Prism: A Distributed VLSI Design System


Neil Sullivan, Jonathan B. Rosenberg, Mark T. Jones, David Kotz, R. James Nusbaum, James W. O’Neil, and Herve Tardif. Prism: A Distributed VLSI Design System. Technical Report number CS-1987-21, Dept. of Computer Science, Duke University, June 1987. ©Copyright the authors.


Chip and cell design take several forms. There are mask-level systems, symbolic-level systems, silicon compilers, and standard cell systems, just to name a few. Many of these forms can be used together to help create an entire design.

This design paper describes a symbolic design system called Prism. The motivation for designing Prism arose from the desire to improve symbolic-to-mask compaction -- specifically in the VIVID system. Current compactors run as totally batch processes. Running batch, a compactor must either smash the chip hierarchy and compact the entire chip as one cell or compact individual cells, making assumptions about the environment and connections for each cell. In either case, the area of the mask suffers. Also, compactors can take an extraordinary amount of time, and one small change -- even if it would make no change in the area of the compacted mask -- requires a total recompaction.

Experiences with using and creating VIVID indicated more reasons to build Prism. VIVID is of the best existing symbolic systems, but strides in state-of-the-art communications, user interfaces, and design automation software engineering have left it behind. Prism is a descendant of VIVID, but Prism is a new model for symbolic design.

Citable with [BibTeX]

Available from the author: [bib] [pdf]
This pdf is the only definitive version available.

[Kotz research]