Can Matlab Coder suport parallel computing toolbox

Hi, I want to develop a parallel application using the parallel computing toolbox. I am wondering if I can convert the application using Matlab Coder for C/C++ and MATLAB Builder JA (for Java language)?

Answer 1

[Ryan Livingston]

As of R2012b MATLAB Coder does offer support for PARFOR code generation with MEX targets only:

http://www.mathworks.com/help/coder/ug/acceleration-of-matlab-algorithms-using-parallel-for-loops-parfor.html

There are some requirements on the compiler:

http://www.mathworks.com/help/coder/ug/acceleration-of-matlab-algorithms-using-parallel-for-loops-parfor.html#btfitu2-9

Also, MATLAB Compiler and the Builder products support PCT in the general MATLAB paradigm:

http://www.mathworks.com/help/compiler/deploy-applications-created-using-parallel-computing-toolbox.html

Answer 2

[Walter Roberson]

No. See http://www.mathworks.com/help/coder/functions-supported-for-code-generation.html?s_tid=doc_12b

Basically if the functionality cannot be compiled into pure ISO/IEC 9899:1999 (known as C99) -- without POSIX.1 -- then MATLAB Coder probably cannot generate it. C99 has no parallel processing capacity.

POSIX.1, the Portable Operating System Interface standard, defines an optional extension, POSIX Threads. But that is not part of C itself, and MATLAB Coder does not know how to code for it. Also, the POSIX Threads model does not really match the programming model used for any of the Parallel Computing Toolbox facilities: MATLAB's facilities are closer to what is provided by POSIX.1b Message Passing, together with some mechanism for creating the additional processes (possibly even just fork())

That said, I do not know what the Simulink Coder with Embedded Coder extensions is able to do, but MATLAB Coder does not support PCT.

Answer 3

[Bill Chou]

If you are going to run the application in MATLAB, you can consider creating MEX files for your application, and then deploy them to different MATLAB Workers using Parallel Computing Toolbox. This will enable you to use as many cores on your machine as you allow.

Check out this webinar for more info on this workflow.