HIGHLY PARALLEL COMPUTER SYSTEM

高度并行的计算机体系结构是实现以下目标的唯一手段 高级生物医学计算问题所要求的计算速度。 高度并行的生物医学计算项目的目标是 确定哪种并行体系结构最适合解决这类问题 在生物医学计算中出现，以开发并行算法 先进的生物医学计算问题，并提供高性能 并行计算机设施，将使NIH的工作人员在他们的 科学计算需求。CSL正在调查以下重大事件 计算要求高的环境下的并行计算问题 生物医学应用：将问题分成多个部分 在不同的处理器上独立执行，这样设计算法 处理器间通信的延迟可以保持在很小的比例内 的计算时间，设计部件以使计算负载可以 均匀分布在可用的处理器上，并设计 算法，因此处理器的数量是一个参数，而 算法可以动态配置。与其合作者一起，CSL正在 开发和实现下列并行算法 生物医学应用领域：电子显微图像处理 结构生物学、蛋白质和核酸序列分析、核 磁共振波谱、X射线结晶学、蛋白质折叠 预测、分子动力学模拟、量子化学方法和 医学影像。 在过去的一年里，CSL安装了128处理器的DARPA Touchstone Gamma 具有10G并发I/O文件系统的原型并行计算机 它是由英特尔超级计算机系统部门建造的。平行 算法被实现来执行计算密集型步骤 在大型二十面体病毒的三维重建中 电子显微照片和计算溶剂可达表面积 一种蛋白质，用来确定这些分子的构象 大分子。在来年，应用程序的工作将继续进行 上面列出的领域。

Highly parallel computer architectures provide the only means to attain the computational rates demanded by advanced biomedical computing problems. The goals of the Highly Parallel Biomedical Computing Project are to determine which parallel architectures are best for the classes of problems that arise in biomedical computing, to develop parallel algorithms for advanced biomedical computing problems, and to provide a high-performance parallel computer facility that will benefit the NIH staff in their scientific computing needs. CSL is investigating the following significant parallel computing issues in the context of computationally demanding biomedical applications: partitioning a problem into many parts that can be independently executed on different processors, designing algorithms so that delays of interprocessor communication can be kept to a small fraction of the computation time, designing the parts so that the computing load can be distributed evenly over the available processors, and designing algorithms so that the number of processors is a parameter and the algorithms can be configured dynamically. With its collaborators, CSL is developing and implementing parallel algorithms for the following biomedical application areas: Image Processing of Electron Micrographs in Structural Biology, Protein and Nucleic Acid Sequence Analysis, Nuclear Magnetic Resonance Spectroscopy, X-ray Crystallography, Protein Folding Prediction, Molecular Dynamics Simulations, Quantum Chemical Methods, and Medical Imaging. During the past year, CSL installed a 128 processor DARPA Touchstone Gamma Prototype parallel computer with a ten Gigabyte Concurrent I/O File System that was constructed by the Intel Supercomputer systems Division. Parallel algorithms were implemented to perform the computationally-intensive steps in the three-dimensional reconstruction of large icosahedral viruses from electron micrographs and to calculate the solvent accessible surface area of a protein that is used to determine the conformation of these macromolecules. In the coming year, work will continue in the application areas listed above.