Parallel Implementation of ADVISE on the Intel Paragon

Environmental models, such as the Pasture Model, require large computing resources to give results within reasonable time frames. This report focuses on the design and implementation of a parallel version of an ADVISE server running the Pasture Model on the Intel Paragon. In order to make the most efficient use of this computing resource, it is first necessary to perform a thorough analysis of the message passing, file I/O and computation performance of this machine. These results are compared to those obtained on a typical workstation such as a SUN SPARC. These results are used to design an efficient implementation of a parallel server version of the Pasture Model. This implementation consists of a single master server to process client requests and re-distribute these requests to a number of sub-servers. The Paragon performance results are used to determine what functionality should be implemented in the sub-servers in order to achieve an efficient use of the Paragon and at the same time optimise server response times for client requests