"Humans and Bits:Designing Holistic Computer Architectures"

 

Abstract:

Electronic systems are taking an increasing role in everyday life, which is arguably one of the main causes for the social changes we have observed in the last two decades.  Hence, there is a growing need for understanding user satisfaction with computing devices.  In the first part of my talk, I will first describe our work in trying to understand the relation between computer hardware performance and user satisfaction.  Our experiments have shown that user satisfaction with computers is highly user-dependent, i.e., there is large variation in expected performance among individual users.  To take advantage of this variation, we are developing biometric input devices for providing information about the user’s physiological traits.  The goal in this work is to understand the users’ involvement/satisfaction by monitoring their physiological traits and making architectural decisions accordingly.  I will describe our initial experiences with three biometric devices as potential sensors: an eye tracker, a galvanic skin response (GSR) sensor, and force sensors.  Our initial experiments show that there are significant changes in human physiological traits as processor performance changes and we can develop a system that captures such changes to efficiently control CPU frequency. 

 

I will then describe our work in understanding smartphone user behavior. We have developed a logger application for Android mobile phones and released the logger into the wild to collect traces of real user activity.  I will describe how these traces can be used to characterize energy consumption and describe a method to minimize energy consumption by taking advantage of “change blindness”, i.e., the inability of humans to understand changes in their environment when focused on another mental activity.

 

In the final part of my talk, I will describe our work on integrating nanophotonic interconnect to future processor architectures.  Specifically, we have argued that power consumption of these communication devices is an important hurdle and have developed several novel techniques to overcome this limitation.  I will finally describe our efficient arbitration method for on-chip nanophotonic networks that can also be used for quality of service.

 

Biography:

Gokhan Memik is an Associate Professor at the Electrical Engineering and Computer Science Department of Northwestern University. He received the B.S. degree in Computer Engineering in 1998 from Bogazici University and PhD in Electrical Engineering from University of California at Los Angeles (UCLA) in 2003 under the direction of William H. Mangione-Smith.  Dr. Memik works in the area of computer architecture.  He is the author of 2 book chapters and over 100 refereed journal/conference publications.  Papers co-authored by him have been nominated for best paper awards at 2005 Design Automation Conference (DAC) and 2008 International Symposium on Microarchitecture (MICRO), and won the Best Student Paper Award at 2007 Supercomputing (SC).  He is also the co-author of NetBench and MineBench, two widely used benchmarking suites for networking and data mining applications, respectively.  He has served in over 40 program committees, was the co-chair for the Advanced Networking and Communications Hardware Workshop (ANCHOR) held in conjunction with ISCA between 2004 and 2006, and the program co-chair of 2007 International Symposium on Microarchitecture (MICRO).  He is the recipient of the Wissner-Slivka Junior Chair (2006-2009), National Science Foundation CAREER Award (2008-2013), Department of Energy Early CAREER PI Award (2005-2008), Searle Teaching Excellence Fellowship (2004-2005), Henry Samueli Excellence in Teaching Award (2002), and Henry Samueli Fellowship (2001-2002).

 

___________________________________________________________________________