UCSD´S COPPER CONTEST PAGE

Embedded Systems Design Automation and Test Group (ESDAT)
Department of Electrical and Computer Engineering
University of California, San Diego

Principle Investigator : Sujit Dey

UCSD´s ESDAT group is involved in the SRC´s copper contest. The point of the contest is to show how copper technology can significantly improve the functionality and performance of integrated circuits and systems. Part of our research in the ESDAT lab is to expand the usage of system-on-chips into the multimedia mobile arena (see our mobile multimedia project web page). Copper technology gives us enough performance to allow for complex, mobile, multimedia system-on-chips. Therefore, the ESDAT group is actively involved in the copper contest.

In this webpage, we:

Motivation for our system-on-chip:

With the recent advent of wireless communications and the internet, there is a growing demand to allow for high-throughput, low-power communications. In addition to voice and data communications, the need for image and video communication over wireless channels is quickly increasing. Therefore, it is critical to minimize the energy consumed by communicating multimedia content while still maintaining other necessary communication parameters (such as image/video quality, transmission delay, tolerable bit error rate, etc). Rather than designing for the worst channel conditions, or the best image/video quality, or the smallest transmission latency, our objective is to enable energy-aware multimedia communication which dynamically adapts current algorithms and algorithmic parameters (and therefore the energy consumption) to the current channel conditions and user demands.

Critical to such adaptive multimedia mobile communications is a computational engine capable of adaptive image and video compression/decompression. By using different algorithms and algorithmic parameters, this computational engine can adjust for the lowest energy consumption for a given set of communication parameters. This computational engine must be flexible enough to execute different algorithms and their parameters, powerful enough to do them quickly, and efficient enough to perform with minimal energy consumption.

As a first step towards this powerful computational engine, we are designing our copper technology system-on-chip which can execute different image compression algorithms with their parameters.

Description of our copper system-on-chip:

 Our image compression system-on-chip was designed with these three goals in mind:
  1. The design must have low energy consumption
  2. The system must be powerful enough to compress images quickly
  3. The same system must be usable for several different image compression algorithms with different parameters.
To meet these three goals, we developed an architecture based around a general purpose processor (picoJava) with hardware accelerators and on-chip memory as shown in Figure 1. The hardware accelerators, which implement compute-intensive transforms (the DCT and DWT) for image compression algorithms, help us to meet goals 1 and 2 by accelerating the image compression routines while consuming less power than a general purpose processor would. Using a general purpose processor, on the other hand, gives us the flexibility we need to execute different algorithms with different parameters (goal 3). Using on-chip memory also allows us to use less energy (no external bus communications) while increasing the speed of image compression. The components of the system-on-chip communicate using the PI-Bus soft core.


Figure 1: System-on-Chip Architecture for reconfigurable image compression

Why copper is advantageous

 Copper is advantageous in this design because it provides us with the speed we need to perform reconfigurable image compression. With large designs, interconnect (such as the system bus) can become the limiting factor in the design. Copper helps to mitigate these problems, providing us with the flexibility we need to build a powerful, reconfigurable image compression system.

Publicity given the copper contest, including us

 Because our team has been advanced to phase II of the copper contest, we have received some positive publicity. These are two articles that have been published concerning the contest.
article 1
article 2

People Involved:

 Sujit Dey, PhD
Pablo Sanchez, PhD
Li Chen
Clark Taylor
Dong-Gi Lee
Debashis Panigrahi
Xiaoliang Bai
Yi Zhao
View the status of our project. (internal web page, password required)
Please direct any feedback to Clark Taylor
Last modified : 27 September 2000