Don't have Telegram yet? Try it now!
https://hgpu.org/?p=3271
Implementing a GPU Programming Model on a non-GPU Accelerator Architecture