People have used the Unitrode PWMs as switching regulators.
Not a great idea for efficiency, the high side voltage drop is high
and the low side, not great. 800mA or so can be had if you don't
mind all that. I am not familiar with the CMOS / BiCMOS versions
output stage but the way it's spec'd makes me think MOS outputs.
With 25 ohms at 200mA in the limit table that's 5V drop on the high
side (peak current does not tell the "on" voltage drop story where
you care about it, if you're using the output as the SW node
directly). At 800mA on this CMOS version I think you'd have
almost no output from a 20V rail.
For driving FETs, all you need to know is in those rise time specs.
50ns at 1nF is pretty leisurely and you can pretty much scale
tr, tf by your_FET_Cgate/1n and see what risetime will look like.
If you want 300ns pulse width, you probably don't want a 3nF
gate capacitance malforming it with 150ns rise and fall. And
forget driving a 10nF beast.