Move from C and Cuda module to Using C++

Good morning,
I have been using a module coded in cuda and C using the possibility to directly use tensor create in python into C and then cuda code. However with the 1.0 I get the “ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.”
How could I still use the same principle of using python created tensor in C++ and cuda now ? I havent found any tutorial or explanations to move my code from C to C++

1 Like