Been there, done that, ended up poorly
There was some formula floating around the net, showing how to calculate joules/second needed to spin a disc of given weight from zero to 6000rpm. It fell somewhere around 616 J/lbs. Then it could be converted from J/s to hp.
Example: by changing a standard-driven pulley from 2.11 lbs steel to 0.66lbs aluminum, gain is only a miserable 0.84 hp at 6000 rpm.
Fluidamprs are about 2 lbs heavier than stock rubber-steel pulley, yet they recover a couple of hp as proven on the dyno, due to dampening effect. It takes energy to vibrate a 30 lbs forged steel crank, and it has nowhere to come from, except from engine itself.
Tried a lightweight, non-clutched pulley for the alternator. Stock pulley is one-way clutched and almost 2 lbs heavy. The purpose was to compensate for the weight of the Fluidampr, which is about 2 lbs above a stock harmonic crank pulley.
There was a slight (almost too slight to feel) increase in engine responsiveness and nothing more. But after 1 year and half, accessories on the serpentine belt began to make grinding noises at idle rpm and alternator voltage went down. There was no noise or vibration above 1500 rpm.
Removed the serpentine belt and checked: alternator slight grinding (bearings worn away), belt tensioner also slight grinding. Replaced both. Also went back to OEM clutched pulley.
In racing, where the engine runs for short time and mostly above 6000 rpm, the slightly better engine responsiveness may help.
In city driving, it will, sooner or later, wear down accessories (alternator, A/C compressor, power steering pump) because solid metal pulley does not allow any compensation for any fault - the smallest wobble in a bearing is amplified by belt tension until it gnaws away at the bearing or the tensioner. Or both.