Nvopencc usage How to use nvopencc directly

Is there any documentation for using nvopencc to compile C for device code? That is, are the parameters the same as nvcc (the device related ones, at any rate), and what are the limitations which nvcc handles during preprocessing?

Specifically, I want to compile kernals at run time without encumbering my app with all the distribution issues that the host side compiler would entail - requiring all users to install Visual Studio 2008 Express really isn’t much of a solution.

Thus, I want to know what preprocessing, specifically, is required in order to compile a .cu file containing only device code. Will this just work with nvopencc? Does anyone have any documentation for nvopencc? For that matter, is there another place I can get the nvopencc source besides Nvidia’s FTP? I have yet to get FTP to work with my network, so I have zero access to the source as of now.