Adaptive and Polymorphic Computer Architectures KEC 1003 Monday, November 24, 2014 - 4:00pm to 4:50pm
Michel Kinsy Assistant Professor Department of Computer and Information University of Oregon Abstract: The current design approach in multicore or many-core computer systems presents application programmers with a great deal of challenges due to their ever-increasing complexity. Unlike frequency-scaling where performance is increased equally across the board, core-scaling pushes the burden of harnessing this additional processing power on the application or software programmer. To make optimal use of the system components, programmers must first learn about system parameters and how to best leverage them for a given application. This approach requires time, effort, and often leads to suboptimal application performance in term of execution time or power. As a solution to this problem, we propose the Helios Multicore System (HMS), an adaptive computer architecture framework that applies machine learning algorithms and control theory techniques to the application execution by reasoning about the system runtime performance trade-offs. It has heterogeneous reconfigurable cores with fast hardware-level migration capability, self-organizing memory structures and hierarchies, an adaptive and quality-of-service aware network-on-chip, and a built-in hardware layer for dynamic, autonomous resource management. Biography: Michel A. Kinsy is an Assistant Professor in the Department of Computer and Information at the University of Oregon and director of the Computer Architecture and Embedded Systems (CAES) Laboratory. He received his Ph.D. in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology (MIT) in 2013. His doctoral work is one of the first to develop algorithms and hardware techniques to emulate and control large-scale power systems at the microsecond resolution. Part of this work is currently being further developed and commercialized through an MIT spin-off. Prior to joining the University of Oregon, Dr. Kinsy was a Technical Staff at the MIT Lincoln Laboratory working on advanced computer architecture concepts. Dr. Kinsy is an MIT Presidential Fellow and holds an M.S. in Electrical Engineering and Computer Science from MIT, a B.S.E. in Computer Systems Engineering and a B.S. in Computer Science from Arizona State University. Prof. Michel Kinsy's research interests lie in the general area of computer system design with particular emphasis on self-aware, adaptive, high-performance many-core architectures, computer hardware security, very large scale integration (VLSI) systems design, intelligent network-on-chip (NoC), cyber-physical systems (CPS), reconfigurable hardware systems, large-scale distributed systems, and hard real-time embedded systems. He is particularly interested in the application of machine learning techniques to hardware execution. -- _______________________________________________ Colloquium mailing list [email protected] https://secure.engr.oregonstate.edu/mailman/listinfo/colloquium
