Distributed Simulation Cluster
CUBINlab has a Mandrake Linux Cluster consisting of a growing number of machines.
These machines can be used for distributed simulation using tools such as MPI & PVM, individually to run MATLAB simulations, or using proprietory techniques to spawn processes on the various machines and collecting/collating processed data at a later stage.
Presentation Material
- "Distributed Simulation in CUBINlab", presented at a CUBIN seminar on 22 January 2003.
Parallelising Jobs
Some rough scripts have been written to help you to run multiple simulations in parallel. Consult the preliminary documentation on spawning jobs.
Cluster Machines
The following table lists the details of each machine we currently have in the cluster.
You may see the status of each machine by selecting the appropriate image. Alternatively, you may see the status of the cluster as a whole.