Другие журналы
|
scientific edition of Bauman MSTUSCIENCE & EDUCATIONBauman Moscow State Technical University. El № FS 77 - 48211. ISSN 1994-0408
Global unconstrained particle swarm optimization on graphics processors with CUDA arhitecture
# 04, April 2010 DOI: 10.7463/0410.0142202
Article file:
Article_4.pdf
(841.86Kb)
Parallel particle swarm optimization method GIPSO for GPUs and appropriate algorithm and implementation with CUDA techology are described.Different aspects of method efficiency are studied. On representative class of test functions it is shown that GIPSO has greater convergence speed in comparison with classic non-parallel method. The experimental results also showed significant speedup and good scalability relative to a problem size.Based on study it can be stated that a number of global optimization problems can be efficiently solved on GPUs with GIPSO method.In future works it is planned to develop and implement GIPSO metaoptimization method. Publications with keywords: global optimization, Method of a plenty of particles, parallel algoritm Publications with words: global optimization, Method of a plenty of particles, parallel algoritm See also: Thematic rubrics: Поделиться:
|
|
|||||||||||||||||||||||||||||
|