APPLICATIONS

Bull and Quadrics today announced clusters designed for large high performance computing centers. These clusters encompass the power of Bull's new Intel(R) Itanium(R) 2-based NovaScale(TM) servers and the performance of Quadrics' clustering solutions. Partnering to deliver power and flexibility Bull's objective is to provide Teraflop computing power to demanding computing centers through standard components-based solutions. Through the Quadrics QsNetII interconnection network, they will combine several hundred large servers each supporting Intel(R) Itanium(R) 2 with 2 to 16 processors (and rising to 32-way servers in the near future). Bull's offer, therefore, answers the needs of most demanding configurations in terms of power, scalability and robustness. It also provides for easy administration and maintenance. With Bull's NovaScale(TM) server range being based on standard components, Bull is able to deliver an extremely competitive price/performance ratio while also supporting the benefits of open software solutions. As a result, Research Centers will benefit from the continuity associated with standards-based HPC solutions. Teaming up with Quadrics, Bull integrated and optimized the best of breed clustering solutions on its Bull NovaScale(TM) servers. These include QsNetII, a high performance interconnection network capable of scaling within SMPs as well as scaling out across large numbers of SMP nodes, and RMS (Resource Management System), a cluster configuration management environment. RMS includes events monitoring as well as launching and controlling parallel tasks through a single point of administration. "We feel very excited and privileged to be chosen by Bull to work with them to develop the Bull clustering solutions for high performance computing. Being chosen by one of Europe's leading manufacturers of computing technologies is another example that the benefits of using Quadrics technologies in today's clustering solutions is of true value to our partners. I am confident that Bull will be able to provide solutions that will be highly appreciated by large laboratories and research centers to fulfill their demanding requirements in high performance computing", said John Taylor, Technical Sales Manager of Quadrics. "The conjunction of Bull's wide experience in the architecture of large servers based on standard components and Quadrics leadership in clustering solutions will provide our customers with highly performing solutions and greatly reduced deployment costs, when compared to proprietary solutions", said Gerard Roucairol, Chief Scientist of Bull.