Compile MATLAB Code

Things under legendu.net/outdated are outdated technologies that the author does not plan to update any more. Please look for better alternatives.

1. MATLAB code can be compile to C/C++ code or stand alone applications. Usually you can run the stand-alone application in Windows system directly, but for Linux system it is a little headachy. The stand-alone application usually takes less than 50% of the time that the original MATLAB code takes to run. This depend on different situations. It can happen that the stand-alone application runs slower than the original MATLAB code.

2. To compile MATLAB code to c code, you can use mcc -mc mainfn sub1fn sub2fn, and to produce an stand alone executable file from c files, you can use mbuild mainfun sub1fn sub2fn.

3. To compile MATLAB code to c++ code and to create a corresponding stand-alone executable file you can use mcc -p mainfn sub1fn sub2fn. To compile MATLAB code that contains Handle Graphics functions into C++ and to create a corresponding stand-alone executable file, you can use mcc -B sglcppp mainfn sub1fn sub2fn.

4. To compile MATLAB code to c and to create corresponding stand-alone executable file, you can use mcc -m mainfn sub1fn sub2fn. To compile MATLAB code that contains Handle Graphics functions into c and to crate a corresponding stan-alone executable file, you can use mcc -B sgl mainfn sub1fn sub2fn.

5. If you use parallel computing, you will not be able to run the compiled application (at least before version 2010b) unless you have MATLAB Distributed Computing Server (MDCS) available on a computer cluster. If you have access to a MDCS, there is little necessary for you to compile your MATLAB parallel code to a stand-alone application, so just do not bother to compile your MATLAB parallel code.