This hasn't really been true in laptops. A major reason for this is that in laptop, the overall power management strategy and the design of the laptop itself matters a lot in total power efficiency, so efficiency benchmarks are often counterintuitive to what you'd think they'd be based on which chip was fabricated on which process node. The R9 8945 HS is, on a flop-per-watt level, under identical loads and identical clock speeds, more efficient than any Intel CPU on the chart (5% better than the Core Ultra 7, according to a chart in the article). But of course, things aren't identical, and in quite a few benchmarks, the specific ASUS ROG laptop it's in ends up gobbling a lot of power.
Example, Intel's new Core Ultras consume significantly less power when idling than AMD's do, despite using a less efficient compute-per-watt chiplet design. That probably won't affect these specific benchmarks, but it's a good example of laptop power management that you don't really care about in desktops or servers.
Source:
https://www.notebookcheck.net/AMD-R...r-crunching-and-GPU-performance.802637.0.html