- * The contended case for MicroLock is likely to be worse compared to
- * std::mutex than the contended case is. Make sure to benchmark your
- * particular workload.
+ * While the uncontended case for MicroLock is competitive with the
+ * glibc 2.2.0 implementation of std::mutex, std::mutex is likely to be
+ * faster in the contended case, because we need to wake up all waiters
+ * when we release.
+ *
+ * Make sure to benchmark your particular workload.