Parallel Cluster Identification for Multidimensional Lattices

Authors:
Stephen J. Fink;Scott B. Baden;Craig Huston;Karl Jansen
Affiliations:
Univ.of California at San Diego, La Jolla;Univ.of California at San Diego, La Jolla;Hewlett-Packard Corp., Escondidio, CA;CERN, Geneva, Switzerland
Venue:
IEEE Transactions on Parallel and Distributed Systems
Year:
1997

Citing 7
Cited 2

Computer simulation methods: in theoretical physics

Computer simulation methods: in theoretical physics
Computer architecture: a quantitative approach

Computer architecture: a quantitative approach
Introduction to algorithms

Introduction to algorithms
Cluster identification algorithms for spin models—sequential and parallel

Concurrency: Practice and Experience
Parallel Algorithms for Geometric Connected Component Labeling on a Hypercube Multiprocessor

IEEE Transactions on Computers
Digital Image Processing

Digital Image Processing
Evaluation of Connected Component Labeling Algorithms on Shared and Distributed Memory Multiprocessors

IPPS '92 Proceedings of the 6th International Parallel Processing Symposium

Software infrastructure for non-uniform scientific computations on parallel processors

ACM SIGAPP Applied Computing Review
Irregular Coarse-Grain Data Parallelism under LPARX

Scientific Programming

Quantified Score

Hi-index	0.00

Visualization

Abstract

The cluster identification problem is a variant of connected component labeling that arises in cluster algorithms for spin models in statistical physics. We present a multidimensional version of Belkhale and Banerjee's Quad algorithm for connected component labeling on distributed memory parallel computers. Our extension abstracts away extraneous spatial connectivity information in more than two dimensions, simplifying implementation for higher dimensionality. We identify two types of locality present in cluster configurations, and present optimizations to exploit locality for better performance. Performance results from 2D, 3D, and 4D Ising model simulations with Swendson-Wang dynamics show that the optimizations improve performance by 20-80 percent.