Abstract
“PI” is an advanced LISP-based placement and interconnect system for custom NMOS or CMOS (single-layer metal) designs. When fully implemented, PI will handle placement of arbitrarily-sized rectangular modules, routing of power and ground, signal routing, and compaction. In this paper we briefly review the structure of PI, and present details on the signal-routing heuristics, focusing on the definition of “channels”, the global router, the “crossing placer”, and the channel routers. The signal router is fully operational; the rest of PI is currently being coded and will be more fully described in later papers and theses.

This publication has 7 references indexed in Scilit: