MathWorks fully supports running 32-bit MATLAB on 64-bit Windows as of MATLAB R2010a. Even prior to R2010a this should still work, albeit with some limitations.
There are a handful of known issues with running 32-bit MATLAB on 64-bit Windows:
- MATLAB Parallel Server may fail when both 32-bit MATLAB and 64-bit MATLAB are installed on the same computer.
- Real-Time Windows Target will not run.
- Data Acquisition Toolbox cannot connect to parallel ports.
- Embedded IDE Link and Target Support Package may fail to start Code Composer Studio.
Similarly, cross-platform compilation is not supported: binaries compiled on a 32-bit system will only run in 32-bit mode, even on 64-bit operating systems. Binaries compiled on 64-bit operating systems will not run on 32-bit operating systems.
If you must run 32-bit R2009a or earlier on 64-bit Windows, be aware that you will encounter issues when invoking external C compilers and other third party applications. For this reason, MathWorks does not recommend running 32-bit R2009a or earlier on 64-bit Windows.
Lastly, MATLAB R2015b was the last release of MATLAB with a 32-bit version. MATLAB R2016a and later are 64-bit only.
To install 32-bit MATLAB on a 64-bit Linux machine, start the installer with the -glnx86 flag:
The -glnx86 flag is also required when starting 32-bit MATLAB, and when using 32-bit network license manager scripts (lmstart, lmdown, lmstat, etc.).
Some users have problems loading the JRE when running in this configuration.
Creating a symbolic link named "glnxa64" in $MATLAB/sys/java/jre that points to "glnx86" should resolve this issue.
You can do this with the following command:
.ln -s $MATLAB/sys/java/jre/glnx86 $MATLAB/sys/java/jre/glnxa64
MacOS does not expose "bitness" to the user and will seamlessly run 32-bit and 64-bit applications side-by-side without issue on 64-bit systems. As of R2010b, MathWorks only builds a 64-bit version of MATLAB for MacOS.