What do you mean by "seemed to really slow things down"? At the most it should be about 7us difference - there should be no noticable effect.

I suspect it may be a combination of shorter initial delay plus playing with the bit timings. I've no idea how much the frequency varies on a ceramic oscillator with temperature, but it's not going to be more than maybe 5-10% (which shouldn't make a difference, really).

Hugo