This is a benchmark between multiple programs doing the task specified here (without file and argument parsing).

Executing run.sh builds the programs (with release profile) and then benchmarks them with Rust's integrated bencher.

As Rust's integrated bencher is unstable, a nightly toolchain is required. Recommended installation method is obtaining rustup.rs (a Rust toolchain manager) and running rustup override add nightly in this directory.

More versions are welcome! Just send a pull request. I can take care of adding it to the benchmark.

Notes:

• The awk and Python versions are currently doing not the exact same work as the other versions.

Credits:

• Rust version by Bobo1239
• D version by Stefan
• awk version by Fabian
• Python version by Simon (modified by lindt)

Results on my workstation (AMD Ryzen 9 3950X):

test awk   ... bench:  34,227,744 ns/iter (+/- 2,163,885)
test d_dmd ... bench:  18,091,353 ns/iter (+/- 1,489,948)
test d_gdc ... bench:  10,590,457 ns/iter (+/- 1,629,433)
test d_ldc ... bench:  10,381,016 ns/iter (+/- 1,387,163)
test py    ... bench:  56,565,560 ns/iter (+/- 6,188,942)
test rust  ... bench:   9,189,600 ns/iter (+/- 1,224,436)

Bonus Round

With test.large which is 389 MB and not included in this repository. Only tested with a simple time cat test.large | <prog> (warmed file cache).

• awk: 2.747s
• Rust: 0.665s
• D (dmd): 1.632s
• D (ldc): 0.670s
• D (gdc): 0.806s
• Python: 4.086s