Efficient and effective packing and analytical placement for large-scale heterogeneous FPGAs

Abstract

As FPGA architecture evolves, complex heterogenous blocks, such as RAMs and DSPs, are widely used to effectively implement various circuit applications. These complex blocks often consist of datapath-intensive circuits, which are not adequately addressed in existing packing and placement algorithms. Besides, scalability has become a first-order metric for modern FPGA design, mainly due to the dramatically increasing design complexity. This paper presents efficient and effective packing and analytical placement algorithms for large-scale heterogeneous FPGAs to deal with issues on heterogeneity, datapath regularity, and scalability. Compared to the well-known academic tool VPR, experimental results show that our packing and placement algorithms achieve respective 199.80X and 3.07X speedups with better wirelength, and our overall flow achieves 50% shorter wirelength, with an 18.30X overall speedup.