Multiple Run Management
The following changes were made to Multiple Run Management.
Distributed MRM Configuration Wording
On the Distributed Runs tab of the MRM Configuration dialog, the wording regarding the options for distributing traces was improved. Now rather than “Distribute Traces Across N Multiple Runs” the new wording is “Distribute traces to N RiverWare instances” in order to make the meaning more clear. The behavior has not changed.
Figure 1.4 Screenshot of Number of RiverWare Instances in Distributed MRM
Distributed MRM Assignment to Processors
Distributed MRM starts a specified number of RiverWare instances and previously allowed the operating system to distribute those instances to individual processors. For some combinations of hardware and operating systems, not all available processors would be utilized for the distributed RiverWare instances. Thus, the full potential benefit of parallelization was not realized. This was filed as issue RW-6783. To address this, options were added to specify directly how many processors to use for the distributed MRM. The MRM then assigns each RiverWare instance directly to a single processor. The following options are now available:
• Let the operating system decide (the default and existing behavior, which may not utilize all processors)
• Assign each RiverWare instance to a single processor, using up to all available processors
• Assign each RiverWare instance to a single processor, using a specified number of processors
For more information, see Assignment of RiverWare Instances to Logical Processors in Solution Approaches.
Figure 1.5 Screenshot of Distributed MRM Processor Assignment Options
Distributed MRM Environment Variables Panel Removed
The Distributed Runs tab of the MRM Configuration dialog previously had a panel for specifying environment variables to use within the distributed MRM. This panel has been removed as it is no longer used. If you used environment variables specified in this panel in your model, please contact riverware-support@colorado.edu.
MRM Output to RDF
See the important note in MRM Output to RDF.
Using RPL Sets in MRM that are Already Loaded
See the important note in MRM with RPL Sets Already Loaded.
Revised: 01/11/2023