How to interpret profile information

[alessiostalla]

Hi, is there a guide or tutorial somewhere on how to interpret profile information? For example, in my parser and problematic inputs I noticed that, for rules with the most time spent in prediction, the max SLL lookahead is in the thousands, while the max LL lookahead is smaller by a factor of 10 to 100. I have no idea if that's normal or a problem, and in general, it's not clear what parameters to look at to identify where all that time is spent.