Implementing density functional theory (DFT) methods on many-core GPGPU accelerators