< Back to previous page

Project

Design Automation and Exploration for Energy Efficient Machine Learning SoCs and Chiplets

The state-of-the-art research for artificial intelligence (AI) aims to bring these power hungry algorithms into edge computing. This necessitates ultra-low energy computing despite requiring complex heterogenous systems. To find the best designs, we need several prototyping of system-on-chips (SoC) architectures. To pave the way for fast experimentation, this research work aims to explore novel algorithms, a library of energy efficient accelerators, reconfigurable network-on-chips (NoC) architectures, design exploration and automation. The goal is to develop a platform that provides convenient ways to explore new AI architectures. Starting from top-level software models that can automatically generate area and energy-efficient SoCs.

Date:31 Jan 2023 →  Today
Keywords:Energy Efficient Computer Architectures, Machine Learning Accelerators, Design Automation, Design Exploration
Disciplines:Computer architecture and organisation, Machine learning and decision making, Computer aided engineering, simulation and design, Digital integrated circuits
Project type:PhD project