cuda + opengl offscreen rendering via ssh

I want to run and application that uses both opengl and CUDA to build a volumetric object (by rendering slices to a texture) and run computations on it. Is it possible to do this via command line only when I am ssh’d into the machine without interrupting anyone else who is using the machine’s screen.


try virtualGL

It uses the remote X to render OpenGL to an offscreen buffer, then sends that over the network, without disturbing other users. The advantage over ‘indirect XGL’ is ofcourse that you use the remote acceleration capabilities.