Parallel greedy graph matching using an edge partitioning approach

Patwary, Md. Mostofa Ali; Bisseling, Rob H.; Manne, Fredrik

doi:10.1145/1863482.1863493

Cited by 29 publications

(9 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Considerable interest in parallel algorithms has been observed recently with work on approximation as well as exact algorithms on shared and distributed memory platforms. Patwary et al [12] have implemented a parallel Karp-Sipser algorithm (in a general graph) on a distributed memory machine using an edge partitioning of the graph. On some real graphs, their algorithm achieved up to 38× speedups on 64 processors, whereas on other graphs their algorithm did not scale at all.…”

Section: Related Workmentioning

confidence: 99%

“…In earlier work, effective serial and parallel algorithms for maximal cardinality matching have been designed and implemented on both shared and distributed memory systems [5,9,10,11,12]. To increase the cardinality of the maximal matching, existing serial and parallel algorithms process only a small fraction of unmatched vertices at a time, which imposes a vertex-processing order.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A matrix-algebraic formulation of distributed-memory maximal cardinality matching algorithms in bipartite graphs

Azad

Buluç

2016

Parallel Computing

View full text Add to dashboard Cite

We describe parallel algorithms for computing maximal cardinality matching in a bipartite graph on distributedmemory systems. Unlike traditional algorithms that match one vertex at a time, our algorithms process many unmatched vertices simultaneously using a matrix-algebraic formulation of maximal matching. This generic matrix-algebraic framework is used to develop three efficient maximal matching algorithms with minimal changes. The newly developed algorithms have two benefits over existing graph-based algorithms. First, unlike existing parallel algorithms, cardinality of matching obtained by the new algorithms stays constant with increasing processor counts, which is important for predictable and reproducible performance. Second, relying on bulk-synchronous matrix operations, these algorithms expose a higher degree of parallelism on distributed-memory platforms than existing graph-based algorithms.We report high-performance implementations of three maximal matching algorithms using hybrid OpenMP-MPI and evaluate the performance of these algorithm using more than 35 real and randomly generated graphs. On real instances, our algorithms achieve up to 200× speedup on 2048 cores of a Cray XC30 supercomputer. Even higher speedups are obtained on larger synthetically generated graphs where our algorithms show good scaling on up to 16,384 cores.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A matrix-algebraic formulation of distributed-memory maximal cardinality matching algorithms in bipartite graphs

Azad

Buluç

2016

Parallel Computing

View full text Add to dashboard Cite

show abstract

“…[9]. The authors proposed an MPI (Message Passing Interface)-based algorithm that runs the Karp-Sipser algorithm in parallel for each edge-partitioned subgraph.…”

Section: Related Workmentioning

confidence: 99%

A Parallel Maximal Matching Algorithm for Large Graphs Using Pregel

Lim

Chung

2014

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYGraph matching is to find an independent edge set in a graph. It can be used for various purposes such as finding a cover in a graph, chemical structural computations, multi-level graph partitioning and so on. When a graph is too large to be handled by a single machine, we should use multiple machines. In this paper, we use Pregel, a cloud graph processing architecture which is able to process massive scale graph data in scalable and fault-tolerant ways. We propose a parallel maximal matching algorithm described in the Pregel's vertex-centric BSP model. We test our algorithm on an 8 node cluster and the results show that our algorithm can realize high quality matching for a large graph in a short time. Also, our algorithm is linearly scalable with the number of machines.

show abstract

“…A more in-depth discussion and comparison of such matching strategies can be found in [11]. A distributed-memory parallel implementation of the Karp-Sipser algorithm is presented in [13].…”

Section: Algorithm 1 Serially Creates a Matching Of A Graphmentioning

confidence: 99%

A GPU Algorithm for Greedy Graph Matching

Auer

Bisseling

2012

Facing the Multicore - Challenge II

Self Cite

View full text Add to dashboard Cite

Abstract. Greedy graph matching provides us with a fast way to coarsen a graph during graph partitioning. Direct algorithms on the CPU which perform such greedy matchings are simple and fast, but offer few handholds for parallelisation. To remedy this, we introduce a finegrained shared-memory parallel algorithm for maximal greedy matching, together with an implementation on the GPU, which is faster (speedups up to 6.8 for random matching and 5.6 for weighted matching) than the serial CPU algorithms and produces matchings of similar (random matching) or better (weighted matching) quality.

show abstract

Parallel greedy graph matching using an edge partitioning approach

Cited by 29 publications

References 19 publications

A matrix-algebraic formulation of distributed-memory maximal cardinality matching algorithms in bipartite graphs

A matrix-algebraic formulation of distributed-memory maximal cardinality matching algorithms in bipartite graphs

A Parallel Maximal Matching Algorithm for Large Graphs Using Pregel

A GPU Algorithm for Greedy Graph Matching

Contact Info

Product

Resources

About