Compiling CUDA applications typically requires you to pass the level of Compute Capability to the compiler. For example, Genga is compiled by typing

$ make SM=XX

where XX is the Compute Capability of the cards the code will run on. For cards available at the ICS, use the following.