different amounts of power and have different efficiency tradeoffs depending on how they are used.
What, then, is the correct metric? I don't know of a single number that tells the whole story. I can tell you that the M1 is efficient both at idle and under load. I will make no effort to quantify how efficient it is other than to say TDP is irrelevant.