Maybe you will find the (crude but very practical) oscilosscope-and-testpin method useful.

Designate a I/O pin as a temporary test pin and hook your scope up to it. Now decide which part of the program you want to monitor. Put a "clr testpin" instruction at the beginning and a "setb tstpin" at the end. The oscilloscope trace will now show you in real time approximately how much time is used.