A faster implementation of finding sets of 5 words of length 5 that all contain different letters from a provided list of words. See https://www.youtube.com/watch?v=_-AfhLQfb6w for more info. The algorithms presented in the video appeared optimizable.
Currently this takes about ~1.3 seconds to generate the "correct" 831 words output according to the linked video on my i9-12900K @ 3.2GHz, which is admittedly a little overkill.
This primarily achieves the performance it has by:
- Being written in Rust, a language that compiles to machine code
- Using multiple threads
- After choosing a word, only considering words that are lexicographically after that one, trusting another iteration would capture when that appears before
- Compare word validity using bitstrings and bit operations
There are likely more improvements to make. But another few orders of magnitude improvement over the existing solutions is sufficient for now. :)
- Prune fully-checked words from initial filtering - Not successful
Install Rust https://rustup.rs/.
Clone this repository.
Run:
cargo run --releaseIf you'd like to provide an alternate word list, you can specify the input file as an argument to the program:
cargo run --release -- your_word_file.txtThe program expects that file to be a newline separated list of words. It will only consider 5 letter words.