Type of Talk: UTCS Colloquium/Architecture

Speaker/Affilia

tion: Doug Burger/Microsoft Research

Date/Time: Monday, April 26

, 2010, 3:30 p.m.

Location: ACES 2.402

Host: Kathryn McKinley<

/p>

Talk Title: How CMOS Scaling Will Affect Architectures and Applicatio

ns

Talk Abstract:

We are reaching the end stages of Moore?s Law
for CMOS.  After decades of
proportional scaling, imbalances a

re emerging that provide both peril and
opportunity going forward.&nbs

p; In this talk, I will describe how CMOS scaling
is likely to affec

t both processors and memory systems, and the implications
for future
systems.  As an illustrative example, scaling imbalances will

drive the emergence of resistive memories, which will replace Flash and
DRAM, and which will create new challenges for architects.  Of th

ese
resistive memories, Phase-Change Memory is the closest to commerc

ial
adoption, but has the challenge of early and frequent failures of
memory
cells.  I will show how Error-Correcting Pointers (ECP)

, invented at
Microsoft Research, are likely to replace Error-Correct

ing Codes (ECC) for
these resistive memories.  With these techni

ques and others, we may
eventually reach terabytes of non-volatile st

orage in mobile devices.  I
will conclude with a description of

the SONGO system, where we built a web
index cache on a mobile phone

to use these massive capacities to reduce
mobile search latencies.

Speaker Bio:

Doug Burger is Director of Client and Cloud Appl

ications research at
Microsoft Research, where he also manages the co

mputer architecture group.
He is also an Adjunct Professor at The Univ

ersity of Texas at Austin, where
he taught on the faculty from 1999-2

008.  His research interests span
processor architectures, memo

ry systems, mobile devices, cloud computing,
new semiconductor tech

nologies, and human/computer interfaces.  At
UT-Austin, he co-

led the TRIPS project with Steve Keckler and Kathryn
McKinley. 

His current main project in MSR is called Natural Computing.