|
When performing design studies involving Design Of Experiment or Optimization methods, OPTIMUS is able to launch several simulations simultaneously, each simulation with different values for the design variables.
Especially when the OPTIMUS workflow contains multiple simulation components, OPTIMUS smartly tracks the dependencies between the simulations and launches as many simulation jobs as possible, but will only start those jobs for which the upstream jobs are completed.
Currently OPTIMUS offers an advanced deep-level parallelization scheme that not only parallelizes for the requested virtual experiments, but also tries to run each simulation code on a different network node, all depending on the number of available CPUs and software licenses.
The parallel execution of OPTIMUS benefits greatly from the presence of a supported network management system, such as Platform/LSF, PBS Pro and Sun GridEngine, that manages all OPTIMUS jobs in a cluster or network of computers.
|
