Repo: https://github.com/shulaoda/rust-benchmark-test
Benchmark: https://codspeed.io/shulaoda/rust-benchmark-test

On lower-end processors, the performance of phf_set is significantly worse than array. On my CPU (13th Gen Intel(R) Core(TM) i5-13600KF), the worst-case performance difference between the two is minimal, but in the best-case scenario, phf_set performs significantly worse than array.

PS: The data size is 25

------2025.04.20------

I’ve updated the repo — it now automatically generates the most frequent case and theoretical worst case for more precise and reliable optimization. Feel free to clone it, tweak the basic dataset, and run cargo bench --bench basic to try it out!