Tightly-coupled Heterogeneous Supercomputing

本项目将研究高性能计算(HPC)领域内紧耦合的异质超级计算的性能极限。拟议的项目将开发准确的模型来预测不同类型系统的性能。发展准确的模型来预测异质系统的性能，将为这一领域提供科学的严谨性，并将对这些系统的潜在用户有很大的好处。这项拟议的工作涉及计算机科学和应用领域的多学科，并将由现实世界中的高性能应用程序推动，这些应用程序预计将在这些系统上运行，来自生物和医学领域。这项研究的正式传播将通过在主要会议和期刊上的介绍和发表以及学术界和工业界之间的互动来进行，从而加强这项工作对新兴商业产品的潜在影响。

This project will study the limits to the performance of tightly-coupled heterogeneous supercomputing within the field of high-performance computing (HPC). The proposed project will develop accurate models to predict the performance of heterogeneous systems. The development of accurate models to predict the performance of heterogeneous systems will impose a scientific rigor on this field and will be of greatbenefit to potential users of these systems. The proposed work is multidisciplinary across the areas of computer sciences and applications, and will be driven by real world high performance applications expected to run on such systems, drawn from the fields of biology and medicine. Formal dissemination of this research will take place through presentations and publication in the major conferences and journals, and through interactions between academia and industry thus enhancing the potential impact of this work to emerging commercial products.