Hacker News Reader

What are the target GPUs for this? Will it run only on NVIDIA cards? What about mobile GPUs?

I presume it will run everywhere CUDA is supported. Draw your own conclusions.

It currently generates NVIDIA's PTX only.