See Tokio's Loom as an example: https://github.com/tokio-rs/loom https://github.com/tokio-rs/loom In development, you import Loom's mutex. In production, you i

See Tokio's Loom as an example: https://github.com/tokio-rs/loom https://github.com/tokio-rs/loom In development, you import Loom's mutex. In production, you import a regular mutex. This of course has zero overhead, but the simulation testing itself is usually quite slow. Only one thread can execute at a time, and many iterations are required.