（毕业设计论文）并行算法在多核环境下的设计与实现.doc

精品 PAGE 精品 毕业设计（论文） 题 目 并行算法在多核环境下的设计与实现 学 院 计算机与电子信息学院 专 业 计算机科学与技术 学 号 0807100321 姓 名 指导教师 二零一二年五月二十五日 精品 摘 要 在信息激增的年代，我们强烈地要求计算机的处理器更快更强，在这种压力之下，双核乃至多核应运而生，多核时代已经来临。多核技术在多线程多任务等领域发挥着极大的作用，因而对软件的运行效率以及软件的开发起到了很大的影响。在多核对软件起着重大的影响的同时，它也面临着一些问题，如能耗、性能瓶颈等都是未来要解决的问题。本文根据多核的特点，简略分析快速排序及顺序查找算法的思想及其描述，并详细分析其内在并行性，提出算法基于多核虚拟环境下的改进方法，设计出适合多核平台的并行算法，使用OpenMP并行编程模型来实现；重点分析PAM聚类算法，PAM算法是一种寻找初始聚类中心的新方法，算法首先计算样本间的距离，然后根据样本点之间的距离寻找有可能是一类的数据，依据这些样本点形成初始聚类中心从而得到更好的聚类效果。实验结果都表明并行算法对多核环境有很好的适应性，算法在多核环境下比在传统环境下具有更高的运行效率。 关键词：多核 OpenMP 并行计算 快速排序 顺序查找 PAM算法 Research and Implementation of Parallel algorithm in Multicore Environment Abstract In the information explosion era, we strongly demand computer processors faster and stronger, in this kind of pressure, binucleated and multinucleated emerge as the times require, multi-core era has arrived. In the field of multi-threaded multi-tasking, multi-core technology play a great role, and thus play a very significant impact on the operating efficiency of the software and software development. In a multi-check software plays a significant impact at the same time, it is also facing some problems, such as energy consumption, performance bottlenecks are the future issues to be addressed. Based on the characteristics of multi-core, This article will briefly analyze the idea of quick sort and sequential search algorithm and its description of its inherent parallelism, and present the improved algorithm is based on multi-core virtual environment, to design a parallel algorithm for multi-core platform. This article will focus on analyzing the PAM clustering algorithm, the PAM algorithm is looking for a new method of initial cluster centers, the algorithm first calculate the distance between samples, then based on the distance between sample points to find a class data, based on these sample points to form the initial cluster centers in order to get better clustering results. The experimental results show that parallel algorit