You are here
Cube Cluster
Cube Cluster distributes model run processes across multiple computer processors dramatically cutting model run times. With Cube Cluster, a travel forecasting model that currently takes a day or more to run might only take a few hours to run. Cube Cluster can distribute run processes across the processors in one or in multiple PCs.
Cube Cluster brings a major advance to travel forecasting--the ability to run models across mutiple processors on a network of multiple computers. Cube Cluster greatly reduces run times, increasing a modeler’s ability to respond quickly to management and client demands and to efficiently run and test many more scenarios.
Methodology
Cube Cluster uses two methods to distribute model processes:
Intrastep distributed processing (IDP):
Cube Cluster splits the zone-based processing from a single step into zone groups, and distributes the groups to available computer processorss. For example, when running a mode choice model, Cube Cluster might process origin zones 1–100 on one PC and origin zones 101–200 on a second PC.
Multistep distributed processing (MDP):
Cube Cluster distributes blocks of one or more independent modeling steps across available processors. For example, Cube Cluster might run the trip generation model on one PC and the network building and skimming on a second PC.
Cube Cluster is an extension to Cube Base. In the initial version, Cube Cluster supports MDP with all Cube-initiated processes and IDP with two Cube Voyager programs:
- Highway — Used for traffic assignments and skimming
- Matrix — Used for demand model and matrix manipulation
Many travel models use these two programs for large and complex processes.
