As far as I'm aware, as long as there's no output dither (at least until the 3rd/final pass) the results will essentially be the same.
You could quibble about needing to work at 32bit fixed if you're going to do multiple passes through SFP, or 32bit flt. for using compressor plugins in a sequencer or audio editor, but I recommend just getting on with your music.
Durring the second compression stage, some/all of the signal you compressed first will be compressed again, and some/non of the signal, that was below the first threshold will be compressed now.
Doing many ligth (and similarly dialed) compressions in a row makes it possible to achieve a compression, that goes gradualy from ligth at low levels to very close to infinite at higher levels.
Soft knee does not necesarily go all the way up to limiting. Soft knee just uses 2 (or sometimes more?) compression stages , wich kick in closely to easch other.
I was saying that there shouldn't be any difference between applying compression in 3 passes to a file in a wav editor, vs. chaining 3 compressors and routing the audio through them.