[Abstract]
The conventional way of scaling performance is no longer viable due to the slowdown of Moore’s Law and the end of Dennard Scaling. Meanwhile, AI workloads have become the first-class citizen in today’s computing landscape, requiring substantial compute and memory resources to process massive amounts of data. We now reached a point where computing systems truly require innovations across the layers of the system stack with key emerging technologies considered. Will this be challenging? Absolutely. However, this is in fact a once-in-a-lifetime opportunity for computer architects to shape and lead the next 50 years of computing! In this talk, I will first discuss the challenges in designing current and future computing systems for key emerging workloads, and how computer architects in industry and academia attack the problems. I will then present some of our research efforts and share my vision towards enabling new compute paradigms in the AI computing era.

[Biography]
Jaewoong Sim is an assistant professor in the Department of Electrical and Computer Engineering at Seoul National University. Prior to that, he worked as Senior/Staff Research Scientist at Intel Labs for five years. His research interests are in the area of computer architecture and systems, including CPU/GPU microarchitectures, domain-specific accelerators/compilers/languages, and near-data computing. His work has been recognized with several awards including IEEE MICRO Top Picks in Computer Architecture and the Best Paper Award at PACT. Jaewoong received his Ph.D. in Electrical and Computer Engineering from Georgia Tech and his B.S. in Electrical Engineering from Seoul National University.