https://hgpu.org/?p=12060
Parallelization of Shape Diameter Function Computation using OpenCL