A GPU-Accelerated Algorithm for Self-Organizing Maps in a Distributed Environment.
2012 (English)Conference paper (Refereed)
In this paper we introduce a MapReduce-based implementation of self-organizing maps that performs compute-bound operations on distributed GPUs. The kernels are optimized to ensure coalesced memory access and effective use of shared memory. We have performed extensive tests of our algorithms on a cluster of eight nodes with two NVidia Tesla M2050 attached to each, and we achieve a 10x speedup for self-organizing maps over a distributed CPU algorithm.
Place, publisher, year, edition, pages
self-organizing maps, gpu computing, mapreduce, graphics hardware, Computer Science
Research subject Library and Information Science
IdentifiersURN: urn:nbn:se:hb:diva-6782Local ID: 2320/10966OAI: oai:DiVA.org:hb-6782DiVA: diva2:887484
20th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. Bruges, Belgium.
Amazon Web Services2015-12-222015-12-22