Feedback  Site Map  
  Other Researches

   Speech Recognition
   Optical Flow
   Neural Network

  Request More Info

Custom hardware is faster and more compact than general-purpose hardware. Integrated circuits (ICs) are the fastest and smallest computing hardware, but they can only be customized for a particular set of tasks, they require significant non-recurring engineering (NRE) costs, and they quickly become obsolete. For many applications, the cost of replacing obsolete ICs is prohibitive.

Fortunately, technological developments have substantially increased gate densities, making reconfigurable computers, based on field programmable gate arrays (FPGAs), an attractive alternative. An FPGA blurs the distinction between hardware and software. The "hardware" fabricated by a foundry is general purpose; consequently they are mass produced and affordable. The logic of an FPGA is customized by loading a "configuration," which is similar to a software program. The resulting FPGA combines the best of both worlds. It is faster and smaller than truly general-purpose hardware such as a workstation, yet compared to an IC, it has smaller NRE costs and transition costs, for it can be easily recustomized without modifying the hardware by designing and loading a different configuration. A reconfigurable computer could be upgraded, or even reconfigured for a completely different function, from a remote location.

Furthermore, reconfigurable computers can serve as powerful research and development tools for sophisticated electronic systems such as ICs and printed circuit boards. Simulation tools for these systems do not always exist, and when they do exist, they are not always capable of simulating large, complex designs. In addition, prototype fabrication is expensive and time consuming. A reconfigurable computer, on the other hand, can serve as an affordable, fast, and accurate tool for verifying electronic designs.

Tanner Labs Research and Development

Tanner Labs has developed three reconfigurable computers. The first was for speech recognition. The other two, for optical processing and large neural network processing, are even more flexible than a typical reconfigurable computer because they are modular. They can easily be adapted for other applications by designing a few specialized components.

We plan to continue our reconfigurable computer research and development by refining and enhancing our technology, as well as by developing user-friendly configuration design tools that enable the user to quickly and easily develop and modify the logic using VHDL. In addition, we are using our reconfigurable computer systems as research and development tools for algorithm development and simulation.


Back to top

Home | Products | MEMS | R & D | Contact Info | Feedback | Site Map

Copyright © 1999-2001 by Tanner Research, Inc. All Rights Reserved