perl style regular expression are pretty fast...
if you ever tried to cobble togther complex regular expressions in say microdoft word, you will find their expression parser to be very very very inefficient
would you mind posting your benchmarking code for anyone else following this thread so they can try on their machines?