https://hgpu.org/?p=4844
Hybrid OpenCL: Enhancing OpenCL for Distributed Processing