what are you all using to bench jvm operations? One of our data people would like to do some jmh profiling