I've been using manufacturing for years and years and to be honest, I always just hit normal. I don't understand why all those options are there - maybe it's a legacy thing, from the days where computers were orders of magnitude slower than they are now. You wouldn't want to just regenerate, because it would take forever, so you just step through in reasonable chunks? I don't know.
Maybe next time I'm doing a bit of programming I'll try one of them...
I work on large files and it would be useful to reduce the regen time.
My fear is the regen normal might leave something incomplete?
In an automatic process, I wouldn't like to find an incomplete NC path and that errors go unnoticed.