Expanding on this. If you have a lot of concurrent tasks, you will overflow[0] the task local queue and be bottlenecked by the global queue mutex[1] [0]: https

Expanding on this. If you have a lot of concurrent tasks, you will overflow[0] the task local queue and be bottlenecked by the global queue mutex[1] [0]: https://github.com/tokio-rs/tokio/blob/8897885425bf3d89053f896319eeb8777cf255fc/tokio/src/runtime/scheduler/multi_thread/queue.rs#L63 https://github.com/tokio-rs/tokio/blob/8897885425bf3d89053f8... [1]: https://github.com/tokio-rs/tokio/blob/8897885425bf3d89053f896319eeb8777cf255fc/tokio/src/runtime/scheduler/inject/rt_multi_thread.rs#L77 https://github.com/tokio-rs/tokio/blob/8897885425bf3d89053f8...