Paul, > yes, my experiment with fft6 was mainly for use in the recursive calls of > SSA, > where is has to be compared to mul nxn->2n. But for that I have to get rid > of > the constraint that n is divisible by 2^11. Then it will be useful also for Niels's mul_2nm1, for the _2np1 part! Marco -- http://bodrato.it/