I was not playing the precision game - it's even nonsense given the dependency on the particular instruction mix in the application.

And it did not took me a hour. The few figures I needed I have at hand in the oh-so-many-times-mentioned table, just three clicks away, and the rest was a guesswork anyway.

I want the OP be aware that the 100x is a hype which might fall short of expectations should it come to the edge. There are applications where this might matter; I don't say this IS the case (although I can't tell exactly not knowing the details).

And, I also said that the SiLabs mix is the best I know, in other words, other singleclockers are even worse off (and not only clock-frequency-wise).

I think this all are informations worth knowing when making an initial decision on which derivative to use.

JW