I ran a small performance test that feeds in 1,000,000 lines of the non-matching text to three regexes to show the difference a few characters make.