Piston efficiency varies in a strange way with rpm. Some, apparently, are more efficient at a higher rpm.
There are a lot of different factors influencing the effect of RPM choice for a given power:
* frictional losses (Deakin provides numbers)
* pumping losses (high RPM => throttle more closed for same power => engine does more useless work)
* crank angle (direction and magnitude of effect depend on other engine parameters)
* prop efficiency (direction and magnitude of effect depend on all sorts of stuff!)