I am trying to find out that my system supports nvidia graphics card and cuda or not. My system has Intel graphics card there is no nvidia card installed on my system but i want to check because i am writing a generic lib.
Statically link to cudart library (which is the default when compiling with nvcc, but you can statically link with your host compiler as well).
Then run cudaGetDeviceCount at the beginning of your program.
[url]CUDA Runtime API :: CUDA Toolkit Documentation
check both the returned device count as well as the runtime error code.