C++ Design Patterns: Matrix Multiplication Library

A matrix multiplication library written in C++. This project serves as a learning exercise for implementing and understanding common C++ design patterns, primarily:

• Factory Pattern
• Registry Pattern
• Meyers' Singleton Pattern

Key Feature: New multiplier implementations (e.g., single-threaded CPU, CUDA-accelerated GPU, OpenCL) can be registered and used via the factory without requiring modifications to the factory itself. Each multiplier is generated using a config, and such, it may be useful, for example, in performance testing multiple algorithms/hyperparameters.

Read more about this project on my blog!

Current Status:

• Includes a registry that contains all Multiplier implementations.
• Includes a single-threaded CPU multiplier implementation using Eigen.
• Includes CUDA and OpenMP implementations for comparison (with SIMD-> AVX2+NEON).
• Includes a globally accessible Logger singleton for handling application-wide logging.
• All backends are created and configured via a central JSON config file.

Dependencies:

• C++17 (for thread-safe static variables, move semantics, smart pointers, and try_emplace)
• CMake 3.22.1 or higher
• Eigen (included as a git submodule)
• Json (included as a git submodule)

Building and Running

1. Clone the Repository:

   git clone git@github.com:AkeelMedina22/cpp-design-patterns.git
   git submodule update --init --recursive

2. Configure with CMake:

   mkdir build
   cd build
   cmake ..

3. Build the project:

   make

4. Run the example application:

   src/matrix_app ../src/config.json

How to Use

#include <memory>
#include <iostream>
#include <string>
#include "matrix_transform/matrix_types.hpp"
#include "matrix_transform/interfaces.hpp"
#include "matrix_transform/factory.hpp"

using MatrixTransform::Factory;
// ...
str::string configFilePath = argv[1]; 
auto multiplier = Factory::createMultiplier(configFilePath);
Matrix a = /* ... */;
Matrix b = /* ... */;
Matrix result = multiplier->multiply(a, b);
// ...