There is a sign up sche

dule for this event, UT EID required:

 Type of Talk:&nbsp

;  UTCS Colloquium/Architecture

Speaker/Affiliation:  Da

vid Brooks/Harvard University

Date/Time:  Thursday, January 29

, 2009  11:00 a.m.

Location: ACES 2.302, Avaya Auditorium

Host:  Steve Keckler

Talk Title:  "Computer Desi

gn in the Nanometer Scale Era: Challenges and  Solutions"

<

p>Talk Abstract:

Technology scaling has enabled tremendous growth in

the computing industry over the past few decades.  However, recent t

rends in power dissipation, reliability, thermal constraints, and device

  variability threaten to limit the continued benefits of device scal

ing and curtail performance and energy improvements in future technology ge

nerations.  The temporal and spatial scales of these  effects m

otivate holistic solutions that span the circuit,  architecture, an

d software layers.  In this talk, I will describe several projects t

hat seek to address technology scaling issues.    These p

rojects include efforts in the areas of a) power and   performa

nce modeling and design space optimization for future chip-multiprocessor s

ystems, b) variability-tolerant microarchitectures  that are flexibl

e in both latency and localized supply voltage, and  c) accelerator-

based architectures for power/performance efficiency.  The talk will

also discuss our chip prototyping efforts that support this work.

Speaker Bio:
 David Brooks joined Harvard University in

September of 2002 and is an Associate Professor of Computer Science. 
Dr. Brooks received his  B.S. (1997) degree from the University of S

outhern California and his M.A. (1999) and Ph.D (2001) degrees from Princet

on University, all in Electrical Engineering.  Prior to joining Harv

ard University,  Dr. Brooks was a Research Staff Member at the IBM T

.J. Watson  Research Center.  Dr. Brooks received an IBM Facult

y Partnership  Award in 2004, an NSF CAREER award in 2005, and a DA

RPA Young Faculty Award in 2007.  His research interests include arch

itecture and runtime software approaches to address power, reliability, a

nd variability issues for embedded and high-performance computer systems. <

/p>