|
|
Design and implementation of communication system
of the Dawning 6000 supercomputer |
Qiang LI1,Bo LI1,Zhigang HUO2,Ninghui SUN2, |
1.National Research Center
for Intelligent Computing Systems, Beijing 100190, China;Key Laboratory of Computer
System and Architecture, Chinese Academy of Sciences, Beijing 100190,
China;Graduate University of
Chinese Academy of Sciences, Beijing 100190, China; 2.National Research Center
for Intelligent Computing Systems, Beijing 100190, China;Key Laboratory of Computer
System and Architecture, Chinese Academy of Sciences, Beijing 100190,
China; |
|
|
Abstract An increasing number of supercomputers adopt a heterogeneous architecture, consisting of both general purpose CPUs and specialized accelerators. Such design is beneficial for scalability and power, but on the other hand, heterogeneity brings new challenges in communication systems to connect heterogeneous components and provide support for programming. The communication system of the Dawning 6000 connects two kinds of heterogeneous processors, Loongson and AMD, and adopts a three layer architecture with an intra-node layer between heterogeneous components. To efficiently connect heterogeneous components, the system forms a global address space and provides a mechanism for message transmission via an in-node global store; and employing Infiniband network, provides an OS-bypassing virtualization method to share an Infiniband card between nodes. To facilitate programming on heterogeneous processors, it supports unified parallel C (UPC), with a modified complier based on global address space. Also, a special collective network is implemented for collective operations. Results obtained from a prototype system prove these features to be both feasible and efficient.
|
Keywords
hyper parallel processing (HPP)
global address space (GAS)
virtualization
Dawning 6000
unified parallel C (UPC)
|
Issue Date: 05 December 2010
|
|
|
Sun N, Li K, Chen M. HPP: an architecture for high performanceand utility computing. Chinese Journalof Computers, 2008, 31(9): 1503–1508
doi: 10.3724/SP.J.1016.2008.01503
|
|
Carrera E V, Rao S, Iftode L, Bianchini R. User-level communication in cluster-based servers. In: Proceedings of 8th International Symposium on High-Performance ComputerArchitecture. 2002, 275–286
|
|
Ries R. Communication patterns [message-passing patterns]. In: Proceedings of 20th International Paralleland Distributed Processing Symposium, 2006
|
|
Consortium UPC. UPC Language Specifications v1.2. Lawrence Berkeley National Lab Tech Report LBNL-59208, 2005
|
|
Buyya R, Cortes T, Jin H. Single system image. International Journal of High Performance ComputingApplications, 2001, 15(2): 124–135
doi: 10.1177/109434200101500205
|
|
Gara A, Blumrich M A, Chen D, et al. Overview of the blue gene/lsystem architecture. IBM Journal of Researchand Development, 2005, 49(2): 195–212
doi: 10.1147/rd.492.0195
|
|
ref="http://www.lanl.gov/orgs/hpc/roadrunner/pdfs/Roadrunner-tutorial-session-5-web1.pdf" target="blank">
|
|
rdware/clearing-up-the-confusion-over-intels-larrabee.ars" target="blank">
|
|
Zhang P Y, Meng D, Huo Z G. Research of collectives optimizationon modern multicore clusters. Chinese Journal of Computers, 2010, 33(2): 317–325
doi: 10.3724/SP.J.1016.2010.00317
|
|
Barham P, Dragovic B, Fraser K, et al. Xen and the art of virtualization. In: Proceedings of the 19th ACM symposium on Operatingsystems principles. 2003, 164–177
|
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
|
Shared |
|
|
|
|
|
Discussed |
|
|
|
|