I also have a similar problem.
My programm works only in emulator mode, are You solved this problem ?
You can try to compile it without Makefile by command
nvcc --link *.cu -o test -I/usr/local/cuda/include -L/usr/local/cuda/lib -I/root/NVIDIA_CUDA_SDK/common/inc -L/root/NVIDIA_CUDA_SDK/common/lib/linux libcutil.a
You can watch how many devices supported by code
CUT_DEVICE_INIT(argc, argv);
r=cudaGetDevice(&dev);
r=cudaGetDeviceCount(&devices);
printf(“Devices=%d\n”,devices);
for(i=0;i<devices;i++)
{
r=cudaGetDeviceProperties(&prop,i);
printf(“\nDevice[%d].name=%s\n”,i,prop.name);
printf(“\nDevice[%d].totalGlobalMem=%d\n”,i,prop.totalGlobalMem);
printf("\nDevice[%d].sharedMemPerBlock=%d\n",i,prop.sharedMemPerBlock);
printf("\nDevice[%d].warpSize=%d\n",i,prop.warpSize);
}
In my case I see only emulator device, but CUDA works,
I tesed it by screensavers and it show about 4000 FPS on some.