All VST SDK examples use the fast floating point calculation model. Is this model enough for real DSP algorithms? Which floating point calculation model are you use in your projects? Fast, precise or strict? Thanks.
years back I switched from precise to fast floating point model. This improved performance resp. reduced CPU consumption. I never had any of my DSP algorithms misbehaving or getting instabil (e.g. calculating coefficients). I believe this can have an impact on DSP computations but I never ran into a situation where it did. I never tried the ‘strict’ one though.