SI2-SSI: Collaborative Research: STORM: A Scalable Toolkit for an Open Community Supporting Near Realtime High Resolution Coastal Modeling

SI2-SSI：协作研究：STORM：支持近实时高分辨率海岸建模的开放社区的可扩展工具包

• 批准号: 1339782
• 负责人: Hartmut Kaiser
• 金额: $ 97.08万
• 依托单位: Louisiana State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-10-01 至 2019-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1339782&HistoricalAwards=false
• 关键词: SI2; SSI; Collaborative; Research; STORM

The ADCIRC coastal circulation and storm surge model has a long standing track record of extremely high societal impact. It has been used to define risk (e.g., 100-yr, 500-yr coastal flood levels) for the FEMA National Flood Insurance Program in coastal states from New York to Texas, it has been used to design the multi-billion dollar Hurricane and Storm Damage Risk Reduction System around greater New Orleans and southern Louisiana by the Army Corps of Engineers, it is currently run operationally by NOAA/NWS National Center for Environmental Prediction to forecast storm surge, to name just a few of its current and recent applications. Thus there is a well-established user network in place to convert improvements in ADCIRC into significant broader impacts. The proposed research provides transformative intellectual contributions that focus on applying new parallelization schemes to enable major advances in the algorithms, implementation and utilization of the ADCIRC model. The broadening of ADCIRC to a multi-algorithmic framework and the resulting performance gains that are anticipated will help ensure ADCIRC's sustainability as a core community model for at least the next 20 years. In addition, the proposed collaboration will impact computer science by serving as a high impact use case to inform the design of new approaches to efficient scalable computing. Together, the advancements in coastal modeling and parallelization technology will make a significant contribution to the science of modeling and HPC. The results of the proposed research will be disseminated to a wider community through ongoing educational outreach activities at the participating organizations as well as through refereed conference and journal papers, and invited presentations. The involvement of graduate students and post-doctoral fellows will be crucial towards the success of this project. The PIs have a long history of training and mentoring students and post-docs in computational science and engineering, coastal engineering and marine science. The recruitment and involvement of underrepresented groups in these efforts has always been a high priority. In addition, aspects of the proposed research will be incorporated into the curricula of several courses taught by the PIs in the areas of finite element methods, scientific computation, hydrology and oceanography.The aim of this project is to broaden the ADCIRC coastal circulation and storm surge model from a successful, but somewhat static coastal modeling tool that is tied to a single solution algorithm and the MPI parallelization paradigm, to a dynamic computational platform that is comprised of multiple solution algorithms, that readily admits new solution algorithms and that is built on a transformational new parallelization scheme that will allow us to scale to at least 256k compute cores on modern high performance computing (HPC) systems. We will do this by creating a living, evolving coastal modeling framework that will continue to lead the community in merging physical science / engineering and high performance computing and we will make the framework available to the broader community as a sustainable long term solution for its coastal modeling needs. In addition we will utilize these advancements in the highly demanding coastal storm surge forecasting system that we presently operate to demonstrate both improved robustness and speed of the model solution. We expect this effort will shorten the time required to provide reliable forecasting results and improve our ability to provide highly resolved, accurate, and physically complete predictions on an unprecedented scale. Concurrently, it should enable the use of smaller resources for simulations of increased scale which improves the usability and widens the applicability of ADCIRC in a broader community. The development of tightly integrated web-oriented products like CERA (www.coastalemergency.org) will enable the wide and timely dissemination of forecast modeling results to reach a broad audience.

ADCIRC 沿海环流和风暴潮模型有着极高社会影响的长期记录。它已被用于定义从纽约到德克萨斯州沿海各州的 FEMA 国家洪水保险计划的风险（例如，100 年、500 年沿海洪水水平），它已被陆军工程兵团用于设计大新奥尔良和路易斯安那州南部周围数十亿美元的飓风和风暴损害风险降低系统，目前由 NOAA/NWS 国家环境预测中心运行以预测风暴 激增，仅举其当前和最近的一些应用。因此，有一个完善的用户网络可以将 ADCIRC 的改进转化为更广泛的重大影响。拟议的研究提供了变革性的智力贡献，重点是应用新的并行化方案，以实现 ADCIRC 模型的算法、实现和利用方面的重大进步。 ADCIRC 扩展到多算法框架以及由此带来的预期性能提升将有助于确保 ADCIRC 至少在未来 20 年作为核心社区模型的可持续性。此外，拟议的合作将通过作为一个高影响力的用例来影响计算机科学，为高效可扩展计算的新方法的设计提供信息。总之，沿海建模和并行化技术的进步将为建模和高性能计算科学做出重大贡献。拟议研究的结果将通过参与组织正在进行的教育外展活动以及经过审阅的会议和期刊论文以及受邀演讲向更广泛的社区传播。研究生和博士后的参与对于该项目的成功至关重要。 PI 在计算科学与工程、海岸工程和海洋科学领域的学生和博士后培训和指导方面有着悠久的历史。招募和代表性不足的群体参与这些工作一直是高度优先事项。此外，拟议研究的各个方面将纳入 PI 在有限元方法、科学计算、水文学和海洋学领域教授的几门课程的课程中。该项目的目的是将 ADCIRC 沿海环流和风暴潮模型从与单一解决方案算法和 MPI 并行化范例相关的成功但有些静态的沿海建模工具扩展到动态计算 该平台由多种解决方案算法组成，可以轻松接纳新的解决方案算法，并且构建在变革性的新并行化方案之上，该方案将使我们能够在现代高性能计算 (HPC) 系统上扩展到至少 256k 计算核心。我们将通过创建一个活跃的、不断发展的沿海建模框架来实现这一目标，该框架将继续引领社区融合物理科学/工程和高性能计算，并且我们将向更广泛的社区提供该框架，作为满足其沿海建模需求的可持续长期解决方案。此外，我们将在目前运行的要求极高的沿海风暴潮预报系统中利用这些进步，以展示模型解决方案的稳健性和速度的提高。我们预计这项工作将缩短提供可靠预测结果所需的时间，并提高我们以前所未有的规模提供高分辨率、准确且物理上完整的预测的能力。同时，它应该能够使用较小的资源进行更大规模的模拟，从而提高可用性并扩大 ADCIRC 在更广泛社区中的适用性。 CERA (www.coastalemergency.org) 等紧密集成的面向网络的产品的开发将使预测建模结果能够广泛而及时地传播到广大受众。

项目成果

Methodology for Adaptive Active Message Coalescing in Task Based Runtime Systems

基于任务的运行时系统中自适应主动消息合并的方法

• DOI: 10.1109/ipdpsw.2018.00173
• 发表时间: 2018
• 期刊: 2018 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW
• 作者: Wagle, Bibek;Kellar, Samuel;Serio, Adrian;Kaiser, Hartmut
• 通讯作者: Kaiser, Hartmut