xpcbench - xPC Target benchmark

Synopsis

MATLAB command line

xpcbench(model)
xpcbench(model,P1)
xpcbench(model,P1,P2)
xpcbench(model,P1,P2,P3)

Arguments

Argument	Value	Description
`model`	`'minimal'`	Benchmark the default minimal model.
	`'f14'`	Benchmark the default f14 model (one f14 system).
	`'f14*5'`	Benchmark the default f14*5 model (five copies of the f14 system).
	`'f14*10'`	Benchmark the default f14*10 model (ten copies of the f14 system).
	`'f14*25'`	Benchmark the default f14*25 model (25 copies of the f14 system).
	`'this'`	Benchmark all five default models (the default minimal model plus all four default f14 models).
	`'usermdl'`	Benchmark your model, usermdl.mdl.
Up to three optional arguments: `P1`, `P2`, and `P3`	`'-reboot'`	Reboot the target computer after testing each model.
	`'-cleanup'`	Delete build files after running benchmarks.
	`'-verbose'`	Plot and display results in the MATLAB Command Window.

Description

xpcbench benchmarks the real-time execution performance of xPC Target applications on your target computer and compares the result with prestored benchmark results from other computers.

The prestored target computer benchmark results of five models (applications) are provided, each model compiled using a sampling of the currently supported compilers (see http://www.mathworks.com/support/compilers/current_release/). Compare these results with those for your target computer. Results are labeled by CPU type, CPU clock rate, and the compiler used to compile the application.

The five benchmark models are:

Benchmark	Description
minimal	Based on a minimal model consisting of just three blocks (Constant, Gain, Termination). This model has neither continuous nor discrete states. The result of this benchmark gives an impression about the target computer interrupt latencies.
f14	Based on the standard Simulink example model f14. Type `f14` in the MATLAB Command Window to open the model and view the model (62 blocks, 10 continuous states).
f14*5	Five f14 systems modeled in subsystems (310 blocks, 50 continuous states). This benchmark is five times more demanding than benchmark F14.
f14*10	Ten f14 systems (620 blocks, 100 continuous states).
f14*25	25 f14 systems (1550 blocks, 250 continuous states).

xpcbench without an argument displays two plot figures, each containing different representations of the prestored target computer benchmark results. The first plot lists, for each target computer tested, the smallest achievable sample time for the five benchmarks, in microseconds. The second plot contains a bar graph of all computers, ranked by relative performance.

Note

The prestored benchmark results were collected with Multicore CPU support disabled. This allows for direct comparison with previous release results.
A sample time is achievable if any smaller sample time causes CPU overload.

xpcbench(model) benchmarks your target computer using argument model. You can specify:

All five default benchmarks ('this')
One of the five default benchmarks ('minimal', 'f14', 'f14*5', 'f14*10', 'f14*25')
Your model ('usermdl')

Before you run xpcbench, you must connect the host machine to the target computer and run the xPC Target test, xpctest, with no failures.

Benchmark execution can take several minutes to complete, including:

Generating the benchmark models
Building and downloading the xPC Target applications
Searching for the smallest achievable sample time
Displaying the performance results in the MATLAB Command Window, along with the prestored results for the other target computers

res = xpcbench returns the prestored benchmark results in a structure array with fields:

Field Name	Contents
`Machine`	Target computer information string containing CPU type, CPU speed, compiler
`BenchResults`	Target computer benchmark performance for the five default models 'minimal', 'f14', 'f145', 'f1410', 'f14*25'
`Desc`	Target computer descriptor string containing machine type, RAM size, cache size

res = xpcbench(model) returns the benchmark results for model in a structure with fields:

Field Name	Contents
`Name`	Name of `model`
`nBlocks`	Number of blocks in `model`
`BuildTime`	Elapsed time in seconds to build `model`
`BenchTime`	Elapsed time in seconds to run benchmark for `model`
`Tsmin`	Minimal achievable sample time in seconds for `model`

Examples

xpcbench

Prestored benchmark results showing what to expect of representative processors.

Boot the target computer using your chosen method.

Connect it to the host computer.

xpctest

res = xpcbench;
res(1)

ans = 

         Machine: 'Intel Celeron M  600MHz (VisualC10.0)'
    BenchResults: [1.0000e-05 1.4000e-05 2.5000e-05 
                  4.4000e-05 1.0800e-04]
            Desc: [1x70 char]

The array contains benchmark results for each processor type, in arbitrary order.

xpcbench

xpcbench('this')

Benchmark using the five default models.

Boot the target computer using your chosen method.

Connect it to the host computer.

xpctest

res = xpcbench('this');
res(1)

ans = 

         Name: 'Minimal'
      nBlocks: 3
    BuildTime: 28.2558
    BenchTime: 23.3845
        Tsmin: 8.3770e-06

xpcbench('this')

xpcbench('f14*5')

Benchmark using model f14*5 only.

Boot the target computer using your chosen method.

Connect it to the host computer.

xpctest

res = xpcbench('f14*5')
res

res = 

         Name: 'F14*5'
      nBlocks: 310
    BuildTime: 32.8377
    BenchTime: 29.7613
        Tsmin: 1.3148e-05

xpcbench('f14*5')

Benchmark results for model:               F14*5
Number of blocks in model:                 310
Elapsed time for model build (sec):        38.3
Elapsed time for model benchmark (sec):    29.7
Minimal achievable sample time (microsec): 14.1

xpcbench('this','-reboot')

Benchmark the target computer using the five default models, rebooting after each test.

Boot the target computer using your chosen method.

Connect it to the host computer.

xpctest

xpcbench('this','-reboot')

xpcbench('xpcosc','-cleanup','-verbose');

Benchmark the target computer using model xpcosc.mdl, delete the build files when done, and display full results in the MATLAB Command Window.

Boot the target computer using your chosen method.

Connect it to the host computer.

xpctest

xpcbench('xpcosc','-cleanup','-verbose');

Benchmark results for model:               xpcosc
Number of blocks in model:                 287
Elapsed time for model build (sec):        28.7
Elapsed time for model benchmark (sec):    26.1
Minimal achievable sample time (microsec): 5.5

res = xpcbench('this','-reboot','-cleanup','-verbose');

Benchmark the target computer using the five default models, reboot after each test, store the results in res, delete the build files when done, and display full results in the MATLAB Command Window.

Boot the target computer using your chosen method.

Connect it to the host computer.

xpctest

res = xpcbench('this','-reboot','-cleanup','-verbose');

res(1)

ans = 

         Name: 'Minimal'
      nBlocks: 3
    BuildTime: 51.2185
    BenchTime: 23.4161
        Tsmin: 4.9727e-06

Search MATLAB Documentation
R2012a Documentation → xPC Target
View documentation for other releases	Learn more about xPC Target

xpcbench - xPC Target benchmark

Synopsis

Arguments

Description

Examples

xpcbench

xpcbench('this')

xpcbench('f14*5')

xpcbench('this','-reboot')

xpcbench('xpcosc','-cleanup','-verbose');

res = xpcbench('this','-reboot','-cleanup','-verbose');

See Also

Related Products & Applications