SoundPLAN Distributed Computing
Scale up your model processing power

Distributed Computing speeds up lengthy calculations by distributing the application on any number of PCs found in the network. Install a regular version or the demo version as a calculation server and scale your application to the speed you need. With just this module on one PC, there is no limit to the number of PCs used for mapping calculations. Multi-Threading is also an option to speed up calculations on a multi core PC. This option is included in the most basic packages.

Computers that participate in the Distributed Computing Calculation Network do not have to be fully licensed copies of SoundPLAN. If the module for DC is present in the Master, it can distribute the calculations amongst an unlimited number of non-licensed slaves. With modern multi core PCs it is even feasible to edit data on the same PCs that is the Master to the DC network while the calculations are in progress. DC works with all calculation types, but mostly it makes sense to use it for Grid Noise Maps, Cross-Sectional Noise Maps and Facade Noise Maps.

How does the Distributed Computing work?

For a "Slave" to do the computation intensive work, several things must happen:

  1. At least the demo software must be installed
  2. The version of SoundPLAN must be identical for Master and all Slaves (same compile date)
  3. On the Slaves DC calculations must be enabled

In the settings of both the Master and the Slave the port numbers in the Client Socket and the Socket Server Socket must be identical, without these ports DC cannot communicate. In case a port number is used by another software, you can choose a different socket number but these must be the same throughout the DC network.

For the Slaves you can establish how many of the socket Servers to start and how many threads each should use. This setting depends on the question if anyone wants to work on the PC while a DC calculation is in progress. The total of the instances x threads should not exceed the number of logical cores the PCs processor has.

The last setting fro a DC Slave is to customize where the Masters intermediate data shall be hosted for the duration of the calculations. (Path for distributed computing data)