"Silicon Galaxy" system area network for collective communication in supercomputers

The paper focuses on the latest trends in supercomputer and high-performance computer architectures with emphasis of the important parameters for parallel computer performance. The main objective is to propose the high-speed network architecture for collective communications in supercomputers "Silicon Galaxy" that provides high-bandwidth, low latency and high tolerance to failures. The proposed network architecture "Silicon Galaxy" is composed of "Milky Way Galaxy" network for intercluster communication and "Dwarf Galaxy" network for communication within the cluster segments (intracluster communication in supercomputers). Communication performance of the proposed collective network has been evaluated on the base of parallel computer simulations on IBM Blade Center.

[1]  Plamenka Borovska,et al.  Architectural Design of Grand Clos Collective Network for Supercomputers , 2013 .

[2]  Plamenka Borovska,et al.  Communication performance evaluation and analysis of a mesh system area network for high performance computers , 2010 .

[3]  Pierre Fraigniaud,et al.  Degree hunter: on the impact of balancing node degrees in de Bruijn-based overlay networks , 2010, 2010 IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum (IPDPSW).

[4]  Torsten Hoefler,et al.  Optimization principles for collective neighborhood communications , 2012, 2012 International Conference for High Performance Computing, Networking, Storage and Analysis.

[5]  Laxmikant V. Kalé,et al.  A Hierarchical Approach for Load Balancing on Parallel Multi-core Systems , 2012, 2012 41st International Conference on Parallel Processing.

[6]  Valentin Puente,et al.  TOPAZ: An Open-Source Interconnection Network Simulator for Chip Multiprocessors and Supercomputers , 2012, 2012 IEEE/ACM Sixth International Symposium on Networks-on-Chip.