EBSILON®Professional Online Documentation
Calculations / Parallelization
In This Topic
    Parallelization
    In This Topic

    Parallelization 

    The computing core of Ebsilon has been modified in such a way that certain processing steps can be run in parallel on computers with several processors:   

    A noticeable reduction of the computing time will mainly result when using complex material data, e.g. mixtures from the Refprop library.

    The disadvantage of the parallelization is that the sequence in which the individual components are processed is no longer fixed. In particular, this has consequences for the controllers and the programmable components 65 and 93. Setting up the equation system is basically independent of the processing sequence, but as the compositions are not integrated into the equation system by default, it previously depended on the processing sequence which composition a component saw on a line.

    Usually the controllers were built in after the other components, so that the controllers could already use the composition that the other components had determined in the same iteration step. To further ensure this behaviour, there is a flag FSEQ for each controller by means of which the user can specify whether the controller can be edited in parallel with the other components or only after the other components (see Release 10, Chapter 1.3, component 12, 39, 65, 69, und 93). By default this flag is set in such a way that the controllers "see" the new composition. The same applies for the new components 65 and 93.

    Whether the parallelization is to be used and how many legs are to be calculated in parallel at most can be defined in the general settings under “Calculation“. There you also find the option to allocate the execution of Ebsilon to certain processors.

    These settings can also be made via the EbsOpen interface.