There is a lot specialazed uC for DSP tasks - for example ADSP-21XX , TMS320C54XX. They calculate simple FIR filter with single instruction. Other big advantage of these DSP chips is that they have very good overflow handling algorithm and conseqently is "dificult" to lose information. Just look at the "ADSP218X instruction set reference" from analog.com ( chapter 3- software examples))
and read the instructions for syntax. "Remez" is an equiripple FIR. And at the end of this help file there are other filter types to choose from. look for the:
"Also see FIRLS,......."
line at the bottom.
You can also see the code for each file by typing (at the Matlab command line)
"edit remez", and study the methodology