【24h】

Parallel Job Scheduling - A Status Report

机译:并行作业调度-状态报告

获取原文
获取原文并翻译 | 示例

摘要

Parallel job scheduling has been useful for parallel processors. Recently, the strategies and algorithms have been adapted to the grid and to clusters. In the future, microprocessors are likely to contain several processors on a chip with the numbers of processors per chip rapidly increasing over the years. Servers and even personal computers will be parallel processors. These may then be organized into clusters, grids, or tightly coupled microprocessors. To a degree, this is already happening with the introduction of hyperthreading in Intel microprocessors. Job schedulers will have to deal with at least two layers of scheduling and even more if several processor chips are aggregated together into a server. Current research projects are actively building processors with tens or even hundreds of processors. When used as a fast personal computer, some gang scheduling variant makes sense. When used as a supercomputer, some variant on batch scheduling will be used. This leads to the obvious convergence of both types of scheduling strategies.
机译:并行作业调度对于并行处理器很有用。最近,这些策略和算法已适应于网格和集群。未来,微处理器可能会在一个芯片上包含多个处理器,并且随着时间的推移,每个芯片的处理器数量迅速增加。服务器甚至个人计算机将是并行处理器。然后可以将它们组织为群集,网格或紧密耦合的微处理器。在某种程度上,英特尔微处理器中引入超线程已经在发生这种情况。作业调度程序将必须处理至少两层调度,如果将多个处理器芯片聚合到一个服务器中,则调度甚至还要更多。当前的研究项目正在积极地构建具有数十甚至数百个处理器的处理器。当用作快速个人计算机时,某些帮派调度变体是有意义的。当用作超级计算机时,将使用批处理调度的某些变体。这导致两种调度策略的明显收敛。

著录项

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