Previous research in my field has been done on GeForce 8800s which can produce direct real time graphics. But reading the tech spec for the C870 I am using, in the display support section on page 2 it states that direct display output is not supported, but graphics remoting is supported.
How can I produce real-time graphics from a C870?
Its not clear what you mean by “real-time graphics”. Tesla products have no display device connectors.
I meant visualizing the output from a numerical simulation as it made, rather than saving results in, for example, numerous vtk files for visualization in Paraview.
You could try to use VNC, however I’m not sure whether that constitutes a ‘real time’ visualization mechanism.
Well, you can copy your results back to the CPU after each frame/iteration and then just use any graphics API with your other GPU (the one that your screen is connected to). It can be slow of course.
I think graphics remoting is supported if one wanted to use some GPGPU code written as shaders in OGL/D3D or something. It’s not really for displaying anything.