Collaborative Proposal: The IESP: Planning to create next generation software infrastructure for exascale science

这是一项关于启动亿级计算软件的协调国际计划的建议。这将采取一系列三次研讨会的形式，以吸引适当的国家和国际合作伙伴，制定路线图和制定积极的计划，以确保软件得到规划、预算，并有望及时开发，以迎接艾级计算的到来。该项目得到了能源部以及欧盟和日本的国际合作伙伴的大力支持。目前正在请求国家科学基金会的支持，以帮助招募国家科学基金会社区作为该计划的创始合作伙伴。该项目是在我们仍在处理Petascale的时候启动的，非常及时，因为它针对的需求的重要性和紧迫性奠定了基础，并启动了用于非常高性能计算的未来几代软件的实施。软件鸿沟--即在计算机问世数年后才能在最新体系结构上使用的高效软件--一直困扰着从千兆级到千万亿级的每一代新的高性能计算机。为艾斯卡设想的截然不同的计算机体系结构可能会使这一差距变得更加戏剧性，并可能使许多艾斯卡投资在没有这种软件协调和预先投资的情况下用处很小。

This is a proposal to launch a coordinated international program on software for exascale computing. This will take the form of a series of three workshops to engage the appropriate national and international partners, develop the roadmap and create an aggressive plan for making sure the software is planned, budgeted and hopefully developed in time for the arrival of exascale computing. The project has deep support from the Department of Energy and international partners in the European Union and Japan. The support of NSF is being requested to help recruit the NSF community as founding partners in the program. The project, being started at a time when we are still coming to grips with Petascale, is very timely in "the importance and urgency of the need it targets laying the foundations and initiating the implementation of future generations of software for very high performance computing." The software gap -- whereby efficient software to use on the latest architecture is only available years after the computer is available has bedevilled every generation of new high performance computers from the gigascale to the petascale. The dramatically different computer architectures that are envisioned for the exascale is likely to make this gap even more dramatic and perhaps render much of the exascale investments of little use without this type of coordianted and advance investment in the software.