TL;DR
This paper demonstrates that lattice cluster construction can be efficiently vectorized, significantly accelerating algorithms like Swendsen-Wang on high-performance computing systems.
Contribution
It introduces a vectorization approach for lattice cluster construction, enabling substantial speedups in cluster algorithms.
Findings
CPU time reduced by a factor of 3.5 to 7 on Cray Y-MP
Vectorization applies to single cluster variants of Swendsen-Wang
Significant performance improvements in practical applications
Abstract
Contrary to conventional wisdom, the construction of clusters on a lattice can easily be vectorized, namely over each ``generation'' in a breadth first search. This applies directly to, e.g., the {\it single cluster} variant of the Swendsen-Wang algorithm. On a Cray Y-MP, total CPU time was reduced by a factor 3.5 -- 7 in actual applications.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.