Those 100W+ numbers are total system power. And that system has the CPU TDP set to 80W (far above AMD's official max of 54W). It also has a discrete 4070 GPU that can use over 100W on its own.
if x86 laptops have 90w of platform power, that’s a thing that’s concerning in itself, not a reasonable defense.
Remember, apple laptops have screens too, etc, and that shows up in the average system power measurements the same way. What's the difference in an x86 laptop?
I really doubt it's actually platform power, the problem is that x86 is boosting up to 35W average/60W peak per thread. 120W package power isn't unexpected, if you're boosting 3-4 cores to maximum!
And that's the problem. x86 is far far worse at race-to-sleep. It's not just "macos has better scheduling"... you can see from the 1T power measurements that x86 is simply drawing 2-3x the power while it's racing-to-sleep, for performance that's roughly equivalent to ARM.
Whatever the cause, whether it's just bad design from AMD and Intel, or legacy x86 cruft (I don't get how this applies to actual computational load though, as opposed to situations like idle power), or what... there is no getting around the fact that M2 tops out at 10W per core and a 8840HS or HX370 or Intel Meteor Lake are boosting to 30-35W at 1T loads.
> In our peak power test, the Ryzen AI 9 HX 370 ramped up and peaked at 33 W.
https://www.anandtech.com/show/21485/the-amd-ryzen-ai-hx-370...