• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Adaptive Scheduling Using Support Vector Machine on Heterogeneous Distributed Systems.
【24h】

Adaptive Scheduling Using Support Vector Machine on Heterogeneous Distributed Systems.

机译：支持向量机在异构分布式系统上的自适应调度。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Since the advent of the modern von Neumann computer in the 1940s, startling advances have been made in computing technology with the creation of innovative, reliable, and faster electronic components, from vacuum tubes to transistors. Computing power has risen exponentially over relatively brief periods of time as Moore's law projected a salient trend for the growth in memory-chip performance, estimating the capacity of the integrated circuit to double every 18 months. Although these developments were essential in solving computationally intensive problems, faster devices were not the sole contributing factor to high performance computing. Since electronic processing speeds began to approach limitations imposed by the laws of physics, it became evident that the performance of uniprocessor computers was limited. This has led to the prominent rise of parallel and distributed computing. Such systems could be homogeneous or heterogeneous. In the past decade homogeneous computing has solved large computationally intensive problem by harnessing a multitude of computers via a high-speed network. However, as the amount of homogeneous parallelism in applications decreases, the homogeneous system cannot offer the desired speedups. Therefore, heterogeneous architectures to exploit the heterogeneity in computations came to be a critical research issue. In heterogeneous computing (HC) systems consisting of a multitude of autonomous computers, a mechanism that can harness these computing resources efficiently is essential to maximize system performance. This is especially true in mapping tasks to heterogeneous computers according to the task computation type, so as to maximize the benefits from the heterogeneous architecture.;The general problem of mapping tasks onto machines is known to be NP-complete, as such, many good heuristics have been developed. However, the performance of most heuristics is susceptible to the dynamic environment, and affected by various system variables. Such susceptibility makes it difficult to choose an appropriate heuristic. Furthermore, an adaptable scheduler has been elusive to researchers. In this research, we show that using a support vector machine (SVM) an elegant scheduler can be constructed which is capable of making heuristic selections dynamically and which adapts to the environment as well. To the best of our knowledge, this research is the first use of SVM to perform schedule selections in distributed computing. We call the novel meta-scheduler, support vector scheduler (SVS). Once trained, SVS can perform the schedule selections in O(n) complexity, where n is the number of tasks. Using rigorous simulations, we evaluated the effectiveness of SVS in making the best heuristic selection. We find that the average improvement of SVS over random selection is 29%, and over worst selection is 49%. Indeed, SVS is only 5% worse than the theoretical best selection. Since SVS contains a structural generalization of the system, the heuristic selections are adaptive to the dynamic environment in terms of task heterogeneity and machine heterogeneity. Furthermore, our simulations show that the SVS is highly scalable with number of tasks as well as number of machines.
机译：自1940年代现代冯·诺依曼（von Neumann）计算机问世以来，计算机技术取得了惊人的进步，从真空管到晶体管的创新，可靠和更快的电子组​​件的创建。随着摩尔定律预测内存芯片性能的显着趋势，计算能力在相对较短的时间内呈指数增长，估计集成电路的容量每18个月翻一番。尽管这些进展对于解决计算密集型问题至关重要，但更快的设备并不是影响高性能计算的唯一因素。由于电子处理速度开始接近物理定律所施加的限制，因此很明显单处理器计算机的性能受到限制。这导致并行和分布式计算的显着兴起。这样的系统可以是同质的或异质的。在过去的十年中，同类计算通过高速网络利用大量计算机解决了计算量大的问题。但是，随着应用程序中同类并行度的减少，同类系统无法提供所需的加速比。因此，在计算中利用异构性的异构体系结构已成为一个关键的研究课题。在由大量自治计算机组成的异构计算（HC）系统中，有效利用这些计算资源的机制对于最大化系统性能至关重要。在根据任务计算类型将任务映射到异构计算机上以最大程度地利用异构体系结构的好处时尤其如此。;将任务映射到计算机上的一般问题众所周知是NP完全的，因此很多启发式技术已经得到发展。但是，大多数启发式方法的性能容易受到动态环境的影响，并受各种系统变量的影响。这种敏感性使选择合适的启发式方法变得困难。此外，对于研究人员而言，一种可调整的调度程序已经难以捉摸。在这项研究中，我们表明，使用支持向量机（SVM）可以构建优雅的调度程序，该调度程序能够动态地进行启发式选择，并且也能适应环境。据我们所知，这项研究是首次使用SVM在分布式计算中执行计划选择。我们将这种新型的元调度程序称为支持向量调度程序（SVS）。经过培训后，SVS可以以O（n）复杂度执行计划选择，其中n是任务数。使用严格的模拟，我们评估了SVS在做出最佳启发式选择中的有效性。我们发现，相对于随机选择，SVS的平均改进为29％，而较差选择的SVS的平均改进为49％。实际上，SVS仅比理论上的最佳选择低5％。由于SVS包含系统的结构概括，因此启发式选择在任务异质性和机器异质性方面适应动态环境。此外，我们的仿真表明，SVS具有很高的可扩展性，可以处理任务数量和机器数量。

著录项

  • 作者

    Park, Yongwon.;

  • 作者单位

    Auburn University.;

  • 授予单位 Auburn University.;
  • 学科 Engineering Mechanical.;Computer Science.;Artificial Intelligence.
  • 学位 Ph.D.
  • 年度 2011
  • 页码 115 p.
  • 总页数 115
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号