coder.profile.test.runTests
Automate execution-time or stack usage analysis of code generated from Simulink models or MATLAB functions
Since R2024a
Syntax
Description
resultsSet = coder.profile.test.runTests(testComponent)testComponent, by using inputs provided by the
model. The function runs in two stages. First, the function runs the tests with only task
profiling enabled. Then, the function runs the most demanding tests again with detailed
function profiling enabled.
In the first stage, runTests captures the execution times of tasks
for the given inputs. In the second stage, runTests identifies how code
is executed. For example, it generates a visualization of the function-call stack.
runTests does not run detailed profiling in the first stage because
instrumentation overhead can be significant and invalidate the acquired time
measurements.
The function stores execution-time results in the resultsSet object.
Use the Code Profile
Analyzer app to analyze the results.
For information about importing test cases into a model, see Overview of Signal Loading Techniques and Load Input Data for Basic Test Cases.
resultsSet = coder.profile.test.runTests(testComponent,CodeMetric="stack")resultsSet
object. For stack usage analysis, there is only one stage because instrumentation does not
affect the results. Use the Code Profile Analyzer app to
analyze the results.
resultsSet = coder.profile.test.runTests(testComponent,TestFile="myTestFile")myTestFile. As explained previously, the
function runs in two stages to reduce the impact of instrumentation overhead.
If the test cases cover system requirements and exercise as many as possible code paths, you can use the function and the Code Profile Analyzer app to identify and analyze performance hotspots in the generated code.
resultsSet = coder.profile.test.runTests(testComponent,MATLABConfig=objConfig,MATLABInputs=inputSource)testComponentinputSource
objConfig is a MATLAB configuration object for code generation.
[
runs two SIL or PIL tests using resultsSet secondResultsSet] = coder.profile.test.runTests(___,OverrideSettings=keyValueMapping)keyValueMapping, a
Map object that maps model configuration parameters to pairs of values.
For the first test, the function uses the first value of each pair. For the second test, the
function uses the second value of each pair. The function returns two sets of results,
resultsSet and secondResultsSet.
resultsSet = coder.profile.test.runTests(___,Name=Value)
Examples
You can use coder.profile.test.runTests and the
Code Profile Analyzer app to identify and analyze performance
hotspots in generated code.
For an example, see Identify Hotspots in Generated Code.
This example shows how you can perform execution-time and stack usage profiling of code generated from a MATLAB function.
Open the example that contains the required files.
openExample('ecoder/KalmanFilterExample')Load the inputs.
load position.mat numPts = 300; inputs = cell(1, numPts); for idx = 1: numPts inputs{idx} = {position(:,idx)}; end
Create a MATLAB configuration object.
config = coder.config('lib'); config.VerificationMode = 'SIL'; config.EnableOpenMP = false;
Run tests with execution-time and stack usage profiling enabled.
myFcn = 'kalman01'; resTime = coder.profile.test.runTests(myFcn, ... MATLABConfig=config, ... MATLABInputs=inputs, ... NumIterations=5); resStack = coder.profile.test.runTests(myFcn, ... CodeMetric="stack", ... MATLABConfig=config, ... MATLABInputs=inputs);
To view execution-time results in the Code Profile Analyzer app, run:
coder.profile.show(resTime);
The Cumulative Results tab displays the Task Summary and Test Details views. When you select a function in the Task Summary view, the Test Details view lists the SIL tests performed in order of execution time.
For more information about analyzing execution-time results, see Automate Identification of Hotspots in Generated Code.
To view stack usage results in the Code Profile Analyzer app, run:
coder.profile.show(resStack);
The Cumulative Results tab displays the Task
Summary and Test Details views. If you select
kalman01 in the Task Summary view, the
Test Details view shows constant stack usage for the SIL tests.
This example shows how you can use key-value mapping to loop over pairs of configuration parameter options.
Open the model SILTopModel.
openExample('ecoder/SILPILVerificationExample', ... supportingFile='SILTopModel.slx') model = bdroot;
Disable Simulink Coverage™ and third-party code coverage analysis.
set_param(model,'CovEnable','off'); covSettings = get_param(model,'CodeCoverageSettings'); covSettings.CoverageTool = 'None'; set_param(model,'CodeCoverageSettings',covSettings);
Use key-value mapping to specify loop-over values for the
BuildConfiguration configuration parameter.
params = containers.Map('KeyType','char', 'ValueType','any'); params("BuildConfiguration") = {"Faster Builds", "Faster Runs"};
Run SIL tests with execution-time profiling enabled.
[resTime1, resTime2] = coder.profile.test.runTests(model, ...
OverrideSettings=params);To compare execution-time results, use the Code Profile Analyzer app. Display the first set of results.
coder.profile.show(resTime1)
On the Time Profiling tab, select the
Comparison tab. From the Select a profile
variable list, choose resTime2, which contains
the second set of results. Then, click Compare.
The values in the Longest Execution Time (%) and
Average Execution Time (%) cells indicate that, when
BuildConfiguration is set to "Faster Runs", the
longest step execution time is slightly longer but the average step execution time is
significantly reduced.
Specify loop-over values for an additional configuration parameter.
params("MaxStackSize") = {"64","256"};
Run SIL tests with stack usage profiling enabled.
[resStack1, resStack2] = coder.profile.test.runTests(model, ... CodeMetric="stack", ... OverrideSettings=params);
Display the first set of results.
coder.profile.show(resStack1)
On the Memory Profiling tab, select the
Comparison tab. From the Select a profile
variable list, choose resStack2, which contains
the second set of results. Then, click Compare.
The display shows values of zero in the Highest Stack Usage (bytes)
(%) and Average Stack Usage (bytes) (%) cells. In this
example, changing the configuration parameter combination from
BuildConfiguration="Faster Builds" and
MaxStackSize="64" to BuildConfiguration="Faster
Runs" and MaxStackSize="256" does not affect stack usage
metrics for the generated code.
This example shows how you can use a coder.profile.test.TestHook object to customize automated execution-time
profiling.
Open the model SILTopModel.
openExample('ecoder/SILPILVerificationExample', ... supportingFile='SILTopModel.slx') model = bdroot;
Disable Simulink Coverage and third-party code coverage analysis.
set_param(model,'CovEnable','off'); covSettings = get_param(model,'CodeCoverageSettings'); covSettings.CoverageTool = 'None'; set_param(model,'CodeCoverageSettings',covSettings);
In your working folder, provide MATLAB code that specifies a subclass of
coder.profile.test.TestHook and defines functionality for hooks.
For example, myTestHook.m.
classdef myTestHook < coder.profile.test.TestHook methods(Access = public) function TestSetup(this) disp("-------------> PRETESTS_START_HOOK"); end function TestStart(this, modelName, isTaskProfilingStage) disp("-------------> PRETEST_RUN_HOOK"); end function TestEnd(this, modelName, isTaskProfilingStage) disp("-------------> POSTTEST_RUN_HOOK"); end function TestCleanup(this, aggregatedResults) disp("-------------> POSTTESTS_PREEXIT_HOOK"); aggregatedResults.addDetails("Item1","Description for Item1"); aggregatedResults.addDetails("Item2","Description for Item2"); end end end
Run coder.profile.test.runTests, which invokes the test hooks.
resTime = coder.profile.test.runTests(model, ... "TestHooks", myTestHook());
-------------> PRETESTS_START_HOOK -------------> PRETEST_RUN_HOOK ### Searching for referenced models in model 'SILTopModel'. ### Total of 1 models to build. ### Starting build procedure for: SILTopModel ### Successful completion of build procedure for: SILTopModel Build Summary Top model targets: Model Build Reason Status Build Duration ========================================================================================== SILTopModel Generated code was out of date. Code generated and compiled. 0h 0m 9.7842s 1 of 1 models built (0 models already up to date) Build duration: 0h 0m 10.145s ### Preparing to start SIL simulation ... Building with 'Microsoft Visual C++ 2022 (C)'. MEX completed successfully. ### Starting SIL simulation for component: SILTopModel ### Application stopped ### Stopping SIL simulation for component: SILTopModel -------------> POSTTEST_RUN_HOOK -------------> PRETEST_RUN_HOOK ### Searching for referenced models in model 'SILTopModel'. ### Total of 1 models to build. ### Starting build procedure for: SILTopModel ### Successful completion of build procedure for: SILTopModel Build Summary Top model targets: Model Build Reason Status Build Duration ========================================================================================== SILTopModel Generated code was out of date. Code generated and compiled. 0h 0m 10.157s 1 of 1 models built (0 models already up to date) Build duration: 0h 0m 10.448s ### Preparing to start SIL simulation ... Building with 'Microsoft Visual C++ 2022 (C)'. MEX completed successfully. ### Starting SIL simulation for component: SILTopModel ### Application stopped ### Stopping SIL simulation for component: SILTopModel -------------> POSTTEST_RUN_HOOK -------------> POSTTESTS_PREEXIT_HOOK
The build output contains hook messages.
To view results in the Code Profile Analyzer app, run:
coder.profile.show(resTime)
On the Cumulative Results tab, the Test Execution
Information section displays the information that you provided through the
addDetails function.
In the Results section of the toolstrip, click Create
PDF. The PDF report displays the information that you supplied through the
addDetails function.
