GPU Computing SDK 4.0 contains many examples for CUDA.
However some of the examples don’t work when two graphics chips present.
I have ATI radeon 3300 integrated graphics chip connected to monitor.
I have Asus/nvidia GT 520 graphics card to do cuda computing.
However I suspect some of the examples assume that the cuda computations happen on the same card as the monitor is connected too ?!?
But this is not necessarily the case.
So perhaps the examples should be changed so it can run opengl/gui on the graphics chip that’s actually connected to the monitor and leave the cuda computations to the cuda device ?!?