Small correction, Tang's classical algorithm considers only low rank matrices, HHL still is more efficient for higher rank matrices.