That sort of comparison isn't so interesting, significant or impressive to me as an engineer. Likely the difference in each test would be reduced if a number of runs are performed. Too much opportunity for experimental variation / error. Doesn't matter that they used a high-speed high-frame rate camera. Reporting more digits provides precision, but that is not a substitute for accuracy or validity in the measurement or comparison.
How many of us experience an application loading fast on one day, and less fast on another? Likely with the OS designed the way it is, any of these individual experiments (or measurements) will have a mean and a standard deviation associated with them (i.e. they will have a "distribution").
I'm not saying the comparison is completely without value...
But, I'm thinking w.r.t. operations we are doing on a typical day, the differences in responsiveness for micro operations like these are not likely to be perceived as significant.
I assert, in the end, on a fast device, the differences aren't going to be as significant as they will be on a slower / less powerful one.