https://hgpu.org/?p=2810
Comparing CUDA and OpenGL implementations for a Jacobi iteration