Although it is entirely possible that a 55nm shrink could use more power than the 65nm 'original' chip (depending on inner workings of the process itself, I assume), the use of an engineering sample doesn't tell us much, does it? Surely such an engineering sample could be running at different voltage to the final chips also?