PETASCALE MOLECULAR DYNAMICS DATA PROCESSING

这个子项目是许多研究子项目中利用 资源由NIH/NCRR资助的中心拨款提供。子项目和 调查员(PI)可能从NIH的另一个来源获得了主要资金， 并因此可以在其他清晰的条目中表示。列出的机构是 该中心不一定是调查人员的机构。 与生物医学相关的细胞尺度过程发生在由以下物质组成的分子组件中 几百万到几亿个原子。原子分子动力学模拟 提供了对它们的功能机制的洞察；这样的模拟 都是极其苛刻的，并且需要千万亿级的计算资源。这个 产生的数据集如此之大，以至于现有的工作站在 物理内存、处理和可视化能力以及海量存储，以高效地 处理好他们。常见的可视化计算，例如三维 长轨迹上的静电势场地图，使用 一个标准的台式工作站。因此，研究人员的互动分析 大的模拟结果被中断和减慢，以至于虚拟 不可能；Petascale分子动力学数据处理系统(MDDPS) 旨在克服台式计算机工作站和 提供必要的硬件功能，以实现实际的交互分析 以及具有挑战性的PB级数据集的可视化。Petascale MDDPS是一款 紧密耦合的计算机群集，作为一个聚合单元运行，以提供 Petascale MD模拟所需的高性能数据分析能力。 该系统由协同存储、分析和可视化节点组成 通过单独的高速网络连接内部和外部资源。 虽然资源的两个软件包NAMD和VMD都进行了很好的调整 到当前的计算机硬件，利用 Petascale MDDPS硬件需要额外的软件开发。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Biomedically-relevant cell-scale processes take place in molecular assemblies made of millions to hundreds of millions of atoms. Atomistic molecular dynamics simulation of these structures provides insight into their functional mechanisms; such simulations are extremely demanding and require petascale computational resources. The datasets that result are so large that existing workstations are too limited  in physical memory, processing and visualization power, and mass storage  to efficiently handle them. Common visualization calculations, such as three-dimensional maps of electrostatic potential fields across a long trajectory, can take minutes using a standard desktop workstation. As a result, a researcher's interactive analysis of a large simulation result is interrupted and slowed so much as to be virtually impossible; the Petascale Molecular Dynamics Data Processing System (MDDPS) is designed to overcome the shortcomings of desktop computer workstations and provides the necessary hardware features to enable practical interactive analysis and visualization of challenging petascale datasets. The Petascale MDDPS is a cluster of tightly coupled computers that operate as a cohesive unit to provide high-performance data analysis capabilities required by petascale MD simulations. The system is composed of synergistic storage, analysis, and visualization nodes connected internally and to external resources by separate high-speed networks. While both of the Resource's software packages, NAMD and VMD, are well-tuned to current computer hardware, harnessing the next-generation capabilities of the Petascale MDDPS hardware requires additional software development.