I am implementing the CUDA function for concating tensors.
Is there any cuDNN Library function for it?
NCHW[1,10,10,10] and [1,10,10,10] = then [2,10,10,10] which I want to get.
My CUDA code what I implemented is inefficient even I did it.
Please tell me any good Idea to utilize exist functions of CUDA Lib, (cuBlas or anthing else is okay)