30sec seems excessive - I wouldn't see what can take that long 
Unless the neopixel library is a lot slower than I'd think or my loops aren't dropping out as planned.
I think this calls for some debugging 
Update:
I guess (without actually debugging ) the strip.show() does cause the "unexpected" delay since the rest of the code should be in the sub-millisecond region.
So you could try to keep the logic unaltered but only call strip.show() every stepSpeed turns.
I usually use APA102 strips without any library (overhead) which makes things really fast