Write a device kernel that performs the double precision BLAS operation
dcopy, i.e. y = x using GPU threads in a 2D grid.
Assume that the vectors are used to store a 600x400 matrix (in row-major format). Initialise the matrix with some values on the CPU and confirm the correctness of the calculation, e.g. by comparing to reference values calculated on the CPU or by printing out some of the results.
You may start from a skeleton code provided in copy2d.cpp.