Skip to main content

Über dieses Buch

This guide provides a comprehensive overview of High Performance Computing (HPC) to equip students with a full skill set including cluster setup, network selection, and a background of supercomputing competitions. It covers the system, architecture, evaluating approaches, and other practical supercomputing techniques.

As the world’s largest supercomputing hackathon, the ASC Student Supercomputer Challenge has attracted a growing number of new talent to supercomputing and has greatly promoted communications in the global HPC community. Enclosed in this book, readers will also find how to analyze and optimize supercomputing systems and applications in real science and engineering cases.





Chapter 1. Development and Application of Supercomputing

A supercomputer is a computer system that provides significantly higher computing power than normal personal computers. To provide a high computing performance, supercomputers are generally in the form of highly parallel systems. While supercomputers were mainly used by national labs and national defense agencies when they first came out, more and more industries are starting to use supercomputers for their research and development.
ASC Community

Chapter 2. Construction and Power Management of Supercomputing System

On the TOP500 list in November 2016, most HPC systems are either clusters (432 machines, 86.4% of the total) or MPP systems (68 machines, 13.6% of the total). The cluster architecture is now the most popular architecture among mainstream systems.
ASC Community

Chapter 3. Network Communication in a Supercomputing System

Mainstream network technologies for supercomputers include Ethernet, FC, and InfiniBand. FC is primarily designed for connecting storage devices. Due to its technical limitations, FC is only used widely in the storage domain. Both Ethernet and InfiniBand are open-network interconnect technologies. Ethernet focuses more on the versatility of the network protocols, with a unified protocol for both local area networks (LAN) and wide area networks (WAN). Therefore, Ethernet has been used widely in almost all network data transfer domains. With the data transfer speed of Ethernet gradually increasing to the same level or even above the FC, a lot of storage devices start to use Ethernet for interconnect as well. An example is the storage protocol ISCSI, which supports data transfer on Ethernet. On the other hand, InfiniBand is designed for making up the disadvantages of Ethernet and FC, as well as to meet the demands on performance and intelligence of the communication network and storage network. The performance of InfiniBand is far better than Ethernet and FC and includes intelligent features such as SDN. In recent years, InfiniBand has gradually become the dominating network technology in supercomputer systems.
ASC Community

Chapter 4. Building the Environment for HPC Applications

Six decades have passed by since the emergence of the first digital computer, and the computer industry has experienced several major technology revolutions. From the initial batch computer to the current parallel processing system based on SMP, MPP, NUMA, and clusters, computer systems have changed greatly. With the popularity of multi-core CPUs, parallel applications have gradually become the mainstream. Therefore knowing how to build an environment for parallel applications has become an urgent issue.
ASC Community

Chapter 5. Supercomputer System Performance Evaluation Methods

For high-performance computers or supercomputers, the powerful computing performance is the most important value and advantage. How to evaluate the performance of a high-performance computer system is of general concerns to users, manufacturers and research institutions. Performance evaluation is also an indispensable step in the development, selection, introduction and effective utilization of a computer. The answers to the following frequently asked questions are all more or less related to the system performance evaluation.
ASC Community



Chapter 6. An Introduction to International Student Supercomputing Competitions

There are three international supercomputing competitions for undergraduate students including the ASC Student Supercomputer Challenge, the ISC Student Cluster Competition, and the SC Student Cluster Competition.
ASC Community

Chapter 7. History and Prospects of ASC Student Supercomputer Challenge

In 2012, with the guidance of the Ministry of Science and Technology of China and the Ministry of Education of China, and with the joint support of the Inspur Group, the International Supercomputing Conference Organizing Committee (ISC), and the international HPC Advisory Committee (HPC-AC), the “Student Supercomputer Challenge and ISC Student Cluster Challenge Regional Preliminary Stage” was introduced into China for the first time. This competition received strong support from the Chinese government and the scientific community in China. Yilian Jin, Xubang Shen, Xin-gui He and Zuoning Chen, four senior academicians of the Chinese Academy of Engineering, served as the chairman and vice-chairmen of the competition steering committee. This competition also received full support from the Department of High and New Technology Development and Industrialization of Ministry of Science and Technology of China, and Tsinghua University.
ASC Community

Chapter 8. Rules of ASC Student Supercomputer Challenge

The 2014 Asia Student Supercomputer Challenge (ASC14) was the first supercomputing competition open to college students from around the world. Applications came from over 100 universities from all over the world. In this chapter, we explain the rules of the preliminary stage and the final stages of ASC14.
ASC Community



Chapter 9. Competition Proposal

The ASC student cluster preliminary competition requires a proposal submission as a perquisite for evaluation by the competition judges for finalist entry selection. This chapter will provide some insights to the efforts and suggested methods required for preparing the proposal and guiding a team to overcome these challenges at a first attempt. The proposal demonstrates the students’ theoretical knowledge on cluster competition subjects ranging from hardware architecture, parallel programming, and software applications, to code optimization. This preliminary stage also provides an excellent chance to learn practical aspects of high performance computing (HPC) or supercomputing in a very competitive learning environment.
ASC Community

Chapter 10. Design and Construction of the Clusters for the Competition

The history of cluster use can be dated to the middle 1990s. In recent years, cluster-based architecture has become more and more popular due to its cheap price, good flexibility, and good scalability. There are more and more cluster-based supercomputers in the Top500 list, since the first cluster entered the list in June 1997. In November 2003, there are already 208 cluster-based supercomputers in the Top500 list. In the recent 23 publications of the Top500, the top supercomputers are all based on cluster architecture. For example, the current top supercomputer is the Tianhe-2 supercomputer in Guangzhou China, designed and developed by the College of Computer, National University of Defense Technology. Therefore, the cluster is definitely the most successful system architecture in the field of high-performance computing.
ASC Community

Chapter 11. Optimization for the High Performance LINPACK Benchmark

HPL (High Performance LINPACK) benchmark can reflect the system’s capacity to do floating-point operations, and is the most popular way to evaluate the performance of the system. The two releases of the Top500 list every year use the HPL test to rank different systems.
ASC Community

Chapter 12. Optimization for the Molecular Dynamics Software GROMACS

In this chapter, we will first give a brief introduction about the GROMACS scientific computing software, and its experiment results based on a four-node HPC cluster. Furthermore, we propose and analyze the optimizations for GROMACS.
ASC Community

Chapter 13. Optimizations for Ocean Model LICOM

LICOM[14-1] (LASG/IAP Climate system Ocean Model) is an ocean model in the climate system developed by the State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG), Institute of Atmospheric Physics (IAP), and the Chinese Academy of Science. LICOM is designed by the global ocean-atmosphere coupler group to simulate large-scale wind circulation and thermohaline circulation. Besides the ocean model, there are other models such as the atmosphere model, ice mode, and land model. They are connected by the coupler to form a complete climate model. LICOM is a big contribution by LASG to the development of the climate system.
ASC Community

Chapter 14. Optimization of Three Dimensional Elastic Wave Modeling Software 3D_EW

The 3D_EW (3D Elastic Wave Modeling) team of Shanghai Jiao Tong University (SJTU) in ASC14 contains two advisors, including Xinhua Lin, who is the deputy director of the High Performance Computing Center of SJTU, and Minhua Wen, who is an engineering specialist, and three students that are in charge of the work of code porting and optimizations.
ASC Community
Weitere Informationen

Premium Partner

BranchenIndex Online

Die B2B-Firmensuche für Industrie und Wirtschaft: Kostenfrei in Firmenprofilen nach Lieferanten, Herstellern, Dienstleistern und Händlern recherchieren.



Best Practices für die Mitarbeiter-Partizipation in der Produktentwicklung

Unternehmen haben das Innovationspotenzial der eigenen Mitarbeiter auch außerhalb der F&E-Abteilung erkannt. Viele Initiativen zur Partizipation scheitern in der Praxis jedoch häufig. Lesen Sie hier  - basierend auf einer qualitativ-explorativen Expertenstudie - mehr über die wesentlichen Problemfelder der mitarbeiterzentrierten Produktentwicklung und profitieren Sie von konkreten Handlungsempfehlungen aus der Praxis.
Jetzt gratis downloaden!