Code for GPU-accelerating arbitrary-sized matrix-matrix multiplication in Python by exposing C++ and CUDA code to Python using Pybind11.

• Cuda installed in /usr/local/cuda
• CMake 3.3 or later
• Python 3.8.10 or later
• PythonInterp 3.6 or later
• PythonLibs 3.6 or later

Should compile out of the box by doing the following:

sudo chmod +x bind_code.sh
./bind_code.sh

python3 matmul.py