Does/will Cuda support template class/struct?

I actually need only

template<int N>

struct TMyData {

	const static int Size = N; //N for optimization purposes will be 32*k, k is int

	unsigned char Data[N];

};

template<class Data> 

__global__ void Operation(const Data* data) {

   for (int i = 0; i < Data::Size; ++i)

	   ....

}

P.S. But even this simple usage combined with template specialization can lead to complex compile time calculation.

Can anybody say in which version of Cuda it will be supported?