MATLAB: How is Simulink 6.6 (R2007a) affected by the use of the new multithreading feature in MATLAB 7.4 (R2007a)

multi-coremulticoresimulink

With the new multithreaded BLAS option in R2007a, how does this affect Simulink simulation?

Best Answer

In general, Simulink is not affected by the use of multithreaded BLAS in MATLAB 7.4 (R2007a). The Simulink engine runs its own solver routines separate from the MATLAB environment. However some toolbox and blockset functions as well as user-written MATLAB Fcn and MATLAB file S-Function routines may use MATLAB commands in their algorithms that take advantage of the multithreaded BLAS, but a noticeable increase in speed is unlikely.

An exception to this is Simulink's implicit solvers, which make use of multithreaded BLAS to perform linear algebra calculations. For more information about implicit solvers please refer to the "Choosing a Solver" section of the Simulink documentation. This can be viewed by executing the following command in MATLAB:

web([docroot '/toolbox/simulink/ug/f11-69449.html'])

Related Solutions

MATLAB: How to limit the number of computational threads used by MATLAB in MATLAB 7.10 (R2010a)

Multithreaded computations, introduced in MATLAB 7.4 (R2007a), are turned on by default as of MATLAB 7.6 (R2008a) release. The followings are options to change the number of computational threads in MATLAB:

1. Setting the Number of Threads from Multithreading menu in Preferences Panel (available in R2007a through R2008b releases)
2. Using the MAXNUMCOMPTHREADS function (available in R2007b through R2010a). This function issues a warning as of R2009b release.
3. Start MATLAB with ‘-singleCompThread’ option
4. Setting CPU affinity for a MATLAB process at the operating system level

Note that the capability to adjust the number of computational threads in a single MATLAB session using Preferences Panel is no longer available in MATLAB 7.8 (R2009a) or later releases. Also, as of MATLAB 7.9 (R2009b), using MAXNUMCOMPTHREADS throws the following warning:

ERROR: maxNumCompThreads will be removed in a future release. Please remove any instances of this function from your code.

The primary reason for this change is that products that MATLAB is dependent upon have the ability to spawn threads from the currently-executing thread. This makes it infeasible to monitor and/or limit the number of computational threads at any given time in a MATLAB process.

There will be no direct replacement that allows users to specify the number of computational threads. Suggested methods to control threading levels in MATLAB are as below:

1. Set the '-singleCompThread' option when starting MATLAB to limit MATLAB to a single computational thread.
2. Specify a given set of CPUs (cores) that are available for use by the MATLAB process by setting CPU affinity provided by the operating system. This allows MATLAB to use more than one computational threads.

For more information on setting CPU affinity for a MATLAB process, refer to the Related Answers:

https://www.mathworks.com/matlabcentral/answers/94049-how-do-i-set-processor-affinity-for-matlab-7-7-r2008b-in-linux

and

https://www.mathworks.com/matlabcentral/answers/93867-how-can-i-permanently-set-the-processor-affinity-for-an-application-on-multi-cpu-machines-on-a-windo

MATLAB: How to read WMV (Windows Media) video files into MATLAB 7.4 (R2007a)

The ability to do read WMV files is not available in MATLAB 7.4 (R2007a).

To work around this issue, you can use the From Multimedia Block in the Video and Image Processing Blockset 2.3 (R2007a) and read in the video file through Simulink.

Best Answer

Related Solutions

MATLAB: How to limit the number of computational threads used by MATLAB in MATLAB 7.10 (R2010a)

MATLAB: How to read WMV (Windows Media) video files into MATLAB 7.4 (R2007a)

Related Question