Robert correctly pointed out that there is a clear problem with correctly estimating memory latency. 8.1ns at lpddr5 8500? This is impossible for any modern architecture, otherwise x86 would not have such a failure by more than an order of magnitude, especially Meteor Lake with the fastest x86 in lpddr5 7500 laptops.
Otherwise, this is an ordinary, deliberately failed device for operation - because. There is a strong strain on the eyes with AMOLED, the noise is increased, despite the fairly modest PL2/PL1.
Apparently, such "laptops" are intended to simply sit on the desks of middle managers, like beautiful furniture, although hardly anyone will buy it in this color.
Regarding the topic of SoC performance:
Compare the shame of Intel - laptops with Meteor Lake with PL1=45W+ are shamefully compared with AMD 7840 with PL1=18W, i.e. laptop with AMD is 2.5 times more economical in PL1 mode! With a performance difference of only 60% in favor of Meteor Lake. This is a complete failure for Intel, even taking into account the fact that part of the SoC is made by TSMC.
What is needed here is not just unification in terms of performance per 1W in multi-threaded mode (especially since there are huge differences in different tests - for example, AIDA Ray-Trace 64 is always in favor of AMD), but unification in terms of performance per 1W taking into account the noise of the laptop, each model, and taking into account the weight of the laptop, not even a laptop - but the pure weight of the cooling system installed on the processor SoC.
Here's what you need - a general report card taking into account 4 criteria - performance in multi-threaded tests (average taking into account certain weights) normalized taking into account consumption and this normalized taking into account the weight of the radiator/cooler and noise in PL1 mode long-term. Here's what those choosing a laptop need.
It is this table that will definitely reveal the real favorites.