I’m glad CUDA and “open source” are in the same sentence again.

We’d all prefer cross platform programming, but if you’re going to do platform specific, I prefer open source to closed source.

Thank you NVIDIA!