首页> 外文会议>Proceedings of the 43rd annual symposium on frequency control 1989 >Achieving a balanced low-cost architecture for mass storagemanagement through multiple fast Ethernet channels on the Beowulfparallel workstation
【24h】

Achieving a balanced low-cost architecture for mass storagemanagement through multiple fast Ethernet channels on the Beowulfparallel workstation

机译:通过Beowulf parallel工作站上的多个快速以太网通道,实现平衡的低成本体系结构,用于海量存储管理

获取原文
获取原文并翻译 | 示例

摘要

A network of workstations (NOW) seeks to leverage commercialnworkstation technology to produce high-performance computing systems atncosts appreciably lower than parallel computers specifically designednfor that purpose. The capabilities of technologies emerging from the PCncommodity mass market are rapidly evolving to converge with those ofnworkstations while at significantly lower cost. A new operating point innthe price-performance design space of parallel system architecture maynbe derived through parallelism of PC subsystems. Then“Pile-of-PCs” (PopC) approach is being explored through thenBeowulf parallel workstation, developed to provide order-of-magnitudenincreases in disk capacity and bandwidth for a single-user environmentnat costs commensurate with conventional high-end workstations. Thisnpaper explores a critical aspect of the architecture trade-off space fornBeowulf associated with the balance of parallel disk throughput andninternal network bandwidth. The findings presented demonstrate that thenparallel channels of a commodity 100 Mbps Ethernet are both necessarynand sufficient to support the data rates of multiple concurrent filentransfers on a 16-processor Beowulf parallel workstation
机译:工作站网络(NOW)试图利用商用工作站技术来生产高性能计算系统,而其成本却明显低于为此目的而专门设计的并行计算机。 PCncommodity大众市场新兴技术的功能正在迅速发展,以与工作站的技术融合,同时成本大大降低。并行系统体系结构的性价比设计空间中的新工作点可能无法通过PC子系统的并行性获得。然后,通过当时的Beowulf并行工作站探索“ PC堆”(PopC)方法,该并行工作站的开发旨在为单用户环境提供磁盘容量和带宽的数量级增加,而成本却与传统高端工作站相当。本文探讨了Beowulf架构权衡空间的关键方面,它与并行磁盘吞吐量和内部网络带宽的平衡相关。提出的发现表明,商品100 Mbps以太网的并行通道既必要又足以支持16处理器Beowulf并行工作站上多个并行文件传输的数据速率。

著录项

相似文献

  • 外文文献
  • 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