A practical and efficient vector quantization called partitioned vector quantization (P-VQ) based non-linear precoder for MIMO downlink broadcasting channels has been analyzed in this paper. P- VQ has been found to be an efficient way of reducing the memory requirements and search complexity in conventional VQ systems, especially for large MIMO. Simulation results reveal that P-VQ technique can enhance the overall quantization performance in terms of achieved capacity, bit error rate (BER) at a cost of reasonable additional complexity under certain feedback budgets. Tomlinson-Harashima precoding (THP), a power and complexity efficient precoding technique along with data dependent vector perturbation (VP) has been employed to meet the power constraint requirements of our system. Least-square (LS) channel estimation is performed at the user terminals to acquire their channel state information (CSI).