For more information about item submission and attendance, see About the Technical Calendar.
Friday, October 12
How Much (Execution) Time, Energy, and Power
Richard Vuduc, Georgia Institute of Technology, Atlanta
Will My Algorithm Cost?
Computer Science and Mathematics Division
Future Technologies Seminar
10:30 AM — 11:30 AM, Research Office Building (5700), Room L-202
Contact: Jeffrey Vetter (firstname.lastname@example.org), 865.576.7115
AbstractWhen designing an algorithm or performance-tuning code, is time-efficiency (e.g., operations per second) the same as energy-efficiency (e.g., operations per Joule)? Why or why not?
As a baby-step toward answering these questions, we posit a simple strawman model of the energy to execute an algorithm. Our model is the energy-based analogue of the time-based "roofline" model of Williams, Patterson, and Waterman (Comm. ACM, 2009). What do this model imply for algorithm design? What might computer architects tell algorithm designers to help them better understand whether and how algorithm design should change in an energy-constrained computing environment? Can this model be "operationalized" into development tools? Importantly, this talk is about an idea, rather than a well-developed set of results. As such, your questions, healthy skepticism, constructive feedback, and offers of collaboration may be even more welcome than usual!
This work is joint with Jee Whan Choi, Marat Dukhan, Kenneth Czechowski, and Aparna Chandramowlishwaran, all PhD students at Georgia Tech.
Rich Vuduc is an assistant professor in the School of Computational Science and Engineering at Georgia Tech. His research group, The HPC Garage (hpcgarage.org), works in high-performance computing, with a focus on parallel algorithms, performance analysis, and performance tuning. His work has been recognized by a variety of scholarly awards, including a 2010 IPDPS Best Paper; 2010 Gordon Bell Prize; and 2012 SIAM Data Mining Best Paper. He is a recipient of the National Science Foundation CAREER Award and a member of the DARPA Computer Science Study Group program. He received his PhD from UC Berkeley and served as a postdoctoral scholar at Lawrence Livermore National Laboratory.