Concurrent Number Cruncher on Linux

Hey all,
Has anyone tried running CNC (Concurrent number cruncher - A GPU implementation of a general sparse linear solver (from old URL), International Journal of Parallel, Emergent and Distributed Systems, to appear) on Linux? The source given by them is for Windows and has various windows structs for which I can’t find the equivalent in unix?

If not this, does anyone have any code for running Sparse Matrix operations using CUDA?

Thanks,
Vandhan