A fast and robust network bisection algorithm

Saab, Youssef

doi:10.1109/12.392848

Cited by 47 publications

(34 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The first such family of heuristics is hierarchical clustering [1,7,30,47,48]. The other group of partitioningrelated heuristics is the Kernighan-Lin heuristic [26], which was substantially improved by Fiduccia and Mattheyses [16], and later by many others [11,42]. These heuristics have been found to be appropriate for hardware/software partitioning as well [31,46,49].…”

Section: Previous Workmentioning

confidence: 99%

Finding optimal hardware/software partitions

Mann

Orbán

Arató

2007

Form Methods Syst Des

View full text Add to dashboard Cite

Most previous approaches to hardware/software partitioning considered heuristic solutions. In contrast, this paper presents an exact algorithm for the problem based on branch-and-bound. Several techniques are investigated to speed up the algorithm, including bounds based on linear programming, a custom inference engine to make the most out of the inferred information, advanced necessary conditions for partial solutions, and different heuristics to obtain high-quality initial solutions. It is demonstrated with empirical measurements that the resulting algorithm can solve highly complex partitioning problems in reasonable time. Moreover, it is about ten times faster than a previous exact algorithm based on integer linear programming. The presented methods can also be useful in other related optimization problems.

show abstract

Section: Previous Workmentioning

confidence: 99%

Finding optimal hardware/software partitions

Mann

Orbán

Arató

2007

Form Methods Syst Des

View full text Add to dashboard Cite

show abstract

“…Without balance criterion, we can find the optimal solution in polynomial time by maxflow-mincut algorithm [10]. In a roughly balanced bipartitioning problem, 10% of skewness is usually allowed [20]. Since it is NP-hard for general graphs [11], heuristic algorithms are used practically.…”

Section: Graph Bipartitioningmentioning

confidence: 99%

Investigation of the Fitness Landscapes and Multi-parent Crossover for Graph Bipartitioning

Kim

Moon

2003

Genetic and Evolutionary Computation — GECCO 2003

View full text Add to dashboard Cite

Abstract. An empirical study is performed on the local-optimum space of graph bipartitioning. We examine some statistical features of the fitness landscape. They include the cost-distance correlations and the properties around the central area of local optima. The study revealed some new notable results about the properties of the fitness landscape; e.g., the central area yielded fairly good quality in the local-optimum space. We performed an experiment on a spectrum of different exploitation strengths of the central areas. From the results, it seems attractive to exploit the central area, but excessive or insufficient exploitation is not desirable.

show abstract

“…A number of iterative min-cut methods for graph or hypergraph partitioning have been previously proposed [16], [17], [23], [24], [29], [30]. Kernighan and Lin proposed the wellknown KL graph partitioning algorithm using pair swaps to improve a random initial two-way partition [23].…”

Section: Ery Large Scale Integration (Vlsi) Circuitmentioning

confidence: 99%

“…It gives better partitions than FM, but requires large amounts of memory, as will be discussed shortly, thus, rendering it infeasible or very slow (due to frequent page swaps) for use on medium-to large-size circuits. Some preclustering techniques have been proposed in [29] and [34] to remedy the weakness of FM and KL in getting trapped in local minima for large circuits; the techniques in [29] apply directly to graphs, though they may be extendible to hypergraphs. Very good results have been obtained at the cost of considerable CPU time increases and implementation complexities.…”

Section: Ery Large Scale Integration (Vlsi) Circuitmentioning

confidence: 99%

“…Since two-way min-cut partitioning is NP-complete [19], a number of approximate schemes have been proposed. These include iterative improvement methods [16], [17], [23], [24], [29], [30], simulated annealing [31], [32] and clustering-based techniques [5], [7], [18], [20], [28], [29], [35], [36]. An excellent survey on partitioning techniques appears in [6].…”

Section: Ery Large Scale Integration (Vlsi) Circuitmentioning

confidence: 99%

See 1 more Smart Citation

A probability-based approach to VLSI circuit partitioning

Dutt¹,

Deng²

33rd Design Automation Conference Proceedings, 1996

View full text Add to dashboard Cite

Abstract-Iterative-improvement two-way min-cut partitioning is an important phase in most circuit placement tools, and finds use in many other computer-aided design (CAD) applications. Most iterative improvement techniques for circuit netlists like the Fiduccia-Mattheyses (FM) method compute the gains of nodes using local netlist information that is only concerned with the immediate improvement in the cutset. This can lead to misleading gain information. Krishnamurthy suggested a lookahead (LA) gain calculation method to ameliorate this situation; however, as we show, it leaves room for improvement. We present here a probabilistic gain computation approach called probabilistic partitioner (PROP) that is capable of capturing the future implications of moving a node at the current time. We also propose an extended algorithm SHRINK-PROP that increases the probability of removing recently "perturbed" nets (nets whose nodes have been moved for the first time) from the cutset. Experimental results on medium-to large-size ACM/SIGDA benchmark circuits show that PROP and SHRINK-PROP outperform previous iterative-improvement methods like FM (by about 30% and 37%, respectively) and LA (by about 27% and 34%, respectively). Both PROP and SHRINK-PROP also obtain much better cutsizes than many recent state-of-the-art partitioners like EIG1, WINDOW, MELO, PARABOLI, GFM and GMetis (by 4.5% to 67%). Our empirical timing results reveal that PROP is appreciably faster than most recent techniques. We also obtain results on the more recent ISPD-98 benchmark suite that show similar substantial mincut improvements by PROP and SHRINK-PROP over FM (24% and 31%, respectively). It is also noteworthy that SHRINK-PROP's results are within 2.5% of those obtained by hMetis, one of the best multilevel partitioners. However, the multilevel paradigm is orthogonal to SHRINK-PROP. Further, since it is a "flat" partitioner, it has advantages over hMetis in partition-driven placement applications.Index Terms-Clustering effect, iterative-improvement, min-cut partitioning, probabilistic gain, VLSI circuit.

show abstract

A fast and robust network bisection algorithm

Cited by 47 publications

References 29 publications

Finding optimal hardware/software partitions

Finding optimal hardware/software partitions

Investigation of the Fitness Landscapes and Multi-parent Crossover for Graph Bipartitioning

A probability-based approach to VLSI circuit partitioning

Contact Info

Product

Resources

About