There are various techniques to optimize .m functions. You would usually start by profiling the code execution.
Generally for higher performance than that you would use an (embedded) MATLAB Function block, for which you would need to adapt the code. You can seldom simply copy and paste into such a block, but the work is not necessarily a lot more than adding hints about the expected matrix sizes and types. However, not all routines can be accelerated with such a block. A lot depends on your code. It is difficult to answer without seeing your code.
