Limitations of the Distributed Processing Module

For the most part, if you need to run Monte Carlo simulations for a complex model, the use of the Distributed Processing Module is advantageous and strongly encouraged. There are some situations, however, in which use of the Distributed Processing Module may not be feasible.

In particular, because a distributed processing simulation requires that potentially large quantities of data (i.e., the GoldSim model file and other associated files, the results of each realization) be passed between the Controller and Client processes over a network, effective use of the Distributed Processing Module requires fast network connections (and fast computers). If the connections between the processes are slow, a distributed processing simulation may be impractical.

In addition, there are three situations in which distributed processing is not supported at all:

• You cannot run multiple scenarios (using the Scenario Manager) and utilize distributed processing. That is, the Scenario Manager provides no access to distributing processing capabilities.

• Because distributed processing can not be used for a single realization, and high resolution results (in which all unscheduled timesteps are displayed) are only available for single realizations runs, high resolution results can never be viewed for distributed processing simulations.

• You cannot export results to a Spreadsheet element when doing a distributed processing run (as each realization would result in a different spreadsheet file).