Hey, I started adapting some CUDA code to take advantage of dynamic parallelism in visual studio 2012 and started getting the following linker error
error LNK2001: unresolved external symbol ___fatbinwrap_66_tmpxft_00000b3c_00000000_17_cuda_device_runtime_compute_52_cpp1_ii_8b1a5d37
With some investigation, the code compiles and links fine as long as I don’t add the ChildKernel line below:
dim3 ChildBlock(BLOCK_X_AXISCOUNT, BLOCK_Y_AXISCOUNT, 1);
dim3 ChildGrid( XGridDim, YGridDim, ZGridDim );
ChildKernel<<<ChildGrid, ChildBlock>>>( /parameters/ );
I assume the compiler notices the use of dynamic parallelism and therefore requires additional libraries to be included to support it, but I have included all the .lib files in the SDK … /lib folder and still no-go.