SCREMS: A Parallel Computer Cluster For Multiphysics & Multiscale Modeling of Subsurface & Surface Flows

SCEMS：多物理场并行计算机集群

• 批准号: 0215389
• 负责人: Mary Wheeler
• 金额: $ 8.06万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2002
• 资助国家: 美国
• 起止时间: 2002-08-15 至 2004-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0215389&HistoricalAwards=false
• 关键词: SCREMS; Parallel; Computer; Cluster; Multiphysics

The Center for Subsurface Modeling (CSM), part of the TexasInstitute for Computational and Applied Mathematics at TheUniversity of Texas at Austin, will purchase 64 PCs, acontrol workstation, as well as a storage device, in orderto constuct a parallel computing platform capable of100 Gflops. The network for the cluster (high speed Myrinetnetwork (1Gbit) and a 100Mb ethernet network) is already inplace. This platform will be dedicated to research in themathematical sciences being led by CSM. Projects that willuse this platform include large-scale multiphysics modelingof subsurface processes, multiscale modeling of flow inextremely heterogeneous rocks, finite element coastal andocean modeling, parallel terascale iterative solvers, andDiscontinuous Galerkin discretization methods. A substantialcomponent of each of these projects is the development ofnew scalable parallel algorithms as well as implementationand testing of both proof-of-concept as well as ofproduction-quality codes for simulation of cases driven byenergy and environment applications. A dedicatedlimited-access resource greatly expedites such development.The Center for Subsurface Modeling, part of the TexasInstitute for Computational and Applied Mathematics at TheUniversity of Texas at Austin, will purchase equipment neededto construct a medium size parallel computer capable of100 billion arithmetic operations per second. This computerwill be dedicated to research in the mathematical andcomputational sciences being led by CSM. Researchersinvolved in the project have decades of experience indevelopment of robust, accurate and fast computationaltechniques. The new cluster will allow for large-scalecomputational modeling and testing of such techniques whichpromise to improve ability to understand, predict, monitor,and control natural and engineered processes occuring insurface and subsurface. Significant examples includeoptimization of oil and gas recovery, analysis ofcontamination and remediation scenarios in subsurface andsurface waters like aquifer and coastal environments. Theaccess to a dedicated computing resource willgreatly expedite such development.

德克萨斯大学奥斯汀分校德克萨斯计算与应用数学研究所下属的地下建模中心（CSM）将购买64台PC、一个控制工作站以及一个存储设备，以构建一个能够实现100 Gflops的并行计算平台。 集群网络（高速 Myrinet 网络 (1Gbit) 和 100Mb 以太网）已就位。 该平台将致力于由 CSM 领导的数学科学研究。 将使用该平台的项目包括地下过程的大规模多物理场建模、极不均匀岩石中流动的多尺度建模、有限元沿海和海洋建模、并行万亿级迭代求解器和不连续伽辽金离散化方法。这些项目的一个重要组成部分是开发新的可扩展并行算法，以及概念验证和生产质量代码的实施和测试，以模拟由能源和环境应用驱动的案例。 专用的有限访问资源极大地加快了这种开发。地下建模中心是德克萨斯大学奥斯汀分校德克萨斯计算和应用数学研究所的一部分，将购买建造一台每秒可进行 1000 亿次算术运算的中型并行计算机所需的设备。 这台计算机将致力于 CSM 领导的数学和计算科学的研究。 参与该项目的研究人员在开发稳健、准确和快速的计算技术方面拥有数十年的经验。新的集群将允许对此类技术进行大规模计算建模和测试，从而有望提高理解、预测、监测和控制地表和地下发生的自然和工程过程的能力。重要的例子包括优化石油和天然气回收、分析地下和地表水（如含水层和沿海环境）的污染和修复方案。获得专用计算资源将大大加快这种发展。