Don't have Telegram yet? Try it now!
https://hgpu.org/?p=8538
Early evaluation of directive-based GPU programming models for productive exascale computing