Sign Up to our social questions and Answers Engine to ask questions, answer people’s questions, and connect with other people.
Login to our social questions & Answers Engine to ask questions answer people’s questions & connect with other people.
Lost your password? Please enter your email address. You will receive a link and will create a new password via email.
Please briefly explain why you feel this question should be reported.
Please briefly explain why you feel this answer should be reported.
Please briefly explain why you feel this user should be reported.
In C/CUDA, if I am passed a pointer, how can I programmatically determine whether the pointer points to memory on the CPU or the GPU? Is there a CUDA function for this purpose?
I don’t know how it is with CUDA 4.0, but in 3.2 you simply cannot deduct if it is CPU or GPU pointer based solely on the address stored in it.
A way around it would be to create a class
CudaPtr<T>, hide a pointer there and provide basic functionality (e.g.memset,memload,memstore,memallocetc. which would in turn use CUDA stuff).Then, if you have a function which may operate on host or on device, you can easily overload it for
T*andCudaPtr<T>.Update: Since the Unified Memory Addressing and new CUDA toolkit things have changed a bit. It seems you can now make such distinction at run time via cuPointerGetAttribute