A System Architecture Point Design Study for Exascale Computing

NSF和其他联邦机构将需要在制定未来的使能技术、体系结构和方法学计划之前，进一步了解Exascale计算系统的概念、方法和手段。这样的理解将驱动未知被确定，挑战被处理，策略被引导，任务被执行，在通往Exascale系统实现的道路上。到2018年实现能够持续进行Exaflops计算的系统的目标表明，赞助的研发计划致力于这一目标，不迟于2011年启动，以提供足够的时间来完成必要的研究、设计和实施任务。NSF百亿亿次计算的点设计研究是三个任务的集合，将由这些领域的专家研究三种不同的计算制度。虽然每一方都是独立的，但他们将通过合作和分享成果而受益匪浅，每一方都以来自其他各方的见解为指导。它们是1)应用程序和编程模型，2)系统架构和操作系统，以及3)硬件技术和组件设计。该SGER将研究百亿亿次计算点设计研究的系统架构和操作系统方面。

NSF and other Federal agencies will require further understanding of concepts, methods, and means for Exascale computing systems before formulation of future programs in enabling technologies, architectures, and methodologies. Such understanding will drive the unknowns to be determined, the challenges to be addressed, strategies to guide, and tasks to be performed on the way to Exascale system realization. The objective of realizing systems capable of sustained Exaflops computing by 2018 suggests that sponsored R&D programs dedicated to this goal be started no later than 2011 to provide sufficient time to complete the ensemble of necessary research, design, and implementation tasks. The NSF Point Design Study for Exascale Computing is a collection of three tasks that will look at three different regimes of computing by experts in those areas. While each is stand alone, they will benefit strongly through cooperation and sharing of results, each being guided by the insights derived from the others. These are 1) applications and programming models, 2) system architecture and operating systems, and 3) hardware technology and component design. This SGER will study system architecture and operating systems aspect of the Point Design Study for Exascale Computing.