audio-mobster wrote: Description of denormals etc., and how they're represented in memory: http://en.wikipedia.org/wiki/IEEE_754#Single-precision_32_bit As for tackling denormal issues: http://music.columbia.edu/pipermail/linux-audio-dev/2003-August/004581.html http://music.columbia.edu/pipermail/linux-audio-dev/2004-June/008739.html The FLUSH_TO_ZERO(x) macro can be seen used in TAP's tap_utils.h The first post above suggest that it sometimes may not work due to compiler optimisations, so the second post suggests a workaround similar to what is used by CAPS. I hope this is useful. -- Steven Chamberlain steven(a)pyro.eu.org