Kai Li

Professor of Computer Science
Associated Faculty in the Department of Electrical Engineering
Ph.D. 1986, Yale University

My research interests are in computer architecture, operating systems, parallel systems, and networking. My current research projects focus on architecture and system issues in building high-performance servers with PC clusters, architectural support for graphics, design issues of building scalable storage servers, and extensible network routers.

SHRIMP (Scalable High-performance Really Inexpensive MultiProcessor). This project investigates how to construct high-performance servers with a network of commodity PCs, and commodity operating systems. The cost of such a multicomputer server is substantially less than a commercial, custom-designed multicomputer. The goal is to study how to build such a system to deliver performance competitive with or better than the commercial multicomputer servers. The research project consists of several components: user-level, protected communication, efficient message-passing, shared virtual memory, distributed file system, performance measurement, and applications. This is a joint project with Douglas Clark, Edward Felten, and Jaswinder Pal Singh of the Department of Computer Science, and Margaret Martonosi of the Department of Electrical Engineering.

Immersive Computer System. This project explores research issues on how to build and use immersive computer systems. In particular, we are interested in building immersive systems for users to collaborate across space and time. Current research topics that we are pursuing include seamless imaging, parallel rendering, data visualization, intelligent networking, spatialized sound, camera-based tracking, and design methodologies for creating applications. This is a joint project with Douglas Clark, Perry Cook, Adam Finkelstein, and Thomas Funkhouser.

Extensible Router. This project investigates how to build a router for the next-generation Internet. Driven by pressures to push functionality from end devices back into the network, this project recognizes the need for routers to move from being closed, special-purpose network devices to being open, general-purpose computing/communication systems. The central challenge in making this shift is to support increasingly complex forwarding logic and high performance, while using commercial hardware components and open operating systems. Our approach achieves this through better integration of the router's switching capacity and its compute cycles, and by supporting a hierarchy of paths through the router. This path hierarchy ranges from fast/fixed paths implemented in hardware to slow/programmable paths implemented in software, but also includes intermediate paths that exploit the improved integration of cycles and switching. This is a joint project with Larry Peterson and Edward Felten of the Department of Computer Science.