[README.md]

Primes-Compare

primes_compare

comparing numbercrunching in c++; py; rust

Prime Number Generators Benchmark This project compares the performance of prime number generators implemented in Python, C++, and Rust. It includes various optimization strategies and a comprehensive benchmarking system. Project Structure Copyprimes_benchmark/ ├── src/ │ ├── python/ │ │ ├── p-primes.py # Basic Python implementation │ │ ├── p-primes-fast.py # NumPy optimized version │ │ ├── p-primes-faster.py # Numba optimized version │ │ └── p-primes-smart.py # Adaptive implementation │ ├── cpp/ │ │ ├── c-primes.cpp # Basic C++ implementation │ │ └── c-primes-fast.cpp # Optimized C++ version │ └── rust/ │ ├── r-primes.rs # Basic Rust implementation │ └── r-primes-fast.rs # Optimized Rust version ├── benchmark/ │ └── cs_exe_benchmarker.cpp # C++ benchmarking tool └── README.md Implementations Python Versions

Basic (p-primes.py): Standard Sieve of Eratosthenes implementation Fast (p-primes-fast.py): Uses NumPy for vectorized operations Faster (p-primes-faster.py): Uses Numba for JIT compilation Smart (p-primes-smart.py): Adaptive implementation that chooses:

Basic sieve for n < 1,000 NumPy for 1,000 ≤ n < 100,000 Numba for n ≥ 100,000

C++ Versions

Basic (c-primes.cpp): Standard implementation with basic optimizations Fast (c-primes-fast.cpp): Optimized version with improved memory management

Rust Versions

Basic (r-primes.rs): Standard Rust implementation Fast (r-primes-fast.rs): Optimized version with improved algorithms

Benchmarking Tool The project includes a C++ benchmarking tool that:

Runs multiple iterations of each implementation Calculates average, min, max, and standard deviation Supports command-line arguments for different ranges Handles timeouts and error conditions Provides detailed performance metrics

Building and Running Python Implementations bashCopy# Install requirements pip install numpy numba

Build executables

pyinstaller --onefile src/python/p-primes.py pyinstaller --onefile src/python/p-primes-fast.py pyinstaller --onefile src/python/p-primes-faster.py pyinstaller --onefile src/python/p-primes-smart.py C++ Implementations bashCopy# Build basic version g++ -o c-primes.exe src/cpp/c-primes.cpp -std=c++17

Build fast version

g++ -o c-primes-fast.exe src/cpp/c-primes-fast.cpp -std=c++17 Rust Implementations bashCopy# Build basic version rustc src/rust/r-primes.rs -o r-primes.exe

Build fast version

rustc src/rust/r-primes-fast.rs -o r-primes-fast.exe Building the Benchmarker bashCopyg++ -o cs_exe_bm.exe benchmark/cs_exe_benchmarker.cpp -std=c++17 Running Benchmarks The benchmarking tool can be run with: bashCopy./cs_exe_bm.exe For the smart Python implementation, you can specify the range: bashCopy./p-primes-smart.exe 500 # Find primes up to 500 ./p-primes-smart.exe 50000 # Find primes up to 50,000 ./p-primes-smart.exe 500000 # Find primes up to 500,000 Performance Results Recent benchmark results show:

C++ and Rust implementations perform similarly well (50-60ms range) Python implementations show higher overhead but better scaling with the smart version:

500 numbers: ~0.14ms (algorithm only) 50k numbers: ~0.37ms (algorithm only) 500k numbers: ~840ms (algorithm only)

Under system load, Rust shows excellent consistency with the lowest standard deviation. Contributing Feel free to contribute by:

Adding new implementations Optimizing existing implementations Improving the benchmarking tool Adding new features or fixes

License cc