There are a number of problems with the methodology on this that make any comparison invalid.
I get the whole benchmark it until it breaks mantra and in that respect they achieved their goal however, the variable settings for power, thermal, and clock manipulation doesn't lend itself to an objective conclusion. To say the Max Turbo Clock rate was not maintained is throttling is a misrepresentation. Unless you can pin the configuration to the same settings you cannot make a comparison from one to another. In a multidimensional system you cannot pick one parameter and draw a conclusion based on that alone.
Turbo Boost is variable and not continuous, as evidenced by the XTU setting for microseconds of Turbo Boost. At the same time clock rates are variable, voltage and current are variable, boost duration is variable, thermal parameters are variable. Each of these contributes or detracts from the total work accomplished over time which is the ultimate desired result. The instantaneous parameter values may not be a 100% accurate predictor under dynamically variable conditions.
If as shown on a youtube video, the sustained performance of a system is maintained at 16W and the rated power of that system is 15W how do you conclude it throttled at all even though under temporary boost conditions it may have peaked at 30W? The message being sent is misleading and incorrect.