Real-time CUDA decoding H.264 to OpenGL

Hi - I am attempting to build a solution for real-time decoding of streaming H.264 video to be used by an OpenGL application in Windows. The constraint is that I want to perform the entire decode operation on the GPU in order to avoid memory transfer overhead. Should I be using the CUDA video decoder (Video Decoder :: CUDA Toolkit Documentation)?

Is there any better alternative? I also thought about DXVA, but need the decoded rgb frame to be used by an OpenGL application so it does not seem like a reasonable option.

Oddly, the VideoSource class provided by NVIDIA as part of the CUDA video decoder only contains the capability of reading a file (not a stream) which makes me think that it is not suitable for the streaming case. Has anyone been able to use it for the streaming case? If so, how?

Also, I am nervous about the fact that the CUDA decode has not been updated since 2010. The documentation for the CUDA decoder seem scattered at best. Is it no longer supported??