Don't have Telegram yet? Try it now!
https://hgpu.org/?p=6075
A case study on porting scientific applications to GPU/CUDA