Beowulf Cluster for Computer-Intensive Research

描述（由申请人提供）： 计算机硬件和软件的进步极大地提高了研究人员可用的计算能力，使他们能够解决诸如解码人类基因组之类的问题，这些问题在十年前似乎是无法克服的。Beowulf集群是可用于并行计算的多计算机系统，由于其性能/价格比是传统超级计算机的3至10倍，因此在计算革命中发挥了重要作用。该提案的目标是获得一个Beowulf集群，以缓解限制马萨诸塞州大学医学院NIH资助研究人员研究的计算瓶颈。该集群将构成CORE设施的核心，供医学院所有NIH资助的研究人员以及寻求NIH提交初步结果的研究人员使用。本提案中记录了许多方法，其中拟议集群的计算能力的可用性将推进本提案中各个项目的科学。它们不同，从使研究人员能够使用新的数据收集技术和实验方法，这些技术和方法具有巨大的计算需求，到使用系统的紧密耦合节点来执行计算，这实际上无法在单个处理器上执行。集群的计算能力还将使调查人员能够减少他们在设计拟议调查时所作的近似和妥协，因为他们可用的计算资源有限。随着生物信息学、结构生物学、分子生物物理学和成像领域的新研究人员被招募到大学，对该集群的需求只会在未来增加。最后，CORE设施还将为来自不同学科的研究人员提供互动和共享计算算法和方法的机会。

DESCRIPTION (provided by applicant): Advances in computer hardware and software have dramatically improved the compute capabilities available to researchers allowing them to tackle problems such as decoding the human genome that seemed insurmountable even a decade ago. Beowulf clusters, which are multi-computer systems that can be used for parallel computations, have played a significant role in the compute revolution due to their performance / price ratio being 3 to 10 times that of traditional supercomputers. The goal of this proposal is to obtain a Beowulf cluster to relieve the computing bottleneck restricting the investigations of the NIH funded researchers at the University of Massachusetts Medical School. The cluster that will form the nucleus of a CORE facility available to all NIH funded investigators at the medical school, and investigators seeking preliminary results for NIH submissions. There are a number of ways documented in this proposal in which the availability of the computational power of the proposed cluster would advance the science of the individual projects of this proposal. They vary from enabling investigators to use new data collection technologies and experimental approaches which have immense computation requirements to using the tightly coupled nodes of the system to perform calculations, which could not realistically be performed on a single processor. The computing power of the cluster would also allow investigators to reduce the approximations and compromises they made in the design of their proposed investigations that were necessitated by the limited computational resources available to them. The need for the cluster will only increase in the future as new investigators in the areas of bio-informatics, structural biology, molecular biophysics, and imaging are recruited to the university. Finally, the CORE facility will also provide the opportunity for investigators from diverse disciplines to interact, and share computational algorithms and approaches.