Don't have Telegram yet? Try it now!
https://hgpu.org/?p=9680
A model of dynamic compilation for heterogeneous compute platforms