Show simple item record

dc.contributor.authorSchulz, Christian Ferdinand
dc.identifier.citationSINTEF Rapport A19678, 21 p., SINTEF, 2011nb_NO
dc.description.abstractWe investigate the performance increase potential of GPU implementations of local search. In particular, we report on how we managed to incrementally improve the implementation of a local search algorithm to a given GPU platform for maximum performance. As our target problem we use the well known Vehicle Routing Problem (VRP). The VRP is a family of computationally very hard problems with high industrial relevance. In particular, we investigate the 2-opt and 3-opt neighborhoods for the Distance constrained Capacitated VRP (DCVRP). Our final GPU implementation utilizes the GPU architecture efficiently. It is nearly one order of magnitude faster than the first implementation. Oppdragsgiver: The Research Council of Norwaynb_NO
dc.rightsNavngivelse-Ikkekommersiell-IngenBearbeidelse 3.0 Norge*
dc.titleEfficient Local Search on the GPU. Investigations on the Vehicle Routing Problemnb_NO
dc.typeResearch reportnb_NO
dc.relation.projectStiftelsen SINTEF: 90A36301nb_NO

Files in this item


This item appears in the following Collection(s)

Show simple item record

Navngivelse-Ikkekommersiell-IngenBearbeidelse 3.0 Norge
Except where otherwise noted, this item's license is described as Navngivelse-Ikkekommersiell-IngenBearbeidelse 3.0 Norge