hi all,

i did some profiling on the PD_BADFLOAT macro ... i'm curious, if the second part, the filter for nans is really necessary ... if not, we would have a really nice branch-free macro...

which dsp functions _could_ create nans? if we could locate them, we would be able to use two macros, one that kills denormals and one that kills both denormals and nans ...

cheers... tim