Featured Articles
Tiny technology leads to big development
Creating the laser of tomorrow
Experts discuss power outage study
DeMarco receives Chancellor's Award
Regular Features
New faculty
Charlie Chung-Ping Chen
Assistant Professor - PhD 1998, University of Texas at Austin
|
High-end microprocessor design is facing challenges from many directions. As the feature sizes shrink to 0.18 micron, the resistance and capacitive and inductive coupling of interconnects increase dramatically. As a result, transistors and interconnects need to be carefully designed together to achieve target performance and maintain signal integrity.
As the clock frequency passes one gigahertz, the millions of devices become a veritable jungle of microwave elements and timing budgets become very stringent. Traditional design methodologies can no longer survive in the design domain.
My specific research interests focus on developing breakthrough technologies to enable and facilitate high-end micro-processor design.
I am working on the following topics:
- High-speed circuit design and optimization targeting for high-end microprocessors
- Full-chip level interconnect synthesis and optimization (RLC routing, wire-sizing, repeater insertion and model order reduction)
- Transistor, interconnect, and microwave simulation and RLC extraction
- Signal integrity analysis and optimization
- Special circuit design and analysis (clock, packaging, power delivery, I/O)
- Optimization methods (mathematical programming and combinatorial optimization)
Mikko H. Lipasti
Assistant Professor - PhD 1997, Carnegie Mellon
|
My research interests include compiler optimization, runtime and operating systems and their interaction with computer architecture. I am interested in software techniques, hardware techniques, and combined hardware/software techniques for both improving the absolute performance of microprocessors and computer systems as well as reducing the complexity of implementing such high-performance systems.
The optimization of instruction through-put in super-scalar processors can be broken down into three primary problems: supplying adequate instruction flow, enabling efficient register data flow and providing high-bandwidth and low-latency memory data flow.
Instruction flow is close to being a solved problem for the next five or ten years.
Register data flow has been solved at a conceptual level, but practical implementation continues to restrict its efficiency, hence, interesting research issues remain.
Furthermore, the recent discovery of properties like value locality and program redundancy create additional opportunities for improving register data flow beyond historical limits.
Memory data flow remains the most significant and largely unsolved challenge facing computer architects today. Packaging constraints and technology trends will continue to limit the bandwidth and/or latency to a memory that is large enough to hold the working set of entire programs, forcing computer architects to come up with clever ways of utilizing limited bandwidth and tolerating long latencies.
Daniel van der Weide
Associate Professor - PhD 1993, Stanford
|
My group designs, builds and uses micro-fabricated circuits and structures for high frequency probes and sensors with an eye toward cross-disciplinary applications. Though our instruments work somewhere in the DC-THz frequency range, many are proximal probes, scanned at or very near a sample, while others measure a collective response in the far field, at distances much greater than the wavelengths they use.
Some of the proximal probes are tiny antennas that can be used with scanning probe microscope platforms to make images of an integrated circuit's topography and local electric or magnetic fields. These same probes can also be used to examine sub-surface defects in materials such as silicon and quartz, excite "artificial molecules" made with semiconductor quantum dots, and perhaps map out the structure and dynamics of ion channels in neuronal membrane. Other probes using diodes at their tips can be used for imaging local temperature and topography or directly detecting local microwave and optical fields.
Our far-field sensors are based on picosecond-pulse electronic circuits using nonlinear transmission lines and integrated antennas. We are currently using GaAs versions of these circuits for measuring components of automotive exhaust gases and for making reflection spectra of explosives and weapons for aviation security. We are also designing complete integrated-circuit coherent measurement systems to drive these circuits, and are investigating less expensive silicon realizations of these sensors and systems.
Lei He
Assistant Professor - PhD 1999, University of California, Los Angeles
|
My research focuses on computer-aided design (CAD) for interconnect-limited circuits and systems such as high-performance microprocessors. This work involves interconnect extraction, modeling and optimization for area, performance, power, signal integrity and manufacturability.
Projects that I am currently involved with include full-chip level RLC extraction considering process variations, RLC routing (with wire sizing and spacing, and buffer and shielding insertion) for high-speed circuits, inter-connect-driven floor planning, and design and analysis of clock/power networks.
I also study optimization techniques using local search, and problems to consider interconnect effects in the early stage of system design. This includes computer architecture considering interconnect effects.
In addition, I have worked on computer-aided design (CAD) frameworks and circuit simulation, especially timing analysis and fast-timing simulation.
Giri Venkataramanan
Assistant Professor - PhD 1992, UW-Madison
|
The horseshoe, the water wheel, the windmill, the steam engine ... systems for energy transport and power conversion have always been at the forefront of human imagination. Rooted in that tradition, modern electrical power converters have evolved through the interplay of historical, social, environmental and technological factors, among others. My research team uses careful empirical study, reinforced with analytical modeling to elucidate the role of these factors in determining the shape and form of electrical power converters. We research major aspects of electrical power conversion systems in different application areas including information processing, industrial drives and processes and utility power distribution. Specific research projects focus on characterization of power semiconductor devices and components, development of novel power converters and control strategies, physical realization and packaging, mitigation of converter-induced harmonics, and control of electromagnetic interference.