Record Coverage in Parallel Simulations by Using Parsim
This example shows how to record coverage in multiple parallel Simulink® simulations corresponding to different test cases by using SimulationInput objects and the parsim command. If Parallel Computing Toolbox is installed on your system, the parsim command runs simulations in parallel. Otherwise, the simulations are run in serial.
The slvnvdemo_powerwindow_parsim model contains a power window controller and a low-order plant model. The component slvnvdemo_powerwindow_parsim/power_window_control_system/control is a Model block that references the model slvnvdemo_powerwindow_controller, which implements the controller with a Stateflow® chart.
mdl = 'slvnvdemo_powerwindow_parsim'; isModelOpen = bdIsLoaded(mdl); open_system(mdl);
Set Up Data for Multiple Simulations
Determine the number of test cases in the Signal Editor block by using the NumberOfScenarios parameter. The number of test cases determines the number of iterations to run.
sigEditBlk = [mdl '/Input']; numCases = str2double(get_param(sigEditBlk,'NumberOfScenarios'));
Create an array of Simulink.SimulationInput objects to define the set of simulations to run. Each SimulationInput object corresponds to one simulation and is stored in array simIn. For each simulation, set these parameters:
- ActiveScenario to indicate which scenario of the Signal Editor block to execute
- CovEnable to turn on coverage analysis
- CovSaveSingleToWokspaceVar to save the coverage results to a workspace variable
- CovSaveName to specify the name of the variable.
for idx = numCases:-1:1 simIn(idx) = Simulink.SimulationInput(mdl); simIn(idx) = setBlockParameter(simIn(idx), sigEditBlk, 'ActiveScenario', idx); simIn(idx) = setModelParameter(simIn(idx), 'CovEnable', 'on'); simIn(idx) = setModelParameter(simIn(idx), 'CovSaveSingleToWorkspaceVar', 'on'); simIn(idx) = setModelParameter(simIn(idx), 'CovSaveName', 'covdata'); end
Run Simulations in Parallel by Using Parsim
Use the parsim function to execute the simulations in parallel. Pass the array of SimulationInput objects, simIn, into the parsim function as the first argument. Set the ShowProgress option to on to display the progress of the simulations in the MATLAB Command Window. The output from the parsim command is simOut, |an array of |Simulink.SimulationOutput objects.
simOut = parsim(simIn, 'ShowProgress', 'on');
[10-Aug-2018 21:59:04] Checking for availability of parallel pool... Starting parallel pool (parpool) using the 'local' profile ... connected to 12 workers. [10-Aug-2018 22:01:33] Loading Simulink on parallel workers... [10-Aug-2018 22:01:54] Configuring simulation cache folder on parallel workers... [10-Aug-2018 22:01:55] Loading model on parallel workers... [10-Aug-2018 22:02:06] Running simulations... [10-Aug-2018 22:03:00] Completed 1 of 2 simulation runs [10-Aug-2018 22:03:00] Completed 2 of 2 simulation runs [10-Aug-2018 22:03:00] Cleaning up parallel workers...
Each Simulink.SimulationOutput object contains logged coverage results stored as cvdata objects. These coverage results are stored in a field named covdata, as previously specified by the CovSaveName parameter. Using parsim to run multiple simulations means that errors are captured so that subsequent simulations can continue to run. Any errors are recorded in the ErrorMessage property of the SimulationOutput object.
covdata references a file containing the coverage results. The coverage data from the referenced file is automatically loaded into memory when covdata is used by a coverage function.
ans = ... cvdata file: C:\TEMP\Bdoc18b_931964_4676\ib036F93\10\tp56b7f985\slcov_output\slvnvdemo_powerwindow_parsim\slvnvdemo_powerwindow_parsim_cvdata_1.cvt date: 10-Aug-2018 22:02:59
Compute Cumulative Coverage
Obtain the coverage data from each element of simOut and cumulate the results.
coverageData = simOut(1).covdata; for i = 2 : numCases coverageData = coverageData + simOut(i).covdata; end
View the cumulative coverage results on the model by using coverage highlighting.
Generate a cumulative coverage report.
Close the model if the model was opened by the example script.
if(~isModelOpen) close_system(mdl, 0); end