Markov clustering based placement algorithm for island-style FPGAs

Previous divide-and-conquer methods including partition-based and cluster-based placement algorithms have shown the effectiveness of good quality-runtime trade off. This paper proposes a new cluster-based FPGA placement algorithm called CAP, which is based on a Markov clustering algorithm that defines a sequence of stochastic matrices by operating on a generating matrix from the input FPGA circuit netlist. The core of the proposed CAP algorithm is the approach that tightly couples a Markov clustering process with an deterministic placement process. Experimental results demonstrate CAP´s excellent adaptability to island-style FPGAs. The average wirelength results produced by CAP are 11.1% shorter than the wirelength results of VPR.