IO connection assignment and RDL routing for flip-chip designs

Abstract

Given a set of IO buffers and bump balls with the capacity constraints between bump balls, an O(n <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ) IO assignment and RDL routing algorithm is proposed to assign all the IO connections and minimize the total wirelength with satisfying the capacity constraints and guarantee 100% routability if the capacity constraint is permitted, where n is the number of bump balls in a flip-chip design. Compared with the combination of the greedy IO assignment and our RDL routing, our IO assignment reduces the global wirelength by 7.6% after global routing and improves the routability by 8.8% after detailed routing on the average. Compared with the combination of our IO assignment, the single-layer BGA global router[8] and our detailed routing phase, our RDL routing reduces the global wirelength by 15.9% after global routing and improve the routability by 10.6% after detailed routing on the average for some tested circuits in reasonable CPU time.