| GPU加速的分段Top-k查询算法 |
| |
| 引用本文： | 黄玉龙,邹循进,刘奎,苏本跃. GPU加速的分段Top-k查询算法[J]. 计算机应用, 2014, 34(11): 3112-3116. DOI: 10.11772/j.issn.1001-9081.2014.11.3112 |
| |
| 作者姓名： | 黄玉龙 邹循进 刘奎 苏本跃 |
| |
| 作者单位： | 1. 安庆师范学院 计算机与信息学院,安徽 安庆 2461332. 江西省国土资源厅 信息中心,南昌 330025 |
| |
| 基金项目： | 国家自然科学基金资助项目 |
| |
| 摘 要： | 现有Top-k查询优化算法无法充分利用图形处理器(GPU)强大的并行吞吐量及时获取查询结果,为此提出了一种基于统一计算设备架构(CUDA)模型的大规模分段查询算法。通过划分查询过程以及采用分段并行处理策略,该算法可最大限度地提升查询过程中的计算和比较效率。实验结果表明,与4线程多核优化算法相比,所提算法具有明显的性能优势，当有序列表数量为6,遍历步长为120时,性能达到最优,此时比多核算法快40倍。
|
| 关 键 词： | Top-k查询 通用计算图形处理器 分段处理 并行优化 禁止随机访问 |
| 收稿时间： | 2014-07-28 |
| 修稿时间： | 2014-08-06 |
GPU-accelerated segmented Top-k query algorithm |
| |
| HUANG Yulong,ZOU Xunjin,LIU Kui,SU Benyue. GPU-accelerated segmented Top-k query algorithm[J]. Journal of Computer Applications, 2014, 34(11): 3112-3116. DOI: 10.11772/j.issn.1001-9081.2014.11.3112 |
| |
| Authors: | HUANG Yulong ZOU Xunjin LIU Kui SU Benyue |
| |
| Affiliation: | 1. School of Computer and Information, Anqing Normal University, Anqing Anhui 246133,China;
2. Information Center, Department of Land and Resources of Jiangxi Province, Nanchang Jiangxi 330025,China |
| |
| Abstract: | The existing algorithms of Top-k query can not make full use of the powerful parallel throughput of Graphic Processing Unit (GPU) to timely return the query results. So, a segmented query algorithm based on Compute Unified Device Architecture (CUDA) model was proposed. By dividing the query process and using the strategy of segmented parallel process, the maximal calculation and comparison efficiency in query process could be obtained in this algorithm. The experimental results show that this algorithm has obvious performance advantages compared with four-thread parallel optimization algorithm on multi-core CPU. When the number of ordered lists is 6 and the traversal stride is 120, the optimal performance can be obtained which is 40 times faster than multi-core CPU algorithm. |
| |
| Keywords: | Top-k query General Purpose GPU (GPGPU) segmented process parallel optimization No Random Access (NRA) |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《计算机应用》浏览原始摘要信息 |
|
点击此处可从《计算机应用》下载全文 |
|
