Ok, lets start from NDT3055L
Calculate for worst case 2 amp in parallel with steady on state. We have 12A load.
Rdson = 0.1OHm, so Voltage Drop on transistor is 12*0.1 = 1.2V, This is 6% of 20V, this is more than 5% of amplifier tolerance (as far as I remember) - transistor can't be used in this configuration
Let's try to calculate for 6A load.
Voltage drop is 6*0.1=0.6V. This is 3%, so your power supply should provide 20V +5%/-2% (quite tight but possible).
Power dissipation on transistor: P = I*V = 6*0.6 = 3.6W. This is quite a lot for smd chip without heatsink.
Temperature raise: P*Tja = 3.6*42 = 151C. Die temperature is 151 + 25 (ambient) = 176C. This is more than 150C maximum. So, transistor can't be used.
Now, let's try to evaluate for pulsed mode (2s cycle 50% duty cycle). By Figure 11 give coefficient 0.5 to Tj.
So, Tja will be 21, Die temperature is 3.6*21+25 = 100.6. In limits, but still too hot. Each 10 degrees of temperature raise reduce life time of transistor by 2 times.
And relaying for pulsed mode also dangerous - if some amplifier will be turned on for more than some time (10-100s approximately) transistor will be burned out.
Now check SiSS65DN.
Rdson = 4.6mOhm. Voltage drop (12A load) is 12*4.6e-3 = 55mV. Almost negligible from point of view of power supply for Amplifier.
P = 12*55e-3 = 0.66W. Max T raise 0.66*25 = 16.5C. Quite cold.
PS. 25 is worst case for Tja (from datasheet). But it specified for mounting on PCB with 1" copper area and for pulse less than 10s. DS do not contain information for steady state. But there is a large reserve in power and temperature, so steady state should not be a problem.