How do I organize the source code of a CUDA C project using subfolders?

Question

I develop some parallel computing code using CUDA C. The system is running a Ubuntu based Linux, the IDE of choice is Eclipse Indigo.

I set up the project using the template which is delivered with Cuda. I fail to setup subfolder containing portions of code (say "gui", "io", "net") in a such a way, that the compiler (ncc and/or g++) and/or linker recognizes those.

The goal would be to just type "make" and everything is put together.

Perhaps somebody knows a project template or sample makefile which works with directory structures? I don't feel the status quo with its many files (like gui_myclass.c, net_myotherclass.c, ...) in the projects root directory is the way to go.

Answer 1

This question has nothing to do with CUDA. There is a good answer here to the question "how do I generate a makefile with source in subdirectories using just one makefile".

This applies to any command line compiler, nvcc, g++, whatever.