Specify floating point data types, such as single and half, in MATLAB and Simulink. Emulate target hardware behavior for denormal floating-point numbers, such as flush-to-zero, in simulation and code generation. Temporarily override data types in your model to compare your design against ideal floating-point behavior.