Irrelevant numbers because you don't like it. We are comparing benchmarks and you are trying to shift your goal post to other things. YOU started with benchmarks, I replied you with benchmarks. This is a 1:1 comparison.
There is not even a power test based on any of these game benchmarks, so how can you say it throttled. Furmark and 3dmark is irrelevant, we already know that AMD does better in 3dmark. Not to mention that Carrizo did use 8 watts more than the linked Broadwell device. Irrelevant for Skylake anyways.
There s a power test made with HVinfo displaying over 18W for the linked 5200U during the stress test...
Anyway i already warned about the difficulty to extract numbers from chips that do not have a fixed power limit, not counting the power deliveries paths i mentioned, good luck finding reviews whose numbers can be trusted confidently.
For Carrizo at 15W i dont use HP numbers but the ones of an Acer wich has about 10% lower scores in CPU tests, this way i m sure to not inflate the numbers as this imply that this laptop chip use about 20% less power than the HP s, this corelate with the HP chip being boosted at 19-20W when cold benched.
Now for SKL i m waiting fo more reviews, so far about all laptops use a dGPU, wich render the analysis boring and likely innaccurate, that s why i wont even bother trying as it s wasted time.
Photo Caption: "The graphics performance increases if a second RAM module is added."
Obviously they added another RAM module.
They often do so with some laptops that are delivered to them without the second RAM stick, to know better require comparing with other reviews.