I have built the cppIntegration project successfuly. Then I wanted to have my own .cu file inside my own project.
I have set all the includes properly, and selected this same build role choice as in cppIntegration for the CUDA compiler.
I compiled and then I got this:
I don’t know what to do with “Toolkit\3.2\inc\ -m… etc etc” (Sorry I am not at work now)
First, kudos for the bizzare error “I don’t know what to do with”, it puzzled me for a few minutes what this error is suppose to mean.
You know, something like “Unexpected option\keyword\whatever” would have been more helpful, to say the least. Or even better if you could write “expecting ‘-’\path\whatever”
After one or two hours of trying to understand what is wrong, I figured the nvcc can’t handle the spaces in the NVIDIA toolkit path. It’s weird, because nvcc can handle the same path in the cppIntegration but not in my project.
So I used SymbLink to work around the numerous spaces in the Toolkit path. The cppIntegration worked with this, but now compiling the .cu in my project returns “Cannot find specified path” (or something. not at work). Ok, is it so hard to specify which path it cannot find? That would be helpful.
So I believe that it has to do with something like VS sends unicode to nvcc or something like that, and nvcc doesn’t like it. I think it has to be in the properties of the project somewhere, because nvcc shouts only in my project and not in the cppIntegration. But apart from that, I don’t know. And I had to leave work for today.
Any help will be appreaciate.