The frontpage is marketing, so it uses the most optimistic values, with no discussion whether they are applicable to real-world usage or not. This one may come from Table 29, performing while(1) at 4MHz at Vcore=1.2V. It's typ, so it's at VDD=VDDA=3.0V, "under ambient temperature" (presumably meaning room temperature, i.e. 25deg.C). Note, that would they use the same while(1) value from Tab.31 i.e. running from RAM, they would achieve an even slightly better value.
The pressure to get "nice numbers" onto the fronpage is huge, and the penalty for the "minor marketing lie" is only the outrage of the engineer which is usually constrained to his cubicle, so there's even an evolution in these:
(Strangely, the lowest consumption at 32MHz does not come from while(1) but from Fibonacci!)
JW
Let's just have more fun with the "evolution" of frontpage matter... ;-)
Oh, so they added 2 SPI interfaces meantime? That's nice! Except that....
"SPI mode of USART"... wait, what? Oh, you mean the *synchronous* mode of USART, that's where the S in USART comes from!
So this is a clear indication how marketing tries to wrap reality as far as it goes. It's nice of them they put up the "Up to" as a warning that lie may follow.
To make it clear, this is not all ST's fault, they are under pressure of the competition which does exactly the same. It's a big shame, but again, engineers are in these cases more and more often delegated to the role of the grumbling droid who has to hammer in the nails and can be neglected otherwise.
JW
PS. Not all changes are wrong there - the minimum working voltage of the comparators got corrected, typo corrected in "8x peripheral communication interfaces" and one remains in "Step-up converted on board" ;-)
