I get the following error while trying to build a CUDA/C++ code in Visual Studio 2012. I'm using CUDA v5.0
1> Generating Code... 1>LINK : warning LNK4044: unrecognized option '/MLd'; ignored 1>cublas_device.lib(kepler_sm35_gemm_wrapper.obj) : error LNK2019: unresolved external symbol __cudaRegisterLinkedBinary_59_tmpxft_00001040_00000000_8_kepler_sm35_gemm_wrapper_cpp1_ii_9402ff4f referenced in function "void __cdecl __sti____cudaRegisterAll_59_tmpxft_00001040_00000000_8_kepler_sm35_gemm_wrapper_cpp1_ii_9402ff4f(void)" (?__sti____cudaRegisterAll_59_tmpxft_00001040_00000000_8_kepler_sm35_gemm_wrapper_cpp1_ii_9402ff4f@@YAXXZ) Anybody has any idea of where the problem might be? I've tried googling around, but haven't found a solution for this as yet.
Thanks a lot!
-c/-dc/-rdcswitches getting passed to individual modules and/or the final link step. It would be helpful if you could paste into the original question the complete command lines issued tonvccfor each module and the final link step to build your application. Also, by any chance is this a project that you migrated forward from cuda 4.2 (or something earlier) to cuda 5.0?