摘要
Performance and energy consumption of high performance computing （HPC） interconnection networks have a great significance in the whole supercomputer, and building up HPC interconnection network simulation plat- form is very important for the research on HPC software and hardware technologies. To effectively evaluate the per- formance and energy consumption of HPC interconnection networks, this article designs and implements a detailed and clock-driven HPC interconnection network simulation plat- form, called HPC-NetSim. HPC-NetSim uses application- driven workloads and inherits the characteristics of the de- tailed and flexible cycle-accurate network simulator. Besides, it offers a large set of configurable network parameters in terms of topology and routing, and supports router＇s on/off states. We compare the simulated execution time with the real execution time of Tianhe-2 subsystem and the mean error is only 2.7%. In addition, we simulate the network behaviors with different network structures and low-power modes. The results are also consistent with the theoretical analyses.
Performance and energy consumption of high performance computing （HPC） interconnection networks have a great significance in the whole supercomputer, and building up HPC interconnection network simulation plat- form is very important for the research on HPC software and hardware technologies. To effectively evaluate the per- formance and energy consumption of HPC interconnection networks, this article designs and implements a detailed and clock-driven HPC interconnection network simulation plat- form, called HPC-NetSim. HPC-NetSim uses application- driven workloads and inherits the characteristics of the de- tailed and flexible cycle-accurate network simulator. Besides, it offers a large set of configurable network parameters in terms of topology and routing, and supports router＇s on/off states. We compare the simulated execution time with the real execution time of Tianhe-2 subsystem and the mean error is only 2.7%. In addition, we simulate the network behaviors with different network structures and low-power modes. The results are also consistent with the theoretical analyses.
作者简介
Wenhao Zhou received the BS and MS de- grees in the School of Computer, National University of Defense Technology, China in 2013 and 2015. His research interests fo- cus on energy-aware HPC interconnection networks and parallel software framework.Juan Chen received the PhD degree in Computer Department, National University of Defense Technology （NUDT）, China in 2007. She is now an associate profes- sor in Key Laboratory of High Perfor- mance Computing at NUDT. Her research interests focus on supercomputer systems,energy-aware interconnection network design, and parallel software framework.Chen Cui received the School of Electronics BS degree in the Engineering and Computer Science at Peking University, China in 2015, and now he is a MS student at National University of Defense Technol- ogy, China. His research interests focus on the large scale parallel numerical simula- tion and parallel software framework.Qian Wang received the BS degree in the School of Computer at National University of Defense Technology （NUDT）, China in 2011, and now is a PhD student at NUDT. Her research interests focus on the large scale parallel numerical simulation and par- allel software framework.Dezun Dong received the BS, MS, and PhD degrees from the Nation＇＆ University of Defense Technology （NUDT）, China in 2002, 2004 and 2010, respectively. He is an associate professor in the Collage of Computer, NUDT. His research interests range across high performance computer systems, high speed interconnect networks,wireless networks, and distributed computing algorithms. Currently, he focuses on performance evaluation of high-performance inter- connection networks for supercomputers and data centers. He is a member of the ACM, 1EEE, and CCEYuhua Tang received her BS and MS de- grees in Computer Department from Na- tional University of Defense Technology （NUDT）, China in 1983 and 1986, re- spectively. She is now a professor in National Laboratory for Paralleling and Dis- tributed Processing at NUDT. Her research interests include supercomputer architec-ture and core router＇s design.
