bitemyapp/benchmark-fft

FFT benchmark to compare performance of MoonBit with other languages

FFT Benchmark Project

A multi-language FFT (Fast Fourier Transform) performance benchmark comparing Rust, MoonBit, Swift, and Go implementations.

Features

• Multi-language support: Rust, MoonBit, Swift, and Go implementations
• Automated building: Auto-builds all language programs
• Performance benchmarking: Runs multiple tests with statistical analysis
• Result visualization: Generates performance comparison charts
• Optional verification: MoonBit ↔ Rust result verification
• Verbose logging: Optional detailed command execution logs

Requirements

This project has been tested on the following environments:

• Python 3.13.7
• Rust 1.89.0
• MoonBit v0.6.25
• Swift 6.0
• go 1.22

Quick Start

# Basic benchmark (build + test)
python3 bench_runner.py

# Skip building (if already built)
python3 bench_runner.py --no-build

# Custom parameters
python3 bench_runner.py --runs 20 --inputs 16 18 20 24

# Enable verification
python3 bench_runner.py --verify --verbose

Examples

# Quick test (skip build, 5 runs)
python3 bench_runner.py --no-build --runs 5

# Detailed test (build all, 20 runs, enable verification)
python3 bench_runner.py --runs 20 --verify --verbose

# Custom input sizes
python3 bench_runner.py --inputs 16 18 20 22 24 --runs 15

Output

• Build progress: Shows building status for each language
• Test results: Statistical summary table with min/max/median/average times
• Chart: Generates bench_avg.png performance comparison chart
• Verification: Optional MoonBit ↔ Rust result verification

Default Settings

• Input sizes: 18, 20, 22 (2^18, 2^20, 2^22 complex points)
• Runs per input: 10
• Verification: disabled by default
• Chart output: bench_avg.png
• Verification files: ./.verify_out/

Troubleshooting

• Use --verbose for detailed error information
• Ensure all language environments are properly installed
• Check write permissions for verification directory