Towards Generic Scalability of the Unifed Model

实现统一模型的通用可扩展性

• 批准号: EP/F011571/1
• 负责人: Lois Steenman-Clark
• 金额: $ 15.38万
• 依托单位: University of Reading
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2007
• 资助国家: 英国
• 起止时间: 2007 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FF011571%2F1
• 关键词: Towards; Generic; Scalability; Unifed; Model

SummaryThe Met Office's Unified Model (UM) is a computer model of the processes that determine our weather and the Earth's climate. The UM can be set up to simulate weather on a small, regional scale, or a global scale and the level-of-detail (resolution of the model) can be varied to capture more or less information about the physical processes involved.The model can take a very long time to run, even on very large supercomputers. The Met Office themselves buy a new supercomputer, the fastest they can afford, every few years and this lets them keep up with the increasing complexity of the model the scientists wish to explore.The UM is also used by a large number of academic researchers and scientists in Universities. Reading University are the key providers of the UM to the academic community and they work closely with the Met Office to make the model available. Academics typically run the UM on the supercomputers provided by the government as part of a National Service provided to all scientists in the U.K. Much of the time on these computers is used by the many users of the UM, and anything that can be done to make better use of the time on these computers is of benefit to the community as a whole. For example, if the UM can be made to run faster on a supercomputer, more science runs could be done in the time available or the computer could be used to run some code from another scientist.The scientists in the U.K also have several other large computers on which they wish to run the model. These computers are often owned by the University the scientist works at. There are also scientists in other countries that use the UM on their supercomputers. For instance, the UM is run on the Earth Simulator in Japan.Since the UM is used by so many people on so many computers, there are some quite well known problems with the way the code is written that stop it running as fast as it could. For each of these problems, this project aims to change the code so that it will run faster. We hope to make changes to the code that can be tailored to each of the machines on which the UM is run in as simple a way as possible so that the benefits apply to as many scientists as possible. The project will help the Met Office too because they will have a wider choice of machines that they can consider buying in the future.

英国气象局的统一模型（UM）是一个决定天气和地球气候过程的计算机模型。可以设置UM以模拟小范围、区域尺度或全球尺度的天气，并且可以改变细节级别（模型的分辨率），以捕获有关所涉及的物理过程的或多或少的信息。该模型可能需要很长时间才能运行，即使是在非常大的超级计算机上。英国气象局自己每隔几年就会购买一台新的超级计算机，这是他们能负担得起的最快的超级计算机，这使他们能够跟上科学家们希望探索的日益复杂的模型。UM也被许多大学的学术研究人员和科学家使用。雷丁大学是学术社区的主要提供者，他们与英国气象局密切合作，使该模型可用。学者们通常在超级计算机上运行UM，这些超级计算机是政府提供给英国所有科学家的国家服务的一部分，这些计算机上的大部分时间被UM的许多用户使用，任何可以更好地利用这些计算机上的时间的事情都有利于整个社区。例如，如果可以使UM在超级计算机上运行得更快，则可以在可用的时间内完成更多的科学运行，或者计算机可以用来运行其他科学家的代码。英国的科学家们还有其他几台大型计算机，他们希望在这些计算机上运行这个模型。这些计算机通常为科学家工作的大学所有。其他国家的科学家也在他们的超级计算机上使用UM。例如，UM是在日本的地球模拟器上运行的。由于UM被如此多的人在如此多的计算机上使用，因此编写代码的方式存在一些众所周知的问题，这些问题会阻止它尽可能快地运行。对于这些问题，本项目的目标是修改代码，使其运行得更快。我们希望对代码进行修改，使其能够以尽可能简单的方式为运行UM的每台机器量身定制，从而使尽可能多的科学家受益。该项目也将帮助英国气象局，因为他们将有更多的机器选择，可以考虑在未来购买。