Friday - March 4, 2011
Physics Research Building - Room 595
3 Cummington Street
Professor Ayse Kivilcim Coskun
Electrical & Computer Engineering
Software Stack for Energy-Efficient Computing"
Operators of large-scale datacenters and high
performance computing systems increase cost-efficiency
and throughput by clustering workload on their
systems. In tandem, there is a strong incentive
to lower operational costs by replacing HVAC
systems with less expensive cooling infrastructures.
However, utilizing cheaper cooling is likely
to result in higher temperatures on systems,
degrading system reliability. Thermal problems
accelerate with the integration of more cores
on a single chip and the use of novel integration
and packaging techniques such as 3D stacking.
To reduce the operational costs while keeping
reliability at desired levels, we need new techniques
for improving energy efficiency and thermal
profiles, both at the chip level and at the
This talk focuses on the challenges and the
techniques we developed to cope with increasing
performance demands, high temperatures, and
high energy consumption for multiprocessor systems.
The key message is that software optimization
at various levels, such as workload scheduling,
is an effective tool for substantially improving
thermal profiles while delivering the desired
performance and lowering system-level energy
Ayse K. Coskun is currently an assistant professor
in the Electrical and Computer Engineering Department
at Boston University. She received her MS and
PhD degrees in Computer Science and Engineering
from University of California, San Diego. Ayse’s
research interests are energy-efficient computing,
multicore systems, 3D stack architectures, computer
architecture, embedded systems and software.
Ayse has worked at Sun Microsystems (now Oracle),
San Diego with the Systems Dynamics Characterization
and Control team for 3 years prior to her
current position at BU. Ayse received the best
paper award at IFIP/IEEE VLSI-SoC Conference
in 2009, and won the Dean’s Catalyst Award
in 2010 at College of Engineering, Boston University
for her research in green computing.