Trying to judge SoC power/efficiency from system level measurements with no controls or even way to measure the system level power is a fool's errand. It shouldn't be a surprise that you can get very different results with different test systems (or OS) even when using the exact same SOC.
I agree but think you are missing the plot: this is exactly the point I am implicitly & explicitly making, because that is part of what Geekerwan is intending to do for his audience and everyone knows this, the system level (laptop itself) part is actually a peripheral matter to him.
It’s not clever to point this out in response when the status quo reaction right now is “Lunar Lake has lower idle Qualcomm sux Intel caught up to Apple what a killer job” etc which is truly the dullest take possible as I explained.
Which brings me to my next point I already iterated: we have the king of controls anyway with ecological validity too for discerning about as much as we could via the XPS 9345 for Lunar Lake and Qualcomm’s X Elite, with both obviously coordinating with Dell to reduce every bit of system drain and parasitic taken for granted — read the leaked Dell Docs on that.
And at any rate they’re remarkably similar, Intel wins by an hour or two on a 27-29 hour offline playback test and QC seems to either tie or pull ahead in web browsing.
I don’t fault Geekerwan for any of this as system implementation or OS and drivers is functionally a relevant variable but we have good reason to believe the Galaxy Book on Linux is harming the results and not mirroring what we see in other laptops with the X Elite.