An architecture-driven metric for simultaneous placement and global routing for FPGAs

Abstract

FPGA routing resources typically consist of segments of various lengths. Due to the segmented routing architectures, the traditional measure of wiring cost (wirelength, delay, congestion, etc) based on geometric distance and/or channel density is no longer accurate for FPGAs. Researchers have shown that the number of segments, instead of geometric (Manhattan) distance, traveled by a net is the most crucial factor in controlling the routing delay and cost in an FPGA. Further, the congestion information of a routing channel shall be measured by the available segments of specific lengths, instead of the density in a channel alone. In this paper, we propose an architecture-driven metric for simultaneous FPGA placement and global routing. The new metric considers the available segments and their lengths to optimize the wiring cost for placement and global routing. Experiments by employing a cluster growth placement and maze routing to demonstrate the new metric show respective average reductions of 8%, 20%, and 19% in the number of tracks used (area), maximum net delay, and average net delay based on the Lucent Technologies ORCA2C-like and the Xilinx XC4000EX-like architectures, compared with the traditional metric of geometric distance and channel density.