Extracting bus elements and their signals from a data dictionary has very slow performance.

I have inhertited a script that I recently updated to search a data dictionary instead of the base workspace to extract buses and signals. It takes ~4 hours to process ~140 buses /~7000 signals in the data dictionary. When these signals were in the base workspace the processing time was 1.6 seconds.

The profile viewer shows that this line consumes 99% of the script if (exist('ddatasectobj','var')) && (~isempty(find(find(ddatasectobj,'-value','-class','Simulink.Bus'),'Name',buselements{nelements,2})))

Is it a feature of using the data dictionary or is there a better method and I am just missing it?

%iterate through each element to find any nested buses, then

%recursively get all those bus elements... and so on...

for nelements=1:size(buselements,1)

% Remove Bus: prefix from datatype string

buselements{nelements,2} = regexprep(buselements{nelements,2},'^[Bb]us:(\W*|)','');

if (exist('ddatasectobj','var')) && (~isempty(find(find(ddatasectobj,'-value','-class','Simulink.Bus'),'Name',buselements{nelements,2})))

buselements{nelements,5}='bus';

%if haven't already defined, recursively call this function to get all nested buses

if ~sum(strcmp(buselements(1:nelements-1,2), buselements{nelements,2}))

busobjects=[busobjects;getbusdef(slddname,buselements{nelements,2})];

end

Answer 1

[Ashutosh Thakur]

Hi Dawn,

I understand that you are trying to search through a data dictionary but are facing slow performance while executing these queries. The increase in processing time between accessing the base workspace and a data dictionary in Simulink is likely due to the overhead associated with these queries. As you mentioned, a particular query consumes 99% of the overall time during the execution of your script.

The following steps can be followed to improve performance:

Cache Repeated Searches:

• You can try caching the repeated searches in the form of a MATLAB variable or a MAT file. This will prevent querying the data dictionary again and use in-memory results for better performance. Refer to the following documentation link to improve the performance of MATLAB code:https://www.mathworks.com/help/matlab/matlab_prog/techniques-for-improving-performance.html.

Bulk Data Retrieval:

• If you have a bulk of data, try executing the query to get dictionary data without using loops, and then store this data in a MATLAB variable. For example, the following sample code will retrieve all the Simulink.Bus objects from the data dictionary section object passed to it:

    % ddSectObj is the Simulink.data.dictionary.Section object
    find(ddSectObj, '-value', '-class', 'Simulink.Bus').

• Store the results of the above-mentioned code in a MATLAB variable and loop through only the bus objects.
• Also refer to the following documentation link to use find and getSection functions: https://www.mathworks.com/help/matlab/ref/find.html. https://www.mathworks.com/help/simulink/slref/simulink.data.dictionary.getsection.html

Use Parallel Computing:

• Consider leveraging MATLAB's Parallel Computing Toolbox to parallelize some of the operations to reduce execution time. Refer to the following documentation link for more information: https://www.mathworks.com/help/parallel-computing/quick-start-parallel-computing-in-matlab.html.

I hope this helps you in improving the performance.