I opened up a Sample CUDA project and I could build+run it, but it was not aware of some include files in “C:\ProgramData\NVIDIA Corporation\CUDA Samples\v5.5\common\inc”. This causes errors to appear on the source file editor for include files and unknown functions.
The solution is at the bottom of this post.
I have Visual Studio 2012 Professional and CUDA 5.5 SDK, toolkit, etc.
In Section 3.1 of this document:
http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-microsoft-windows/index.html
It says we can open up the Visual Studio 2012 project in v5.5\1_Utilities\bandwidthTest\
I open up the bandwidthTest VS 2012 project.
Go to the Project Menu and update it.
Open up the bandwidthTest.cu source file.
Immediately I can see an error on this line:
#include <helper_functions.h>
It says it doesn’t know where it is. If I try to open it up, it shows an error:
The helper_functions.h is in “…/…/common/inc/”.
One can see the Current Source File Path does not contain “…/…/common/inc/”
If I open up the project properties:
The CUDA C/C++ menu item on the left shows that
Additional Include Directories contains: “./;…/…/common/inc”
But this is not enough. I don’t know why the CUDA additional include directories don’t work, but that’s how it is.
I can work around this by going to the VC++ Directories menu item on the left and appending the include directory to the end of the
Include Directories: “$(DXSDK_DIR)\include;$(IncludePath);…/…/common/inc/”
edit:
Another example is similar:
"CUDA Samples\v5.5\0_Simple\matrixMul"
It builds and runs, but the source editor has issues.
I added “…/…/common/inc/”, but still #include <cuda_runtime.h> does not exist in the source editor.
I also add "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.5\include"
Then #include <cuda_runtime.h> shows up as valid in the source editor.