Cross-Language Benchmark Suite

A comprehensive benchmark suite comparing performance across multiple programming languages: C, C++, Rust, Python, and Java with an interactive TUI for visualizing results.

Features

• Multi-language support: C (clang), C++ (clang), Rust, Python, Java
• 12 benchmark categories: Matrix, Sort, String, Hash, Regex, JSON, File I/O, Math, Network, Crypto, ML, Concurrency
• Interactive TUI: Graph and table views with statistical analysis
• SQLite storage: Persistent benchmark result history
• Row selection: Navigate and drill down into specific benchmarks
• Statistical view: Standard deviation, quartiles, min/max

Quick Start

# Install dependencies
mise install

# Compile all benchmarks (do this once before running)
mise compile

# Run all benchmarks
mise all

# View results in TUI
mise monitor

Requirements

• mise - Language version manager and task runner (brew install mise)
• C/C++ - GCC or Clang with optimizations
• Python - 3.x with numpy
• Rust - Latest stable
• Java - OpenJDK 21+

Benchmark Commands

# Compile all (before running)
mise compile

# Run all benchmarks (compile + execute)
mise all

# Run only (requires compiled binaries)
mise run-all-benchmarks

# By language
mise c-run          # Compile + run all C benchmarks
mise cpp-run        # Compile + run all C++ benchmarks
mise rust-run       # Compile + run all Rust benchmarks
mise python-all     # Run all Python benchmarks
mise java-run       # Compile + run all Java benchmarks

# Compile by language
mise c-compile
mise cpp-compile
mise rust-compile
mise java-compile

# Individual benchmarks (requires compilation first)
mise c-matrix
mise rust-sort
mise python-hash

By Category

mise matrix-all      # Run all matrix benchmarks
mise sort-all        # Run all sort benchmarks
mise hash-all        # Run all hash benchmarks
mise regex-all       # Run all regex benchmarks
mise json-all        # Run all json benchmarks
mise fileio-all      # Run all file I/O benchmarks
mise math-all        # Run all math benchmarks
mise network-all     # Run all network benchmarks
mise crypto-all      # Run all crypto benchmarks
mise ml-all          # Run all ML benchmarks

Tools

mise monitor         # Launch interactive TUI
mise store           # List stored benchmark runs
mise runner          # Run benchmarks with automatic result storage

Monitor TUI

Run mise monitor to launch the interactive TUI:

Key	Action
[1]	Graph view (line chart)
[2]	Table view (list)
[3]	Statistics view
[j/k] or arrows	Navigate rows in table
[Enter]	Detail view for selected benchmark
[s]	Select benchmarks to run
[r]	Run selected benchmarks
[R]	Refresh data from database
[q]	Quit

Detail & Stats Views

• Detail View: Shows all languages compared for a specific benchmark with min/max
• Statistics View: Shows avg, min, max, std_dev (standard deviation), Q1 (25th percentile), Q3 (75th percentile), runs

Project Structure

Benchmarks/
├── C/              # C benchmarks (.c files)
├── C++/            # C++ benchmarks (.cpp files)
├── Python/         # Python benchmarks (.py files)
├── Java/           # Java benchmarks (.java files)
├── Rust/           # Rust benchmarks (.rs files) + Cargo.toml
├── tools/          # Monitor & database tools
│   ├── monitor.rs  # TUI application
│   ├── runner.rs   # Benchmark runner
│   ├── lib.rs      # Database & types
│   └── Cargo.toml
├── mise.toml       # Tool versions & task definitions
└── README.md

Optimization Flags

• C/C++: -O3 -ffast-math -march=native (maximum compiler optimizations)
• Rust: opt-level = 3, lto = "fat", codegen-units = 1
• Java: JIT warmup enabled before timing

Benchmark Categories

Category	Tests
Matrix	Multiply, Transpose, Add (2000x2000 float matrices)
Sort	Qsort, Quicksort, Heapsort, Mergesort (5M integers)
String	Concatenation, Split, Replace, Search
Hash	SDBM, DJB2, FNV (10K iterations)
Regex	Find, Count, Email pattern match
JSON	Parse, Serialize, Count fields (1MB payload)
File I/O	Write, Read, Read lines, Random access (5MB files)
Math	Trig (sin/cos), Exp/Log, Arithmetic
Network	Pipe throughput, Memcpy 8K
Crypto	XOR, Integer ops
ML	Element-wise mul, Dot product, Lerp, Sigmoid
Concurrency	Thread pool, Atomic counter

Adding New Benchmarks

1. Add source file to each language directory
2. Add tasks to mise.toml with *-compile and * (run) variants
3. Rebuild tools if needed: mise tools-build

Sharing Results

Results are stored in SQLite. To share results between machines:

# Export results
sqlite3 ~/.local/share/benchmarks/benchmarks.db .dump > results.sql

# Import on another machine
sqlite3 ~/.local/share/benchmarks/benchmarks.db < results.sql

Or copy the database file:

~/.local/share/benchmarks/benchmarks.db

License

MIT