II-EN: Virgo 3 -- GPU Upgrade to the UTEP Distributed Computing Lab

II-EN: Virgo 3 -- GPU 升级至 UTEP 分布式计算实验室

• 批准号: 1059430
• 负责人: Patricia Nava
• 金额: $ 13.5万
• 依托单位: University of Texas at El Paso
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2011
• 资助国家: 美国
• 起止时间: 2011-07-15 至 2014-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1059430&HistoricalAwards=false
• 关键词: II; EN; Virgo; GPU; Upgrade

This project focuses on the upgrade of an existing computing cluster, Virgo-2, with Graphics Processing Unit (GPU) capability. The upgraded cluster, Virgo-3, will enhance the computational infrastructure of the Distributed Computing Laboratory (DCL) at the University of Texas at El Paso (UTEP), as well as increasing access to HPC resources by UTEP researchers. Virgo-3 will be utilized by researchers to exploit the potential parallelism that exists in their computational workloads. The GPUs in Virgo-3 will significantly increase the research productivity of UTEP researchers, which will lead to more rapid discoveries in areas such as cardio-visualization and communication networks. A finite element model of the heart, as well as the human torso, can be analyzed more efficiently using GPUs as a result of the inherent data-level parallelism. This more efficient finite element analysis will accelerate the work of UTEP researchers to advance the field of electrocardiography. GPUs can also be used to provide cost effective speedup for discrete event simulation to improve the productivity of communication networks research, where forty percent of computation time for discrete event simulation is consumed by updates to the future event list. This research will explore the utilization of GPUs to speedup future event list manipulation. Additionally, Virgo-3 will be available to other researchers, so that they can utilize the benefits of distributed computing in their research activities.Finally, this project will have broad-based impact through summer computer camps that are targeted to underrepresented minority pre-college students in the El Paso, Texas region.

该项目的重点是升级现有的计算集群Virgo-2，使其具有图形处理单元（GPU）能力。升级后的集群Virgo-3将增强德克萨斯大学埃尔帕索分校（UTEP）分布式计算实验室（DCL）的计算基础设施，并增加UTEP研究人员对HPC资源的访问。研究人员将利用Virgo-3来利用其计算工作负载中存在的潜在并行性。Virgo-3中的GPU将显著提高UTEP研究人员的研究生产力，这将导致在心脏可视化和通信网络等领域的更快发现。由于固有的数据级并行性，可以使用GPU更有效地分析心脏的有限元模型以及人体躯干。这种更有效的有限元分析将加速UTEP研究人员的工作，以推动心电图领域的发展。GPU还可以用于为离散事件仿真提供经济高效的加速，以提高通信网络研究的生产力，其中离散事件仿真的40%的计算时间用于更新未来事件列表。本研究将探索利用GPU来加速未来的事件列表操作。此外，Virgo-3将提供给其他研究人员，使他们可以利用分布式计算在他们的研究活动的好处。最后，这个项目将通过夏季计算机夏令营，针对在埃尔帕索，得克萨斯州地区的少数民族大学预科生的广泛影响。