It would be nice if __setitem__ worked with Numpy arrays as right-hand side. Currently one gets the error

    Traceback (most recent call last):
      File "cupy/core/core.pyx", line 1575, in cupy.core.core.ndarray.__setitem__
      File "cupy/core/core.pyx", line 3043, in cupy.core.core._scatter_op
      File "cupy/core/core.pyx", line 442, in cupy.core.core.ndarray.fill
    ValueError: non-scalar numpy.ndarray cannot be used for fill

A straightforward implementation could first cast to a cupy array and then do the usual setitem. An optimization for same dtype and size (assuming contiguous memory) would be to just use cudaMemcpy cupy_array.data.copy_from_host, but that may not be very relevant.