https://hgpu.org/?p=10171
Matrix Convolution using Parallel Programming