MATLAB Compiler 4.0 (R14) encrypts MATLAB code and wraps it in C. As a result, the source code is still interpreted at run-time, and performance of code should be the same whether run in stand-alone mode or in MATLAB. There is one important difference for compiled applications. Every time you start a compiled executable, you pay the penalty essentially of starting MATLAB. You should be aware of this when comparing performance variations. See the information below regarding MATLAB Compiler 3.0:
MATLAB Compiler 3.0 (R13) has the ability to translate MATLAB Code into C/C++ code. This code can then be compiled into a stand-alone application or C-shared library. The main purpose of the MATLAB Compiler is to provide deployment options for customer applications. It is true that in certain cases you may see a substantial speed increase in your applications after compilation. Though there can be instances where execution actually takes longer. We make no claims as to the speed increase of MATLAB code after compilation. Also, certain MATLAB file programming techniques can help execution speed of your MATLAB files.