High End Computing Resource for Large Memory Data-intensive Biomedical Applicatio

描述（由申请人提供）：提议采购高端计算机（HEC）资源，以满足明尼苏达大学33个研究小组的广泛需求，这些研究小组由91个目前资助的NIH资助，包括47个R01资助。这一建议是由日益增长的需要来解决高影响的问题，包括从高性能计算和高吞吐量技术中获取、分析和可视化千兆级数据。这些巨大的内存和磁盘/IO密集型生物医学应用需要一个专门设计的整体硬件解决方案。已经确定了四个主要的用户群体对这种使能技术有迫切的需求：1)多尺度建模，2)化学动力学，3)生物信息学和计算生物学，以及4)生物医学成像。HEC解决方案基于1,152核/2.3TB SMP SGI紫外（UV）服务器和Virtu VN200可视化节点，具有无限存储4600个磁盘阵列，具有32 TB光纤通道，512 TB SATA容量和CXFS文件系统。这种集成的HEC解决方案是一种独特的、最先进的系统，具有非常大的共享内存和扩展的、超快的数据存储和传输能力，可以满足这四个主要用户群体的特定需求，并将填补明尼苏达大学在明尼苏达超级计算研究所（MSI）集中的计算支持基础设施的关键空缺。拟议的系统与MSI目前的核心硬件有很大的不同，并且将极大地扩展nih支持的生物医学应用的范围，这些应用可以由MSI服务。MSI负责系统的安装、供电和管理，并提供培训和软件支持，以及开发自定义排队系统，以简化主要用户的吞吐量。提出的HEC解决方案将通过直接访问和新颖的网格网络服务，为明尼苏达大学nih支持的研究人员提供按需访问。由明尼苏达大学和梅奥诊所开发的网格服务以及标准网格服务将部署在HEC上，并将支持NIH研究人员使用核心实验室服务的数据密集型和内存密集型需求。结果将是，需要大量内存/磁盘的计算的吞吐量将大大提高，用于传输、操作、分析和可视化非常大的数据集的人力时间将大大减少，并且由于资源限制而受到阻碍的新应用程序将被启用。拟议的HEC解决方案预计将创造大量新的就业机会，促进生物医学研究的突破，并对人类健康产生重大影响。

DESCRIPTION (provided by applicant): The procurement of a high end computer (HEC) resource is proposed to meet demands of a broad range of 33 research groups at the University of Minnesota that are supported by 91 currently funded NIH grants, including 47 R01 grants. This proposal is driven by the growing need to tackle high- impact problems that involve the acquisition, analysis and visualization of petascale data from high performance computing and high-throughput technologies. These tremendously memory and disk/IO intensive biomedical applications require a specially designed total hardware solution. Four major user groups have been identified that have urgent demands for this enabling technology: 1) multi- scale modeling, 2) chemical dynamics, 3) bioinformatics and computational biology, and 4) biomedical imaging. The HEC solution is based on a 1,152-core/2.3TB SMP SGI Ultraviolet (UV) server and Virtu VN200 visualization nodes with Infinite Storage 4600 disk arrays having 32 TB fibre-channel, 512 TB SATA capacity and CXFS file system. This integrated total HEC solution is a unique, state of the art system that has very large shared memory and expansive, ultrafast data storage and transfer capabilities to meet the specific needs of these four major user groups, and will fill a critical niche in the University of Minnesota's computational support infrastructure centralized at the Minnesota Supercomputing Institute (MSI). The proposed system is significantly different from the MSI's current core hardware, and will greatly extend the range of NIH-supported biomedical applications that can be served by the MSI. The MSI is committed to house, power and administer the system, and will provide training and software support, as well as develop a custom queuing system to streamline throughput for major users. The proposed HEC solution will provide on-demand access to a broad community of NIH-supported researchers at the University of Minnesota through both direct access and novel caGrid web services. The caGrid services developed at the University of Minnesota and the Mayo Clinic as well as standard caGrid services will be deployed on the HEC, and will support the data-intensive and memory-intensive needs of NIH researchers using core laboratory services. The result will be that throughput of calculations with intensive memory/disk requirements will be greatly enhanced, human time involved in the transfer, manipulation, analysis and visualization of very large data sets will be significantly reduced, and new applications that were impeded by resource limitations will be enabled. The proposed HEC solution is expected to create a wealth of new job opportunities, facilitate breakthroughs in biomedical research and have significant impact on human health.