My standard 'install a specific ripgrep' benchmark, now with hyperfine and zero install with nix

benchmark.md

With Rust and Hyperfine installed natively

(through https://rustup.rs)

hyperfine --warmup=1 'cargo install -f [email protected]'

Without installation on any machine with Nix

nix-shell -p cargo hyperfine --run "hyperfine --warmup=1 'cargo install -f [email protected]'"

Without hyperfine

This won't be quite as comparable, hyperfine is really great for benchmarking, but if you only have cargo, just run the install twice:

cargo install -f [email protected] && cargo install -f [email protected]

Steamdeck - 48.6s

This seems... weirdly slow. I'm sure it used to run in half this time.

Thinkcentre - 39.6s

Benchmark 1: cargo install -f ripgrep@14.1.0
  Time (mean ± σ):     39.647 s ±  0.210 s    [User: 130.243 s, System: 5.717 s]
  Range (min … max):   39.481 s … 39.883 s    3 runs

An old thin workstation I had lying around with an old i5-6400T

X13 Yoga - 14s

Benchmark 1: cargo install -f ripgrep@14.1.0
  Time (mean ± σ):     14.152 s ±  0.349 s    [User: 81.808 s, System: 3.044 s]
  Range (min … max):   13.777 s … 14.868 s    10 runs

Sometimes thermal throttling pushes this to 24s.

2024 Ryzen 9 7950X PC build - 5.7s

💨