Category I: Bridges-2: Scalable Converged Computing, Data, and Analytics for Rapidly Evolving Science and Engineering Research

第一类：Bridges-2：用于快速发展的科学和工程研究的可扩展融合计算、数据和分析

• 批准号: 1928147
• 负责人: Sergiu Sanielevici
• 金额: $ 1000万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Cooperative Agreement
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-10-01 至 2025-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1928147&HistoricalAwards=false
• 关键词: Category; Bridges; Scalable; Converged; Computing

Unlocking the power of data will accelerate discovery to advance science, improve our quality of life, and enhance national competitiveness. The increasing importance of data is driving a rapid evolution of research, creating opportunities for breakthroughs by combining the strengths of artificial intelligence (AI) and high-performance computing (HPC) simulation. To address this exciting opportunity for the nation's research community, the Pittsburgh Supercomputing Center (PSC), a joint research center of Carnegie Mellon University and the University of Pittsburgh, in partnership with Hewlett Packard Enterprise (HPE), will deploy Bridges-2, a resource that will provide exceptional capacity and transformative capability for rapidly evolving, data- and computation-intensive research. Building on PSC's experience with its very successful Bridges system, Bridges-2 will take the next step in pioneering converged, scalable HPC, AI, and data; prioritize researcher productivity and ease of use; and provide an extensible architecture for interoperation with complementary data-intensive projects, campus resources, and clouds. Bridges-2 will deploy new high-performance technologies to deliver results faster by combining AI with simulation and through exceptional data-handling capabilities and will be integrated with NSF-funded cyberinfrastructure (CI) services for allocation, user-support, monitoring, etc., such as those currently provided by the Extreme Science and Engineering Discovery Environment (XSEDE) and XDMoD.Bridges-2 will accelerate discovery to benefit science, society, and the nation. Its unique architecture will catalyze breakthroughs in critically important areas such as understanding the brain, developing new materials for sustainable energy production and quantum computing, assembling genomes of crop species to improve agricultural efficiency, exploring the universe via multimessenger astrophysics, and enabling technologies for smart cities. A great amount of valuable research is in areas that did not traditionally use HPC until Bridges. Bridges-2 will therefore serve nontraditional communities and applications, which now have workloads requiring great capacity, as well as traditional ones. Its community data collections and user-friendly interfaces will democratize participation in science and engineering and foster collaboration and convergence research. The Bridges-2 project includes areas of focus on developing STEM talent, engaging industry, reaching beyond borders, and broadly engaging stakeholders to develop the nation's workforce, maximize learning through collaboration, and bring the benefits of advanced computing to the general public.PSC is integrating cutting-edge technologies to deploy a uniquely powerful national resource for turning data into knowledge. Bridges-2 will consist of Hewlett Packard Enterprise Apollo 2000 and ProLiant DL560 central processing unit (CPU) servers of three different memory capacities (256GB, 512GB, and 4TB) for extremely broad workloads; Apollo 6500 servers each with two CPUs and eight graphics processing units (GPUs) for AI (particularly deep learning) and to accelerate simulations; and ProLiant DL360/380 servers for gateways, databases, and utility and management functions. A hierarchical data management system using HPE's Data Management Framework (DMF) will provide seamless access with exceptional ease of use to active and archival data, spanning a Lustre parallel filesystem and StoreEver MSL6480 Tape Library. An HPE TierZero high-performance flash array will provide even higher performance for data-intensive applications. A Mellanox HDR InfiniBand interconnect will provide unified, high-performance data transport across compute and storage components. The Bridges-2 user environment will be extremely robust and flexible, supporting HPC applications, AI frameworks, high-productivity programming languages, containers, interactive and batch execution, workflows, databases, and science gateways. Bridges-2's uniquely flexible software environment and heterogeneous hardware resources will serve the needs of rapidly evolving research, opening the door to discovery and the path to prosperity.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

释放数据的力量将加速发现，推动科学发展，提高我们的生活质量，增强国家竞争力。数据的重要性日益增加，推动了研究的快速发展，通过结合人工智能（AI）和高性能计算（HPC）模拟的优势，为突破创造了机会。为了解决这个令人兴奋的国家研究界的机会，匹兹堡超级计算中心（PSC），卡内基梅隆大学和匹兹堡大学的联合研究中心，与惠普企业（HPE）合作，将部署桥梁-2，一个资源，将提供卓越的能力和变革能力快速发展，数据和计算密集型的研究。基于PSC非常成功的Bridges系统的经验，Bridges-2将在融合、可扩展的高性能计算、人工智能和数据方面迈出下一步；优先考虑研究人员的生产力和易用性；并提供一个可扩展的体系结构，用于与互补的数据密集型项目、校园资源和云进行互操作。Bridges-2将部署新的高性能技术，通过将人工智能与仿真结合起来，通过卓越的数据处理能力，更快地交付结果，并将与nsf资助的网络基础设施（CI）服务集成，用于分配、用户支持、监控等，例如目前由极端科学与工程发现环境（XSEDE）和XDMoD提供的服务。桥梁2号将加速发现，造福科学、社会和国家。其独特的架构将催化在至关重要的领域取得突破，如理解大脑，开发可持续能源生产和量子计算的新材料，组装作物物种基因组以提高农业效率，通过多信使天体物理学探索宇宙，以及智能城市的使能技术。大量有价值的研究是在传统上没有使用高性能计算的领域，直到桥梁。因此，Bridges-2将为非传统社区和应用程序提供服务，这些社区和应用程序现在的工作负载需要很大的容量，与传统社区和应用程序一样。它的社区数据收集和用户友好界面将使科学和工程的参与民主化，并促进合作和融合研究。Bridges-2项目的重点领域包括培养STEM人才、吸引行业参与、跨越国界、广泛吸引利益相关者发展国家劳动力、最大限度地通过合作学习，并将先进计算的好处带给公众。PSC正在整合尖端技术，部署独特强大的国家资源，将数据转化为知识。Bridges-2将由Hewlett Packard Enterprise Apollo 2000和ProLiant DL560中央处理器服务器组成，具有三种不同的内存容量（256GB、512GB和4TB），可用于非常广泛的工作负载；阿波罗6500服务器每台都有两个cpu和八个图形处理单元（gpu），用于人工智能（特别是深度学习）和加速模拟；以及ProLiant DL360/380服务器，用于网关、数据库、实用程序和管理功能。使用HPE数据管理框架（DMF）的分层数据管理系统将提供对活动数据和存档数据的无缝访问，并且非常易于使用，跨越Lustre并行文件系统和StoreEver MSL6480磁带库。HPE TierZero高性能闪存阵列将为数据密集型应用提供更高的性能。Mellanox HDR InfiniBand互连将在计算和存储组件之间提供统一的高性能数据传输。Bridges-2用户环境将非常健壮和灵活，支持高性能计算应用程序、人工智能框架、高生产率编程语言、容器、交互式和批处理执行、工作流、数据库和科学网关。Bridges-2独特的灵活软件环境和异构硬件资源将满足快速发展的研究需求，为发现和繁荣之路打开大门。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Rebuilding Bridges: The tools used to deploy and maintain Bridges-2

重建桥梁：用于部署和维护 Bridges-2 的工具

• DOI: 10.1145/3569951.3597589
• 发表时间: 2023
• 期刊: PEARC '23: Practice and Experience in Advanced Research Computing
• 作者: Sablosky, Geoffrey Ryan

AlphaFold2 Workflow Optimization for High Throughput Predictions in HPC Environment

用于 HPC 环境中高吞吐量预测的 AlphaFold2 工作流程优化

• DOI: 10.1145/3491418.3535181
• 发表时间: 2022
• 期刊: PEARC '22: Practice and Experience in Advanced Research Computing
• 作者: Posada, Edwin Fernando;McGee, Francisco;Berger, Richard;Dorrell, Mitchell;Lamanna, Jason;Sanielevici, Sergiu;Carnevale, Vincenzo
• 通讯作者: Carnevale, Vincenzo

Visualizing the Future of Louisiana’s Coastline

路易斯安那州海岸线的未来可视化

• DOI: 10.1145/3569951.3597548
• 发表时间: 2023
• 期刊: PEARC '23: Practice and Experience in Advanced Research Computing
• 作者: Yoder, Matt;Puerto, Juan
• 通讯作者: Puerto, Juan

COSMO: a Research Data Service Platform and Experiences from the BlueTides Project

COSMO：研究数据服务平台和BlueTides项目的经验

• DOI: 10.1145/3491418.3535166
• 发表时间: 2022
• 期刊: PEARC '22: Practice and Experience in Advanced Research Computing
• 作者: Uran, Julian;Huang, Wanting;Puerto, Juan;Chen, Nianyi;Wang, Mei-Yu;Ni, Yueying;DiMatteo, Tiziana;Buitrago, Paola
• 通讯作者: Buitrago, Paola

Deep Learning Benchmark Studies on an Advanced AI Engineering Testbed from the Open Compass Project

开放罗盘项目的高级人工智能工程测试平台的深度学习基准研究

• DOI: 10.1145/3569951.3597596
• 发表时间: 2023
• 期刊: PEARC '23: Practice and Experience in Advanced Research Computing
• 作者: Wang, Mei-Yu;Uran, Julian;Buitrago, Paola
• 通讯作者: Buitrago, Paola