@jzakiya I have some new code. Can you try it?
I checked your Rust code. It doesn’t use channels like Crystal. It does a parallel map to compute things, then uses that to compute the final results. Of course that’s faster than using channels because there’s no need to communicate anything between fibers.
So, one has to be careful about comparing benchmarks that use different algorithms.
Also, I replaced Array with Slice because I think that’s more similar to Rust’s vec, plus they never grow so it’s okay to use Slice.
There’s no parallel map so I’m using an implementation courtesy of @waj
In my machine it ends up running a bit faster and consuming just 20MB of memory.
Can you try it on your machine and compare the time/memory to Rust and Nim? I’m not sure this will beat them, but maybe the times will be similar.