I am assuming you are using MATLAB code to HDL or MATLAB Function Block in Simulink. Can you share your sample model?
Usually fimath is used during a creation of fi object. The fi function with fimath is supported by HDL Coder. fimath helps define rounding and saturation behavior during fixed-point math operations. Please note default fimath object settings put you in nearest rounding mode and with overflow behavior as saturate. Both generate extra hardware. You might want to consider use of hdlfimath function (which recommends use of hardware friendly floor and wrap settings).
Browse through the code in this folder which shows the use of fimath and hdlfimath operations.
>>mlhdlc_demo_setup("fft")
% check out the examples fft40 and fft_channelizer which show use of "fi"
% and "fimath" functions.