We recently started using perfect hashing in V8's parser, also with measurable performance improvement (even over what we had previously, a first character jump table followed by linear search). It's always nice to apply that computer science education to a real world problem.
