Deep compression and EIE: Efficient inference engine on compressed deep neural network