Easy question about splitting code

Accelerated Computing CUDA CUDA Programming and Performance
#1

Hi everyone. This is quite a basic question but im struggling.

I have a .cpp file with a main function and i am calling a function that sets up GPU (allocates memory etc) - however this function is in its own file.

//main.cpp

int main()

{

   set_up_gpu();

   return 0;

}

File: gpu.cu

//gpu.cu

void set_up_gpu()

{

     cudaMalloc()  // balh blah blah

     cudaMemcpy()  // blah blah blah

     callKernel()  // etc etc

}

HOWEVER, i have no idea how to deal with the header files to point to this function in main.cpp. What i have done is: i have a GPU.h that has a decleration for set_up_Gpu()

and the main has

#include “GPU.h”

BUT im getting errors! Its not executing. - ERROR LNK2005

If anyone could tell me how to structure the headers so that it compiles it would be great!!!

Thanks in advance.

#2

It’s link error, it has nothing to do with header file (which you are handing fine, in a way you described it). Make sure that you compile both main.cpp and gpu.cu.

#3

You must describe exactly what you’re doing. What platform/compiler/editor etc. What OS/distribution? Print/attach the exact main and GPU files(where is the #include that you say is there???) Are you using a command line or Microsoft Visual Studio? You must provide exact details.

PS:
You are right: it might be better to ask this question somewhere else.