Friday, October 4

11am - 12pm

Hybrid: SEC LL2.221 & Zoom (p/w: 025207)

Title: Tailored Computing: Domain-Specific Architectures for Next-Generation Systems

Abstract: For over 50 years, rapid innovation in general-purpose microprocessors drove exponential growth in computing performance. Recently, however, this progress has slowed due to critical limitations, including the end of Dennard Scaling and Dark Silicon, where power constraints leave large portions of microprocessors underutilized. In response, the field has shifted toward domain-specific architectures (DSAs), tailoring systems to specific application domains. This talk demonstrates how DSAs can unlock dramatic performance and energy efficiency improvements, often ranging from one to several orders of magnitude. However, it also addresses a key challenge with DSAs: the "whack-a-mole" problem, where each new domain or evolving requirement necessitates creating custom solutions from scratch. Current approaches have largely been reactive and piecemeal, requiring immense engineering effort to develop and redevelop specialized architectures as needs change. To meet future computing demands across diverse domains, we must fundamentally rethink DSA design. This talk introduces a holistic, scalable methodology that integrates application discovery, systems thinking, and hardware-software co-design. Our approach automatically detects key domain characteristics and leverages them to generate optimized processor architectures and system designs. We employ a range of techniques, from system morphology analysis to data-driven methods, to identify important computational patterns. This allows us to efficiently develop optimized designs for specific requirements. Moreover, by considering the entire ecosystem in which these DSAs operate, including software stacks and cross-domain interactions, we optimize both the architectural and algorithmic aspects of our solutions. Through case studies in robotics, machine learning, and the Internet of Things (IoT), the talk demonstrates how our tailored methodology can shape the next generation of computing systems and deliver significant improvements in performance, adaptability, and efficiency across various application domains, all while sustainably shaping our digital future.

Bio: Dr. Vijay Janapa Reddi is an Associate Professor of Engineering and Applied Sciences at Harvard University, where his research focuses on the intersection of computer architecture, machine learning systems, and autonomous agents. His multidisciplinary expertise drives advancements in efficient and intelligent computing systems across scales, from mobile and edge platforms to Internet of Things (IoT) devices. Prior to joining Harvard, Dr. Janapa Reddi was an Associate Professor in the Department of Electrical and Computer Engineering at the University of Texas at Austin. In addition to his academic role, Dr. Janapa Reddi is deeply involved in shaping the future of machine learning and edge AI technologies. He serves as Vice President and co-founder of MLCommons, a nonprofit organization dedicated to accelerating machine learning innovation. In this capacity, he oversees the MLCommons Research organization, sits on its board of directors, and co-led the development of the MLPerf Inference benchmark, which evaluates a wide range of ML systems from megawatt to microwatt scales. Dr. Janapa Reddi also serves on the board of directors for the tinyML Foundation, nurturing nascent technologies and fostering academic-industry partnerships in edge AI. Throughout his career, Dr. Janapa Reddi has earned numerous awards and accolades, including the Gilbreth Lecturer Honor from the National Academy of Engineering (NAE) in 2016, the IEEE TCCA Young Computer Architect Award (2016), the Intel Early Career Award (2013), and Google Faculty Research Awards in 2012, 2013, 2015, 2017, and 2020. He has also received Best Paper awards at the 2020 Design Automation Conference (DAC), the 2005 International Symposium on Microarchitecture (MICRO), and the 2009 International Symposium on High-Performance Computer Architecture (HPCA). Additionally, he has won various honors and awards, including IEEE Top Picks in Computer Architecture (2006, 2010, 2011, 2016, 2017, 2022, 2023). He is included in the MICRO and HPCA Halls of Fame (inducted in 2018 and 2019, respectively). Dr. Janapa Reddi is passionate about expanding access to applied machine learning and promoting diversity in STEM. He has developed an open-source book, "Machine Learning Systems," which is being adopted by institutions worldwide. Additionally, he created the Tiny Machine Learning (TinyML) series on HarvardX, a massive open online course that has trained nearly 100,000 students globally. Dr. Janapa Reddi holds a Ph.D. in computer science from Harvard University, an M.S. in electrical and computer engineering from the University of Colorado at Boulder, and a B.S. in computer engineering from Santa Clara University.