Distributed CCA Components and Grid Services for Scientific Computing

76334S There is an increasing need to design large-scale scientific computations by combining established codes that use multiple physics models. Currently, there is no systematic way to develop and maintain these applications, which require the simultaneous use of high-performance and distributed resources, such as distributed clusters and supercomputers. To address this problem, this project will extend an existing Common Component Architecture (CCA) framework to support the development of distributed and parallel high-performance computing applications. In Phase I, prototype Remoting Components for adding distributed capability were developed, preliminary performance comparisons were conducted, and procedures for passing arrays in the CCA framework were identified. Phase II will develop software tools for assisting both the creation of Remoting Components and the development of distributed and parallel high-performance computing applications. Support for a modern language feature, ¿struct', will be added to the scientific interface definition language (SIDL.) Finally, examples that demonstrate the usage and benefit of distributed and parallel high-performance computing will be developed. Commercial Applications And Other Benefits as described by the awardee: All software developed will be released as Open Source. Training and consulting services will be provided to customers in industry and government that utilize high-performance computation in such applications as weather and climate modeling, automobile development, and battlefield simulations.