Design of programmable, energy-efficient reconfigurable accelerators

Stanford University, 2018


Abstract

This dissertation develops programmable reconfigurable accelerators and compilation techniques that map high-level parallel patterns onto efficient specialized hardware.