Special simulations may require a different executable file, modifications to the Control file that are not supported by the GUI, or interactions with other items under the Advanced Menu tab. More information is provided below for each special simulation that can be run under a multi-processor environment that supports MPI. These special simulations are available only for UNIX operating systems. All the MPI simulations execute the special script run_mpi.sh which can be found in the /exec directory. This script executes the appropriate MPI executable variation of the concentration model and almost certainly will require some customization to match the local operating system environment.
Run MPI Model
The standard concentration model can be run on multi-processor systems. As pollutant particles are released during the simulation, they are parsed out in sequence to the available processors. The calculation proceeds independently until the end of a concentration averaging period. At this point, the concentrations from each processor are summed, and only one concentration output file is updated. The MPI version can be quite effective in speeding up simulations requiring the release of many particles. No special configuration or control file is required and output can be viewed using the standard concentration display menu.
Run Matrix Model
The multiprocessor version of the matrix calculation is configured the same way as for a single processor system. The MPI calculation proceeds as described above for the standard MPI model simulation.
Run Ensemble Model
The ensemble model automatically starts each member on a single processor in a multi-processor environment. The multiprocessor version of the ensemble calculation is configured the same way as for a single processor system. Note that if fewer than 27 processors are available, the ensemble configuration menu permits starting the calculation at any ensemble member number within the valid range of 001 to 027.
The Model Launch Menu
The model launch menu contains four entries: the number of processors requested for the simualation, the name of the executable, the Prep-Code, and the working directory. Each of these is assigned a default value depending upon the calling menu. Normally there should be no reason to change any of these values except the number of processors. If the working directory is changed and does not already exist, a new directory is created, and the script changes to that directory before starting the simulation.
If the Prep-Code field is left blank, the MPI script is not called, and the named executable will be run in background. In this way multiple jobs can be submitted on a multi-processor system in a non-MPI environment. However, in both MPI and non-MPI applications, when simultaneous calculations are desired, it is important that each simulation have its own working directory.
Table of Contents