Abstract:
Computing is in the midst of a data-driven revolution that has exposed significant and costly data movement inefficiencies in conventional system designs. In response, there has been much research on and prototyping of in-memory computing (IMC), which can eliminate data movement energy by performing programmable CPU-like operations directly inside memory arrays. However, while IMC promises large efficiency improvements, critical challenges in system integration have stalled its commercialization. A large contributing factor is the bespoke nature of today’s IMC proposals, making their design and viability intricately dependent on specific memory technologies and very low-level assembly-like interfaces. In many ways, this is reminiscent of the early days of von Neumann computers, prior to the advent of critical abstractions across the system stack.
In this talk, I will showcase three of our recent efforts at breaking IMC free from this bespoke design approach, by introducing cross-stack-layer abstractions that address longstanding system integration issues. These abstractions (1) eliminate technology-specific microarchitecture design for IMC, (2) expose a universal architectural abstraction for programmable IMC that supports CPU-free end-to-end application execution and opens pathways to IMC systems software development, and (3) unify analog IMC and digital IMC under a single programmable hybrid architecture to tackle edge computing problems. I will also briefly highlight our ongoing efforts for memory-driven optimization in both IMC-based and conventional systems, including programming frameworks, improved support for managed languages, coordinated voltage scaling, and in-memory support for edge continual learning.
Bio:
Saugata Ghose is an assistant professor in the Siebel School of Computing and Data Science at the University of Illinois Urbana-Champaign, where he leads the ARCANA Research Group. He holds M.S. and Ph.D. degrees in electrical and computer engineering from Cornell University, and dual B.S. degrees in computer science and in computer engineering from Binghamton University, State University of New York. He has been recognized with the U.S. National Science Foundation CAREER Award, an Intel Rising Star Faculty Award, induction into the ISCA and HPCA Halls of Fame, and a Wimmer Faculty Fellowship at Carnegie Mellon University. Several of his works have been winners or finalists for best-of-venue awards, and one was a finalist for the 2022 Intel Hardware Security Academic Award. His current research interests include data-oriented computer architectures and systems, new interfaces between systems software and hardware, energy-efficient memory and storage, and architectures for emerging platforms and domains. For more information, please visit his website at https://ghose.cs.illinois.edu/.
