SYSTEMS

InfiniCon Systems, the premier provider of clustering and I/O virtualization solutions for next-generation server networks, announced today that numerous customers using its InfinIO family of InfiniBand clustering solutions have been included in the TOP500 list of the world's fastest supercomputers. Further, the company's installation at the Institute of Chemical and Physical Research in Japan (RIKEN) - the seventh-highest ranking supercomputer overall - is rated the most powerful InfiniBand cluster installation in the world, according to the prestigious list. The 23rd edition of the TOP500 list was officially presented today at the International Supercomputer Conference (Heidelberg, Germany), where InfiniCon is teaming with multiple industry partners to showcase the performance advantages of clustering solutions based on the InfiniBand architecture. InfiniCon has enjoyed significant traction as an interconnect provider in the high-performance computing (HPC) clustering market. The high bandwidth and low latency attributes of its InfiniBand-based systems are ideally suited for building powerful clusters from industry-standard servers - at price points far below the cost of single-machine supercomputers. InfiniCon customers attaining TOP500 status are: -- Institute of Physical and Chemical Research (RIKEN), #7 - InfiniCon's InfinIO switching systems are powering, in partnership with Fujitsu, a 512-node fabric that is part of a larger heterogeneous cluster. The heterogeneous system is the first of its kind to attain a top-ten ranking. -- AMD Developer Center, # 277 - AMD selected InfiniCon to supply a 144-node InfiniBand fabric for its Developer Center in Sunnyvale, California. The collaboration lets users profile how the latest advances in processing and networking technologies impact real-world applications as workloads scale. In recent benchmarking performed on the Opteron-based cluster, InfiniCon's InfinIO family of InfiniBand-based solutions attained more than 80% average efficiency and up to 88% peak efficiency for scaling CPUs. -- Shandong High Performance Computing Center, # 365 - InfinIO systems are providing the interconnect for multiple university clusters that are tied into China's National Grid Project, which will ultimately link the computing resources of 100 universities and empower more than 200,000 students and researchers. Eleven InfiniBand-Based Solutions Crack TOP500 List The 23rd edition of the TOP500 shows that eleven of the world's fastest supercomputers are clustering architectures deployed with InfiniBand, whose standards-based performance advantages make it an ideal interconnect for building and scaling HPC and database clusters. Offering unmatched bandwidth (10Gbps to 30Gbps) and extremely low switching latency (below 5 microseconds for message passing applications), InfiniBand-based network fabrics permit end-users to build incredibly powerful clusters from commodity computing components - at costs as much as 90% lower than proprietary supercomputing solutions. The InfinIO 9200: Ultimate Density and Scale For HPC The news of InfiniCon's customers attaining recognition in the TOP500 list comes just as InfiniCon is adding ultra dense, ultra scalable clustering systems to its InfinIO family of solutions. InfiniCon just introduced the InfinIO 9100 and InfinIO 9200 models to its 9000 Series. Supporting from 24 InfiniBand ports to 288 InfiniBand ports in a single system, these powerful new models can deliver up to 5.76 Tbits/s of peak system bandwidth and incorporate Virtual I/O blades that let users integrate HPC clusters seamlessly into enterprise data centers for connectivity to storage and networking resources.