Primality Tester 9000 (prime9k)

Overview

A brute force primality testing program written in Python. It uses a 6k +/- 1 algorithm, that is parallelized using either the CPU or GPU (depending on what's available/selected).

Manual Installation

Requirements

• python>=3.8,<3.13
• pip
• CUDA (optional, for GPU acceleration)

Installation

1. Clone this repository, and change directory.

   $ git clone https://github.com/ralgar/prime9k.git
   $ cd prime9k

2. Create and activate a virtual environment.

   $ python -m venv venv
   $ source venv/bin/activate

3. Install the Python package.

   $ pip install .

Usage

1. Invoke the help dialog to learn how to use the program.

   $ prime9k --help
   usage: prime9k [-h] [-d {cpu,gpu}] [-v] INTEGER
   
   Primality Tester 9000
   
   positional arguments:
     INTEGER               a single integer to test (max value: 9223372036854775807)
   
   options:
     -h, --help            show this help message and exit
     -d {cpu,gpu}, --device {cpu,gpu}
                           override selected compute device
     -v, --verbose         show more detailed logs
   

Running with Docker

If you don't want to mess with venvs and dependencies, a single-command Docker Compose build is provided.

Requirements

• Docker
• NVIDIA Container Toolkit (optional, for GPU acceleration)

Usage

1. Build and start the container using Docker Compose.

   $ docker compose --profile [cpu,gpu] up

2. Once the program has finished execution, bring the compose stack down.

   $ docker compose --profile [cpu,gpu] down

Tests

Prime9k comes with a dynamic testing framework that leverages a well-known algebra library (SymPy) to generate valid prime and composite numbers at runtime. A wide range of values are tested (over 20,000 combined prime and composite numbers), and both the CPU and GPU compute algorithms are tested.

After following the Manual Installation steps:

1. Install the test optional dependencies.

   $ pip install .[test]

2. Invoke pytest from the project root.

   $ pytest

License

Copyright: (c) 2025, Ryan Algar (ralgar/prime9k)

BSD 2-clause License (see LICENSE or BSD 2-clause)