Don't have Telegram yet? Try it now!
https://hgpu.org/?p=8176
Lost in Translation: Challenges in Automating CUDA-to-OpenCL Translation