I am using VS2012 C++/CLI with OpenCV 3.1, CUDA 7.5, NSight 5.2 on a Win10 64 bit machine with a GTX 760 NVidia card.
All of this works fine when compiling from within VS2012. Further, I can use the VS debugger to set breakpoints, etc. within the C++ code. In this configuration, I know that at various times CUDA code is running including my own custom kernel. However, you cannot set breakpoints (at least that are recognized) in the CUDA code.
Now that my CUDA code is getting more complex, I want to use the CUDA debugger to be able to set breakpoints in the CUDA code. When I invoke NSight from within VS, all of the non-CUDA code runs fine. However, as soon as I perform an operation that would invoke CUDA code I just get a "stopped working" message that really does not give me much info as to what the problem might be.
Thus, your task is just to help me get the NSight debugger working.