I just put a new computer together. In it I have a 9800GX2 video card (dual gpu for gaming and other video purposes - play) and a C870 tesla board (for number crunching - work). I installed the sample programs. Saw there was a SobelFilter sample and figured since I had just done some sobel filtering using regular c++ in visual studio that it’d be a good one to start with.
I double clicked the .sln file and it opened up Visual Studio. Throwing caution to the wind, I clicked the build button. Expecting to see 60+ pages of errors and warnings since I hadn’t really done anything to apart from installing the telsa stuff and the studio program, I was pleasantly surprised when it came back saying there were 0 errors and 0 warnings. Yippie.
Feeling bold, I clicked the start button. Up popped a cmd window (old dos style text window) with an error message saying…
Graphics interoperability on multi GPU systems currently not supported.
I suspect this is because it’s trying to load some of this processing onto the 9800GX2 video board instead of just the tesla. What do I have to do to set cuda up to work with only tesla. At the moment, I’m attempting to hunt down how to enable/disable SLI. It doesn’t seem to be in the same place as it was with previous versions of the driver. I don’t even see anything having to do with 3D graphics acceleration.