██╗     ██╗████████╗███╗   ███╗██╗   ██╗███████╗███╗   ███╗ █████╗ ████████╗██╗  ██╗
██║     ██║╚══██╔══╝████╗ ████║██║   ██║██╔════╝████╗ ████║██╔══██╗╚══██╔══╝██║  ██║
██║     ██║   ██║   ██╔████╔██║██║   ██║███████╗██╔████╔██║███████║   ██║   ███████║
██║     ██║   ██║   ██║╚██╔╝██║██║   ██║╚════██║██║╚██╔╝██║██╔══██║   ██║   ██╔══██║
███████╗██║   ██║   ██║ ╚═╝ ██║╚██████╔╝███████║██║ ╚═╝ ██║██║  ██║   ██║   ██║  ██║
╚══════╝╚═╝   ╚═╝   ╚═╝     ╚═╝ ╚═════╝ ╚══════╝╚═╝     ╚═╝╚═╝  ╚═╝   ╚═╝   ╚═╝  ╚═╝
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

A C++ header only library that implements and benchmarks various numerical methods

Dependencies and Requirements

• Compiler : Tested with g++-14 and above, with --std=c++23
• • uses <print>, <span>, and <ranges> .
• MPI : OpenMPI or equivalent

currently under development, not intended for use.

Build instructions

1. Configure CMake: cmake -B build -DTESTING=ON -DCMAKE_BUILD_TYPE={Release|RelWithDebInfo|Debug}

2. Build and Compile: cmake --build build

   • Default CMAKE_BUILD_TYPE = RelWithDebInfo

   • Default Testing = OFF

3. Run Benchmarks: ./run_bench_mm <min_pow_2_size> <max_pow_2_size> <n_iterations>
   

   • default config: ./run_bench_mm 2 9 100

Current Features

• Matrix Multiplication

  • Naive implementation
  • Transpose before multiplication for cache friendly behaviour
  • Strassen's Algorithm

• Statistics

  • Compute basic measures of central tendency

• Benchmarking and Timing

  • Using chrono::steady_clock

• Testing

  • Implemented using GoogleTest and is fetched by CMake during build.

Some Preliminary Benchmarks

Hardware: AMD 5800X3D, 4.1GHz, 32GB DDR4

Compiler: gcc15
Flags: -std=c++23 -O2 -g

Planned Features

• Matrix algebra methods
• More robust benchmarking using RDTSC
• Parallel MPI implementations
• SIMD implementations
• CUDA implementations

Authors

Dr. Abhimanyu Bhadauria, Ph.D