🚀 go-pugleaf

ECE SEMINAR
Thursday, January 13, 2000
12:00
HH 1112

Diana Marculescu
Department of Electrical and Computer Engineering
University of Maryland, College Park

ABSTRACT


Power Modeling and Optimization in High-Performance Systems

             
Power dissipation has become a critical design concern in recent years,
driven by the increased levels of complexity and emergence of mobile
applications. Furthermore, an increasing amount of system functionality
tends to be realized through software, which is leveraged by high
performance of modern processors. Thus, reduction of power consumption
of processors is important for the
power-efficient design of such systems.

In this talk, I will present my most recent work in the area of power
modeling and optimization at behavioral, architectural and software
level.  The talk will be centered around power modeling and optimization
issues in superscalar processors. Specifically, I will present a
compiler-assisted technique for code annotation that adaptively selects
the energy optimal execution rate. Using this technique, significant
savings can be obtained in both energy and power consumption, at the
expense of some decrease in performance. I will continue with presenting
an energy-driven run-time data routing paradigm that is able to reduce
significantly the switched capacitance in the execution path of
superscalar processors. I will conclude the talk with an overview of my
present and future research activities.   


BIO

Diana Marculescu received her M.S. in Computer Science from
"Politehnica" University of Bucharest, Romania and the Ph.D. in Computer
Engineering from University of Southern California in 1998. She is
currently an Assistant Professor in the Department of Electrical and
Computer Engineering at University of Maryland, College Park. Her
reserach interests are in the general area of VLSI/computer
architecture, specifically power modeling and optimization of
high-performance computing systems.

ECE SEMINAR
Thursday, January 13, 1999
12:00
HH 1112

Diana Marculescu
Department of Electrical and Computer Engineering
University of Maryland, College Park

ABSTRACT


Power Modeling and Optimization in High-Performance Systems

             
Power dissipation has become a critical design concern in recent years,
driven by the increased levels of complexity and emergence of mobile
applications. Furthermore, an increasing amount of system functionality
tends to be realized through software, which is leveraged by high
performance of modern processors. Thus, reduction of power consumption
of processors is important for the
power-efficient design of such systems.

In this talk, I will present my most recent work in the area of power
modeling and optimization at behavioral, architectural and software
level.  The talk will be centered around power modeling and optimization
issues in superscalar processors. Specifically, I will present a
compiler-assisted technique for code annotation that adaptively selects
the energy optimal execution rate. Using this technique, significant
savings can be obtained in both energy and power consumption, at the
expense of some decrease in performance. I will continue with presenting
an energy-driven run-time data routing paradigm that is able to reduce
significantly the switched capacitance in the execution path of
superscalar processors. I will conclude the talk with an overview of my
present and future research activities.   


BIO

Diana Marculescu received her M.S. in Computer Science from
"Politehnica" University of Bucharest, Romania and the Ph.D. in Computer
Engineering from University of Southern California in 1998. She is
currently an Assistant Professor in the Department of Electrical and
Computer Engineering at University of Maryland, College Park. Her
reserach interests are in the general area of VLSI/computer
architecture, specifically power modeling and optimization of
high-performance computing systems.