CUDA rendered output display strategy

I want to start to play with ray tracing in CUDA.

My question is not about tracing rays, but how to show the resulting rendered image.

Should I write the result as a 2D array into device memory and then bind it as an OpenGL texture and display a camera-aligned quad with the map on it?

There are some example projects like Nbody and Volume that do this kind of thing… I am sure it’d work.

But I wonder what other methods exist and if they matter or not. Are the OpenGL PBO games the best way? “best” may mean portable, most efficient, simplest, best RAM use, etc, I don’t know what options there are and what the tradeoffs might be.

I’d love an SDK example of raytracing even if it were just three spinning spheres or something easy.