能耗并行加速比:高性能计算系统综合性能的有效度量  被引量:2

Power Parallel Speedup:An Effective Metric for Evaluating the Comprehensive Performance of High-Performance Computing Systems

在线阅读下载全文

作  者:王之元[1] 胡庆丰[1] 陈娟[1] 

机构地区:[1]并行与分布处理国家重点实验室,湖南长沙410073

出  处:《计算机工程与科学》2009年第11期113-116,共4页Computer Engineering & Science

基  金:国家863计划资助项目(2008AA01Z137);国家自然科学基金资助项目(60621003;60633050;60873014;60903044)

摘  要:随着并行系统规模的扩大,高性能计算系统运行时消耗的能耗也在急剧增长,过高的能耗也给系统的可靠性、稳定性等方面带来严峻挑战。在这种情形下,能耗问题受到了前所未有的关注。因此,设计和研究高性能计算系统,需要在考虑高计算性能的同时兼顾系统低能耗的要求,这为高性能计算系统的度量模型提出了新的挑战。于是,大规模并行系统逐渐从"高性能"走向"高效能"的衡量标准。基于此,本文采用加速比度量指标,从系统可扩展角度将计算性能和能量消耗要素进行综合,提出了一种度量高性能计算系统综合性能的能耗并行加速比模型。该模型能够直观地反映并行计算系统的效能,旨在指导系统设计和应用研究。最后,通过对该模型的分析和模拟,验证了模型的有效性。As parallel systems scale up, the energy consumption of high performance computing systems drastically increases. The use of high power and energy leads to many problems, such as low reliability, bad stability and so on. In this case, the energy consumption problem draws unprecedented attention. Thus it is necessary to take into account the energy consumption while pursuing high computing performance in system design and research. This brings a new challenge for building a high performance computing system metric model. The measure index for the performance of the large-scale system demands a shift from "high performance" to "high productivity". Then, this paper uses speedup, synthetically measures the computing performance and energy consumption in view of system scalability, and proposes a power parallel speed- up for evaluating the comprehensive performance of high-performance computing systems. The mode/can measure the productivity of parallel systems and guide system design and application research. Finally, we validate the model of power paralle/speedup through analysis and simulation.

关 键 词:能耗并行加速比 高效能 度量模型 

分 类 号:TP302.7[自动化与计算机技术—计算机系统结构] TP338.4[自动化与计算机技术—计算机科学与技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象