MRI: Development of a High Density, High Performance Beowulf Cluster

德克萨斯A M大学计划设计和建造一个256处理器的中粒度并行机，尽可能接近每美元的最大浮点性能。& 这个系统，通常被称为贝奥武夫集群，将需要修改通常的范例，以解决某些设计问题，在这些系统中固有的。该集群将包含256个处理器，由128个节点组成，使用双AMD主板，每个节点都有1GB ECC DDR DRAM，并将与千兆以太网互连，形成最大和最强大的学术集群之一。 将探索安装和冷却其他板的新技术，以最大限度地提高节点的密度。 还将提供多种操作环境，允许最终用户确定哪种环境可为其应用程序提供最佳性能。这台机器将用于解决数学和物理科学中的各种计算密集型问题，并支持主要位于化学，数学，物理和统计学的研究人员。 一个小组，以数值分析为核心，但重点是流体流动多孔介质，还将包括石油开采和地下水流（来自石油和农业工程）以及并行架构（来自计算机科学）的专业知识研究人员。 除了大规模计算问题之外，这些小组还将解决反问题、非参数函数估计、计算生物学的几何方法、量子动力学、高能物理学、无机和有机金属系统以及多原子分子等方面的问题。这一努力还将在设计和实施阶段提供教育效益。 机械工程将在高级设计课程中使用活动系统组件的新电源和冷却要求。一旦该系统到位，数学系、统计系和计算机科学系将提供有效利用和编程课程，以使指定的研究小组以及大学社区的其他人获得最大利益。

Texas A&M University plans to design and build a 256 processor,medium grain parallel machine with as close to maximum floatingpoint performance per dollar as possible. This system, commonlyreferred to as a Beowulf cluster, will require a modification ofthe usual paradigm in order to solve certain design problemsinherent in these systems. The cluster will contain 256 processorsconsisting of 128 nodes using dual AMD motherboards, each with1GB ECC DDR DRAM and will be interconnected with Gigabit Ethernetto form one of the largest and most powerful academic clusters.Components will be chosen to minimize cost while maximizingperformance. Novel techniques for mounting and cooling themotherboards will be explored in an effort to maximize thedensity of the nodes. Multiple operating environments willalso be made available, allowing the end user to determinewhich environment yields the peak performance for theirapplication. This machine will be used to solve a variety ofcomputationally intensive problems in the mathematical andphysical sciences and support researchers primarily locatedin chemistry, mathematics, physics and statistics. One group,with a core in numerical analysis but with emphasis on fluidflowing porous media, will also include researchers withexpertise in oil recovery and in groundwater flow (fromPetroleum and from Agriculture Engineering) and in parallelarchitectures (from Computer Science). In addition to problems inlarge-scale computation, the groups will be solving problems ininverse problems, nonparametric function estimation, statisticalmethods for computational biology, quantum dynamics, high energyphysics, inorganic and organometallic systems, and polyatomicmolecules.This effort will also provide an educational benefit, both withinthe design and the implementation stages. Mechanical Engineeringwill use the novel power and cooling requirements of the system assource of activity in a senior design class. Once the system is inplace, the departments of Mathematics, Statistics and ComputerScience will provide courses in efficient utilization andprogramming to allow maximum benefit to the designated researchgroups, as well as the rest of the university community.