vector-rs Benchmarks and Setup

Installation

1. Rust (native)

• Install Rust: https://rustup.rs/
• Clone this repo
• Build and run native Rust benchmark:

  cargo run --release --bin bench

2. Python (PyO3/Rust extension)

• Install Python 3.10+
• Create and activate a virtual environment:

  python -m venv .venv
  .venv\Scripts\activate  # Windows
  source .venv/bin/activate  # Linux/macOS

• Install dependencies:

  pip install maturin uv
  uv pip install py-ballisticcalc[exts]==2.2.0rc2
  maturin develop

• Run Rust/PyO3 benchmark:

  python benchmark_vector.py

3. Python (Cython/C extension)

• Ensure py-ballisticcalc[exts] is installed in your venv
• Build Cython benchmark:

  pip install cython setuptools
  python setup_bench_cython_v3d.py build_ext --inplace

• Run Cython/C benchmark:

  python -c "import bench_cython_v3d; bench_cython_v3d.bench_v3d()"

Benchmark Results (1,000,000 calls)

Operation	Rust Native	PyO3 (Rust in Python)	Cython/C (v3d)	Rust (glam)
add	0.0010 s	0.1161 s	0.0010 s	0.0010 s
sub	0.0010 s	0.1149 s	0.0010 s	0.0010 s
neg	0.0010 s	0.0852 s	0.0010 s	0.0011 s
mul_scalar	0.0010 s	0.1196 s	0.0010 s	0.0011 s
dot	0.0010 s	0.0890 s	0.0010 s	0.0010 s
mag	0.0013 s	0.0610 s	0.0011 s	0.0010 s
norm	0.0012 s	0.0923 s	0.0011 s	0.0010 s
eq_false	0.0000 s	0.0654 s	0.0011 s	0.0000 s
eq_true	0.0010 s	0.0653 s	0.0011 s	0.0000 s

Additional: Rust (glam) SIMD Benchmark

• Uses the glam crate for SIMD-optimized vector math.
• To run:

  cargo run --release --bin bench1

• Results show glam is as fast as hand-written Rust and Cython/C for these operations.

Reproducing Benchmarks

1. Follow installation steps for each method above.
2. Run the provided benchmark scripts/commands.
3. Compare the output to the table above.

Notes

• Native Rust and Cython/C are extremely fast for vector math.
• PyO3/Rust in Python is much slower due to Python/Rust FFI overhead.
• For maximum performance in Python, prefer Cython/C or batch operations in Rust.