How Does Distributed Processing Work?
Running a Monte Carlo Simulation with Multiple Processors
The Distributed Processing Module uses multiple copies of GoldSim running on multiple machines (and/or multiple processors or cores within a single machine) to share the computational burden of a Monte Carlo simulation.
A distributed processing simulation is controlled by a single Controller GoldSim executable. The Controller executable is a version of GoldSim running on one machine which has the Distributed Processing Module enabled. Through a dialog accessed via a menu item, the Controller can create connections with multiple Clients.
The Controller assigns and sends individual realizations to the Client processes. Each Client process carries out its assigned realization, sends the results to the Controller, and then makes itself available to work on another realization. Once all of the realizations have been completed, the Controller combines all of the results into a single GoldSim file just as if the entire simulation had been carried out on the Controller’s machine.
The GoldSim model file (and other required files such as linked spreadsheets) need not be present on machines running Client processes. The Controller automatically transfers any necessary files to the Client machines.
When running as a Client process, GoldSim is run in a special mode. Rather than launching GoldSim in the traditional manner on the machine running the Client, a GoldSim Client is launched from the command line (this can be done using a Windows shortcut or the Run option on the Windows Start menu). When running in Client mode, you cannot view or edit a GoldSim model. The only function that a Client serves is to receive, process, and send back realizations sent to it by the Controller. Note that Clients do not require purchase of a separate GoldSim license, since the free GoldSim Player can act as a Client.
Learn More
- Download the GoldSim Distributed Processing Module User's Guide