Assuming that you've setup your CUDA environment all ready... You may need to adjust: NVIDIA_SDK -> The location of the nvidia sdk. LIBS -> The location of cudart.lib and the cufft.lib CUFLAGS -> check the following flags -ccbin --> set to the location of your compiler. -IC:/CUDA/v3.2/include <-- set to the location of cuda header files -IC:/CUDA/v3.2/include/cudart <-- set to the location of cudart header files ######################################################################### # Installation (Linux, CUDA >= 3.0 (might work on older releases, too)) # ######################################################################### Makefile assumes that nvcc is in the path. Redefine NVCC if nvcc is not in your path. Likewise, add the appropriate -I includes and -L libdirs to the COMMON_INCLUDES and COMMON_LIBS, respectively. To build, just run "make" ######################################################################### # Installation Windows ######################################################################### A GNU compatible version of make is also required as the Makefile is not compatible with nmake. GNU Make for Win32 can be dowloaded from http://gnuwin32.sourceforge.net/packages/make.htm. Rename makefile.win64 or makefile.win32 to Makefile Adjust the path to your CUDA installation as desribed above in "Makefile" and run 'make'. Run the Visual Studio command prompt for 32 bit builds or the x64 Command Prompt for 64 bit builds and change into the subdirectory where the source files are and run make. Note: You can not compile using a 64 bit cuda installation on a 32 bit platform.