Advent of Code 2022

These are my solutions to Advent of Code 2022, written in Rust. The tag v1.0.0 marks the state of the code when I completed all of the puzzles. Subsequent commits are continuing work to improve the code's style and performance and to fix bugs.

Each day's puzzle is implemented as a package with a binary and a library crate. All examples provided as part of the puzzle descriptions are used as unit tests, as are the actual puzzle inputs and answers.

Run the tests for a particular day:

$ cargo test -p day01

Run the tests for all the days:

$ cargo test

The example and actual puzzle inputs are in files in the data subdirectory of each day's package. The main programs take no arguments and read their input from stdin.

Run the program to solve both parts for a particular day:

$ cargo run --release -p day01 < day01/data/input

Benchmarks

Initially I just tried to solve the problems without worrying about how fast they ran. But after completing all 25 problems, I worked on optimizing the performance of my slowest solutions. Here are some current benchmarks.

On an AMD Ryzen 7 3700X, the solutions to all 25 problems run sequentially in a cumulative time of 0.21 seconds:

% hyperfine --warmup 10 'for day in day*; do target/release/$day < $day/data/input; done'
Benchmark 1: for day in day*; do target/release/$day < $day/data/input; done
  Time (mean ± σ):     209.6 ms ±   6.3 ms    [User: 220.0 ms, System: 25.8 ms]
  Range (min … max):   200.1 ms … 221.0 ms    14 runs

Every solution completes in less than 100 ms, and all but the following complete in under 10 ms:

Day	Mean [ms]	Min [ms]	Max [ms]
day23	92.5 ± 2.1	89.2	98.9
day24	30.4 ± 0.8	29.7	34.4
day16	20.6 ± 0.6	20.1	23.4
day20	12.1 ± 0.4	11.6	13.9

On a Raspberry Pi 4 Model B aarch64, the solutions to all 25 problems run consecutively in under 1 second:

% hyperfine --warmup 10 'for day in day*; do target/release/$day < $day/data/input; done'
Benchmark 1: for day in day*; do target/release/$day < $day/data/input; done
  Time (mean ± σ):     943.2 ms ±   4.6 ms    [User: 883.3 ms, System: 124.1 ms]
  Range (min … max):   938.7 ms … 951.3 ms    10 runs

These benchmarks were performed using hyperfine. The code was compiled with -C target-cpu=native.