you need a signal generator, odcilliscope, indutive dummy load at specific impedance and power dissapation to handle the load applied, microsoft excell (or a scientific calculator and a whole lotta paper & pen).
set your signal generator to the frequency you want to test, hook up dummy load to output of amp, and the o-scope is hooked up parallel to the load.
Power in watts is = to Voltage X Voltage / resistance
so if the voltage reads 10VPeak and resistance / inductance is 4 ohms it would be (10*10)/4 or 25 watts peak
you adjust the gain till distortion is just starting to appear in the o-scope display, Peak voltage would give you max watts, to figure RMS (Root Mean Square) you multiply the peak by 0.707, to get AVG output multiply the peak by 0.637, you can also use this to figure out how linear your push-pull stages are by figuring the negative side of the sine wave. THD figures are very hard to calculate so I wont go into that.
so if its 50v peak on top and 46v peak on the bottom on a given frequency this would be the outcome: MAX = 625 Watts, RMS = 312 Watts, AVG = 253 Watts and Pulse Linearity = 85%, that would mean that your amps capability to creat negative cone movement would be 85% of its forward ability. Frequency linearity wold be compiled by repeating these steps across the audio spectrum and averaging that #.
Unfortunately the equipment to do a test like this would cost in the neighborhood of 5 to 7 thousand dollars, and it wold take about two working days to fully test one amp. So probably none of us will be doing that any time soon.
Sad to say many of the manufacturers abuse thier ratings, If all had to be measured at 0.01% THD (total Harmonic Distortion) we could more easily test apples to apples. also, some amps such as the older alpine V12's had a regulated power supply so it produced max output at 12Vdc. whereas the Phoenix Gold amps were unregulated so if you had an alternator with an adjustabe regulator you could tweat it to 15.5 ot 16VDC and squeezee more power out of the amp. Thats how the M100 was originally rated as a 20 watt amp, 10Wx2 RMS @ 4ohms @ 0.005%THD @ 12VDC, but the same amp bridged mono into 3.2 ohms and 16VDC would produce around 700 Watts RMS at 0.10% THD. The other manufacturers (rockford, alpine, and others) got pissed off and the next year IASCA changed the rulings and that same amp had to be rated 50x2. (my system went from (5-M100's) being 100Watts Class IASCA to 250 watts). Soon other manufacturers started playing the impedance game to create "cheater" amps (Can I say Orion?) and now no one really rates thier amps the same.
Well that got log and I could keep going but, all I can say is: If it sounds good and you enjoy it, who really cares about the numbers anyways?