Hacker News Reader

Nvidia probably sponsored a lot of high level frameworks that only support CUDA, essentially poisoning them with lock-in.

Nope, it is up to Khronos and other card vendors to actually offer tooling beyond stone age C dialects and printf style debugging.

NVidia cannot be blamed for their incompetency.

Somebody below mentions that nvidia donated free GPUs to the developers of these open source frameworks.

Why didn't AMD do this as well? I thought AMD was strong on supporting open source.