Efficient implementation of a synchronous parallel push-relabel algorithm
From MaRDI portal
Abstract: Motivated by the observation that FIFO-based push-relabel algorithms are able to outperform highest label-based variants on modern, large maximum flow problem instances, we introduce an efficient implementation of the algorithm that uses coarse-grained parallelism to avoid the problems of existing parallel approaches. We demonstrate good relative and absolute speedups of our algorithm on a set of large graph instances taken from real-world applications. On a modern 40-core machine, our parallel implementation outperforms existing sequential implementations by up to a factor of 12 and other parallel implementations by factors of up to 3.
Recommendations
- scientific article; zbMATH DE number 515925
- The Partial Augment–Relabel Algorithm for the Maximum Flow Problem
- On implementing the push-relabel method for the maximum flow problem
- On implementing push-relabel method for the maximum flow problem
- Processor-efficient implementation of a maximum flow algorithm
Cites work
- scientific article; zbMATH DE number 3174052 (Why is no real title available?)
- scientific article; zbMATH DE number 487935 (Why is no real title available?)
- A computational study of the pseudoflow and push-relabel algorithms for the maximum flow problem
- A new approach to the maximum-flow problem
- An O(n2log n) parallel max-flow algorithm
- Efficient implementation of a synchronous parallel push-relabel algorithm
- Engineering multilevel graph partitioning algorithms
- Graph Partitioning and Graph Clustering
- Maximum flows by incremental breadth-first search
- On implementing the push-relabel method for the maximum flow problem
- The Pseudoflow Algorithm: A New Algorithm for the Maximum-Flow Problem
Cited in
(4)
This page was built for publication: Efficient implementation of a synchronous parallel push-relabel algorithm
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3452773)