On Sun, Jan 23, 2005 at 10:26:30 +0100, Tim Blechmann wrote: I believe that my biquad based filters (dj_eq_1901, lcr_delay_1436, single_para_1203 and triple_para_1204) are denormal safe, if not I'd like to know about it as it can be fixed. If anyones looking for a quick project, a denormal tester* would be very useful. Small denormal problems are hard to detect in testing on Athlon and Pentium 3/M machines (all I use for devel and testing). * run a few seconds worth of sinewave into plugin, measure min/avg/max cpu cost, run 0's in, repeat measurement, compare - Steve