I/UCRC Phase II: Center for Hybrid Multicore Productivity Research

I/UCRC 第二阶段：混合多核生产力研究中心

• 批准号: 1439664
• 负责人: Sheldon Brown
• 金额: $ 22.5万
• 依托单位: University of California-San Diego
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-09-15 至 2016-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1439664&HistoricalAwards=false
• 关键词: UCRC; Phase; II; Center; Hybrid

The University of Maryland, Baltimore County (UMBC), the University of California, San Diego (UCSD) and the University of Utah (U Utah) are pursuing a Phase II renewal of the multi-institution I/UCRC for Hybrid Multicore Productivity Research (CHMPR). CHMPR, in Phase II, plans to focus on problems of national priority such as Big Data Analytics for Health IT, Climate Change, Brain and Pluripotent Cell Imaging, Material Genomics and the Human Imagination. It will conduct research to optimize the scalability, performance and productivity of applications to Interactive Digital Media and Data Intensive Parallel Computing Paradigms which are all of direct relevance to industry advisory board members on hybrid multicore clusters. CHMPR will contribute towards development of efficient hybrid multicore computational science applications. It will improve the understanding of constrained parallel processing algorithms, and provide academic researchers access to large hybrid multicore systems to address the challenges in obtaining peak performance of applications on these architectures. CHMPR will further assist in addressing these challenges by providing fundamental understanding of the numerical formulation, structure of the applications and the ability to map known single threaded algorithms from multiple processors onto the underlying hybrid multicore architectures under limitations on local processor memory, and cache. The combined computing facilities at UMBC, UCSD and U Utah respectively, will offer industry advisory board members access to university testbeds consisting of some of the most advanced and diverse computing technologies and Big Data sets in Health and Climate. There is a large demand to explore the application and scalability of such hybrid multicore architectures.CHMPR will provide faculty and students the opportunity to gain hands-on research expertise in a wide variety of hybrid multicore applications in areas of Health, Climate Change, Biomedical Informatics, 3-D Graphic Environments and the use of Social Media for Disaster mitigation. By addressing the future needs of industry advisory board members as the computer industry evolves its hybrid multicore processor technology, the center will contribute towards US leadership in HPC. CHMPR sites will incorporate hybrid multicore computing into their core course curricula for students to address the industry applications. The students will also have the opportunity for internships with industry and the center will educate students how to take advantage of the parallelism offered by tens of thousands of multicore processors to solve a variety of applications.

马里兰州巴尔的摩县大学（UMBC）、加州圣地亚哥大学（UCSD）和犹他州大学（U犹他州）正在寻求多机构I/UCRC混合多核生产力研究（CHMPR）的第二阶段更新。CHMPR在第二阶段计划专注于国家优先事项的问题，如健康IT的大数据分析，气候变化，大脑和多能细胞成像，材料基因组学和人类想象力。它将进行研究，以优化应用程序的可扩展性，性能和生产力，以交互式数字媒体和数据密集型并行计算范式，这些都与混合多核集群的行业顾问委员会成员直接相关。CHMPR将为开发高效的混合多核计算科学应用做出贡献。它将提高对受限并行处理算法的理解，并为学术研究人员提供访问大型混合多核系统的机会，以解决在这些架构上获得应用程序峰值性能的挑战。CHMPR将通过提供对数值公式的基本理解、应用程序的结构以及在本地处理器内存和缓存限制下将已知单线程算法从多个处理器映射到底层混合多核架构的能力，进一步帮助解决这些挑战。UMBC，UCSD和U犹他州的联合计算设施将分别为行业咨询委员会成员提供大学测试平台，包括一些最先进和最多样化的计算技术和健康和气候大数据集。有一个很大的需求，以探索这种混合多核架构的应用和可扩展性。CHMPR将为教师和学生提供机会，以获得在健康，气候变化，生物医学信息学，3-D图形环境和使用社会媒体减灾领域的各种混合多核应用的实践研究专业知识。随着计算机行业发展其混合多核处理器技术，该中心将满足行业顾问委员会成员的未来需求，为美国在HPC领域的领导地位做出贡献。CHMPR站点将把混合多核计算纳入其核心课程，供学生解决行业应用问题。学生还将有机会在工业界实习，该中心将教育学生如何利用数万个多核处理器提供的并行性来解决各种应用。