The benchmarks that we have as of today for methods like IndexOf, StartsWith etc use not very realistic test cases and moreover typically test a single culture.

/cc @jkotas