I’m debugging a large visual studio solution. For some reason though I set breakpoints the CUDA debugger skips them. I’ve checked if the -g option is on in the ProjectProperties > CUDA C/C++ etc but still doesn’t work. Also the .ptx generation is enabled as well.
But still doesn’t work, once I attach the process something runs but I can’t step through the code lines. They says this should be equivalent to run the Cuda debugger, but to me still doesn’t work.
There was an issue with some of the flags of the nvcc compiler (-gencode) they weren’t matching the architecture of my gpu, whose capabilities are 6.1.
About where I set the breakpoints, one was in the CPU code, while the other is in the GPU code, but I got confused at the beginning, I realized later.
It seems to be working now, at least with the examples.