If you use cluster.fork() to spawn new processes instead of using the main process you will be able to solve the same problem without the use of threads. Threads needs a lot of considerations and I think most problems can be better handled with an actor model and cluster instead of introducing threads.

The following code runs on 1.156s on my Macbook Pro:
https://gist.github.com/irony/d44cbda0c71707e89274edd63fcc3d35