We are developing a product based on multiple Tesla cards (currently C1060 and 1070. We may support a low end system using a Quadro card). We need to combine cuda programming with some OpenGL programming. We do not want/need any windows, just offscreen buffers (FBO or pbuffers). Unfortunately, we have to use MS Windows (Windows Server 2008). Do I read correctly that the GPU affinity WGL extension does NOT work on Tesla’s? What happens if I create an OpenGL context on a windows system with Tesla’s. Will the OpenGL commands get replicated to the Teslas or only on the graphics card? What if there is no graphics card, just Tesla’s? Is there some other way to limit OpenGL rendering to a specific Tesla card. Any advice in general on multi-Tesla OpenGL interop?
As a followup on this, I am now understanding that Tesla cannot do OpenGL under windows (specifically Windows Server 2008R2). Can someone please verify whether or not this is true? I have only found comments that people don’t think that it works, but no definitive answer.
i found this post by you from 2010. Did you have any progress on this issue in the last two years? Specifically, did you find a definitive answer on whether Tesla can do OpenGL on Windows Server 2008R2?