Running the Sequential Version

Run the sequential version of Feko with optional parameters.

It is recommended to run the Feko kernel directly from the GUI components CADFEKO, EDITFEKO or POSTFEKO. Once a session or model has been loaded, the sequential Feko solver can be started from the Solve/Run tab, by selecting Feko (the shortcut key Alt+4 can also be used).

While the Feko kernel is running, the status of the calculation phases is indicated on the Executing runfeko dialog, see Figure 1. The output generated by the Feko kernel is hidden by default. The Feko kernel output may be viewed by clicking on the Details button and selecting the Output tab. Similarly, notices, warnings and errors can be viewed by selecting the Notices, Warnings and Errors tabs respectively.

Figure 1. The Executing runfeko dialogs. Output generated by the kernel is hidden by default. Solver output may be viewed by clicking on the Details button.
When the Feko kernel is not executed from within the GUI, it can be started in a command window (on a Windows PC) or a shell (in UNIX) by executing the command:
runfeko example08

where example08.pre must be an existing input file. RUNFEKO executes PREFEKO if the .fek file is missing or older than the .pre file and then executes the appropriate Feko solver.

RUNFEKO accepts the optional parameters listed below. More information regarding additional options for launching and controlling the parallel version of the solver can be found in Running the Parallel Version. Additional options for the remote launching of Feko are found in Running on a Remote Host.

In CADFEKO these settings are available by selecting the Solve/Run tab and clicking on the dialog launcher button on the Run/launch group. For POSTFEKO, select the Home tab and click on the dialog launcher button on the Run/launch group.

Print the version information and then exit.
--priority x
The value x specifies the CPU usage priority of the Feko run:
0 = idle
1 = below normal
2 = normal
3 = above normal
4 = high.
If not specified, the default is 2. This option might not be available for specific systems or for specific Feko versions. In this case it is just ignored.
[NUM_GPUS][:GPU_1[,GPU_N]] Execute Feko using GPU acceleration. The optional parameters are:
NUM_GPUS: The number of devices to use.
GPU_1 ,GPU_N: A comma separated list of specific devices to use. If the option is specified without the optional parameters, all available GPU resources are used. If NUM_GPUS is specified, the first NUM_GPUS devices in the system will be selected. Specifying --use-gpu 0 will completely disable GPU detection and prevent NOTE 35179 from being printed.
Example usage is as follows:
--use-gpu 2:0,2 which uses the first (device 0) and third (device 2) GPU in the system.
This is equivalent to --use-gpu :0,2.
Activates the MPI method on Windows.
Always execute CADFEKO_BATCH first (by-pass automatic checks based on file existence and date stamps.)
CADFEKO_BATCH will not be run to create a new .cfm and .pre file.
Always execute PREFEKO even if the existing .fek file is newer than the .pre.
PREFEKO will not be run to generate a new .fek file before the Feko solver is launched, even if the .fek and/or .cfm files are older than the existing .fek file.
Run the parallel Feko kernel within a queuing system and obtain the number of parallel processes as well as the host list directly from the respective job scheduler.
The Intel MPI library supports the following job schedulers:
Microsoft Windows
  • Altair PBS Professional
  • Microsoft HPC Pack
  • Altair PBS Professional
  • Torque
  • OpenPBS
  • IBM Platform LSF
  • Parallelnavi NQS
  • Univa Grid Engine
Debug mode with extra output (can be used to troubleshoot errors).
All options following, up to the next –xxx-options, are passed to PREFEKO.
All options following, up to the next –xxx-options, are passed to Feko.
All options following, up to the next –xxx-options, are passed to ADAPTFEKO.

The optional command line parameters for Feko (specified after --feko-options) are listed below.

Feko processes and checks the model, but does not start a solution. This is useful to, for example, check an input file on a local computer before submitting it to a cluster.
Feko processes the model and provides an estimate for the memory consumption. The estimated value is provided at the end of the .out file.
Note: An estimate is only available for:
  • MoM
  • PO (not hybridised with any other solution methods).
Tip: For a more accurate estimate, run the estimation with the intended number of processes on the intended host(s).
-e ENV=value
This has the same effect as starting Feko with the environment variable ENV set to value. More than one -e … argument is allowed.
--data-export-format n
Use the nth version format for the data export files (.efe, hfe, .ffe, .os, .ol). Allowed values for n are 1 and 2 where 2 is the latest version (since Feko 6.1). If not specified, the default is to use the latest supported version.
Special execution mode to export SPICE MTL circuit files.