Criticality history guided FPGA placement algorithm for timing optimization

Abstract

We present an efficient timing-driven placement algorithm for FPGAs. Our major contribution is a criticality history guided (CHG) approach that can simultaneously reduce the critical path delay and computation time. The proposed approach keeps track of the timing criticality history of each edge and utilizes this information to effectively guide the placer. We also present a cooling schedule that optimizes both timing and run time when combined with the CHG method. The proposed algorithm is applied to the 20 largest MCNC benchmark circuits. Experimental results show that compared with VPR [1], our placement algorithm yields an average of 21.7% reduction (maximum 45.8%) in the critical path delay and it runs 2.2X faster than VPR. In addition, our approach outperforms other algorithms discussed in the literature in both delay and run time.