Efficient Parallel Simulation

高效并行仿真

• 批准号: 9108448
• 负责人: Paul Reynolds
• 金额: $ 8.04万
• 依托单位: University of Virginia Main Campus
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1991
• 资助国家: 美国
• 起止时间: 1991-07-15 至 1993-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9108448&HistoricalAwards=false
• 关键词: Efficient; Parallel; Simulation

Research in parallel discrete event simulation (PDES) protocols is characterized by a large number of unrelated empirical studies, a small set of analytic studies and to date no realizable approach that is optimal or universally practical. This research project is extending a framework incorporating a low-cost network that supports rapid dissemination of critical synchronization information. Such a framework should, at a minimum, provide speed-ups for all existing protocols. More promising, however, is its potential to support adaptive protocols-those that employ aggressive processing selectively. Among all protocols, these have the greatest likelihood of having provably minimal expected execution times. Success in this research could lead to the construction of PDES engines that would simulate arbitrary applications in minimum expected execution times. Further benefits could be gained from framework-supported, application-dependent performance enhancements based on "look-ahead" and "future lists."

并行离散事件模拟 (PDES) 协议研究的特点是大量不相关的实证研究、少量的分析研究，并且迄今为止还没有可实现的最佳或普遍实用的方法。 该研究项目正在扩展一个包含低成本网络的框架，该网络支持关键同步信息的快速传播。 这样的框架至少应该为所有现有协议提供加速。 然而，更有希望的是它支持自适应协议的潜力——那些有选择地采用积极处理的协议。 在所有协议中，这些协议最有可能具有可证明的最短预期执行时间。 这项研究的成功可能会导致 PDES 引擎的构建，该引擎可以在最短的预期执行时间内模拟任意应用程序。 进一步的好处可以从基于“前瞻”和“未来列表”的框架支持、应用程序相关的性能增强中获得。