Multi-resource aware partitioning algorithms for FPGAs with heterogeneous resources

Abstract

As FPGA densities increase, partitioning-based FPGA placement approaches are becoming increasingly important as they can be used to provide high-quality and computationally scalable solutions. However, modern FPGA architectures incorporate heterogeneous resources, which place additional requirements on the partitioning algorithms because they now need to not only minimize the cut and balance the partitions, but also they must ensure that none of the resources in each partition is oversubscribed. In this paper, we present a number of multilevel multi-resource partitioning algorithms that are guaranteed to produce solutions that balance the utilization of the different resources across the partitions. We evaluate our algorithms on twelve industrial benchmarks ranging in size from 5,236 to 140,118 vertices and show that they achieve minimal degradation in the min-cut while balancing the various resources. Comparing the quality of the solution produced by some of our algorithms against that produced by hMETIS, we show that our algorithms are capable of balancing the different resources while incurring only a 3.3%-5.7% higher cut.

Library-Based Placement and Routing in FPGAs with Support of Partial Reconfiguration

“…For example, [Jamieson et al 2013] introduced the genetic algorithm for solving the heterogeneous FPGA placement. [Selvakkumaran and et al 2004] proposed a multilevel multi-resource partitioning algorithm for heterogeneous FPGA placement. [Hu 2006] employed a multi-layer density system for the heterogeneous FPGA placement.…”

Multi-personality partitioning for heterogeneous systems

“…Netlist/Hypergraph partitioning based on variants of the KLFM algorithm [4] often optimize for metrics such as channel width/congestion, ease of routability, and operating frequency [5]. Some work has also examined multi-resource constraints for heterogeneous devices for KLFM [2] and other algorithms [3], but unlike our work, they do not integrate personality selection for multi-personality nodes into partitioning. In general, our proposed changes do not conflict with other KLFM partitioning extensions.…”

“…1) Statically-Mapped Partitioning (SM): This strategy is based on the Native Multi-Constraint Refinement partitioning method [2] proposed for heterogeneous FPGAs, modified to use Resource-Affinity Buckets. We first apply our global implementation remapping algorithms to achieve the target RUR.…”