Python code to run Rule 110 sequentially and in parallel.
I wrote this because I was wondering how much of a speedup you get from executing Rule 110 in parallel processes, given that a lot of communication between processes is necessary because of how the rules work. As it turns out, there is still a substandial speedup from using multiple processes even for pretty small inputs (>1000-cell tape with a few thousand iterations).
The file rule_110.py contains two implementations of Rule 110 (one that runs
in a single process, one that runs in multiple processes) and a simple
benchmark test that you can play with.
Coming soon, maybe:
- Re-implementation in a lower-level language (C or Rust) with shared memory
- Additional parallelization algorithm that I have in mind, which is greedier than the one implemented here