ABSTRACT As LSI become more complex every year, faster routing programs are re- quired. In ICOT, a new method for LSI routing based on concurrent objects modeling has been developed. We have implemented this on a parallel infer- ence machine PIM, and have evaluated its performance. KEY FEATURES Programming based on concurrent objects model All line segments, including free lines and occupied lines on routing grid lines, are implemented as line processes. Routing is executed by commu- nication between these line processes. Master line processes control the communication of messages between line processes. Applicability to large-scale data Distributor processes are assigned to each processor to reduce the amount of communication between processors. A one hundred percent wiring for large-scale data (2746x3643 grid and 556 nets) is attained.
![]() Process structure and processor mapping |