The Fastest Mutexes by Justine Tunney (discussed on Hacker News) is an excellent work digging into the performance of commonly used mutex implementations, and presenting a new fast mutex in Cosmopolitan Libc. This kind of stuff underpins virtually every piece of software written, but it is complicated and not so widely understood, so it deserves attention!
See also the Rust issue for improving Mutex and other synchronization constructs for a good discussion of the details of various mutex implementations.