| 基于C++ AMP加速并行蚁群算法 |
| |
| 引用本文： | 林超.基于C++ AMP加速并行蚁群算法[J].现代电子技术,2014(23):69-71. |
| |
| 作者姓名： | 林超 |
| |
| 作者单位： | 中国石油大学 华东 网络及教育技术中心,山东 青岛,266580 |
| |
| 基金项目： | 山东省基金《先验信息融合的数据驱动建模方法研究与应用(2013ZRE28090)》资助 |
| |
| 摘 要： | 蚁群算法与同类智能算法相比具有计算速度快、收敛迅速、算法稳定性好等优点,但是随着数据量的增大,计算用时呈现指数型增长。为了更好地解决大数据量蚁群算法计算慢的问题,结合蚁群算法天然的并行性,基于最新的GPU并行化接口C++AMP实现了并行蚁群算法,使计算用时大幅度减少。经试验分析,该算法可以达到3倍的加速效果。
|
| 关 键 词： | 蚁群算法 并行蚁群算法 C++AMP GPU计算 |
Accelerated parallel ant colony algorithm based on C++ AMP |
| |
| LIN Chao.Accelerated parallel ant colony algorithm based on C++ AMP[J].Modern Electronic Technique,2014(23):69-71. |
| |
| Authors: | LIN Chao |
| |
| Affiliation: | LIN Chao;Network and Educational Technology Center,China University of Petroleum; |
| |
| Abstract: | Compared with similar intelligent algorithms,ant colony algorithm has the advantages of faster calculation,more rapid convergence and more perfect stability. However,with the increasing amount of data,its computation time emerges the exponential growth. In order to solve the problem that the large amount of data ant colony algorithm is slow,by utilizing the natural parallelism of ant colony algorithm,the parallel ant colony algorithm was achieved on the basis of the latest GPU parallel interface C＋＋AMP,which realized the substantial reduction of computation time. The analysis result indicates that the reduction extent is up to 3 times the acceleration effect. |
| |
| Keywords: | ant colony algorithm parallel ant colony algorithm C++ AMP GPU computation |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
