On the other hand, with a guitar, the more efficient a radiator it is, the louder it is but the less it sustains, because each pluck imparts a fixed amount of energy to the instrument.
Beyond that we operate on a logarithmic scale when it comes to sound, what sounds slightly louder to us can be heard over much larger areas.
Admittedly, guitars are different from violins here, but... with a guitar, there's a pretty fixed amount of energy transmitted when a string is plucked. This energy may radiate quickly or slowly. In other words, there's a tradeoff of attack and sustain. An exaggerated case it the banjo. A banjo is loud, but has little sustain. An unplugged Les Paul electric makes little acoustic sound, but sustains for many seconds. Within the narrower world of acoustic guitars, small-bodied guitars tend to be apparently louder, but not as rich as large-bodied guitars, due to this attack/sustain tradeoff.
80 decibels, ~= 0.001 watts of energy. https://en.wikipedia.org/wiki/Sound_energy_density_level
PS: And yes most home speakers ~1% efficient. https://en.wikipedia.org/wiki/Loudspeaker#Efficiency_vs._sen... At scale they can get into the 20% range, but even that suggests guitars are probably not that efficient overall.
My cello is particularly resonant, and in hindsight (is this confirmation bias talking?) has very little sustain on pizzicato (plucking) compared to my cheap, less-resonant student cello.
I think the F shape has a lot more to do with limiting mechanical stress points for cracking. A round hole with the same area as the f-holes would weaken the top and create changes in how the soundboard vibrates. It's different on a flat-top acoustic guitar, where bracing rather than carving governs the strength of the top, and bracing also limits the valid locations for soundholes.