Designing an Energy-Efficient HPC Supercomputing Center
Designing an Energy-Efficient HPC Supercomputing Center
复制标题
设计节能 HPC 超级计算中心
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Tom Squillo
中科院分区：
文献类型：
--
作者：
Michael W. Strevell;David Lambiaso;Adam Brendamour;Tom Squillo
本文介绍了驱动节能、高性能计算（HPC）数据中心设施发展的设计考虑因素。关键的电气和机械设计考虑将提出最大限度地提高设施效率，无论是在施工和正在进行的操作。提供了一个案例研究，将这些设计决策应用于洛斯阿拉莫斯国家实验室（LANL）的多种开发方案。我们评估了从20兆瓦到80兆瓦的电力和冷却方案，并将新建筑与升级现有的HPC超级计算机中心进行了比较。指导设计决策的四个原则是支持未来系统的灵活性、可持续性、生命周期成本和初始成本。
This paper presents design considerations that drive the development of an energy-efficient, high performance computing (HPC) data center facility. Key electrical and mechanical design considerations will be presented that maximize facility efficiency, both in construction and in ongoing operations. A case study is provided that applies these design decisions to multiple development alternatives for Los Alamos National Laboratory (LANL). We evaluated alternatives ranging from 20 MW to 80 MW of power and cooling, and compared new buildings to upgrading the existing HPC Super Computer Center. The four principles guiding the design decisions are flexibility to support future systems, sustainability, lifecycle costs and initial cost.