Currently libxm always works internally with float audio frames. Make this a compile-time option (s16, s32, float?) to allow compiling libxm without FPU use. Might yield faster and smaller code, but it's a lot of refactoring work.
Can use the same model as xm_sample_point_t (CMake option + _Generic()).
Currently libxm always works internally with float audio frames. Make this a compile-time option (s16, s32, float?) to allow compiling libxm without FPU use. Might yield faster and smaller code, but it's a lot of refactoring work.
Can use the same model as xm_sample_point_t (CMake option + _Generic()).