The Fast Fourier Transform (FFT) is a powerful method in contemporary computing, with lots of practical applications – from signal processing to cryptography. On multicore platforms, symmetrical load distribution prevails but has efficiency issues with locality creating a gap between theoretical arithmetic complexity and actual performance. Some proposed re-evaluations of Amdahl’s law for parallel speed-up favour asymmetric multi-processing. The approach taken in this work is therefore to let the individual processors/cores specialise in parts of the FFT such as butterfly operation, permuting/transposing, calculating complex roots of unity; thus computation is asymmetrical even on SMP/CMP. Inter-thread synchronisation employed is spin-lock. This research contributes: the notion of incrementality inherent in the application, innovative usage of a shared heap to hold twiddle-factors, and best known arithmetic complexity of computing these. The solution suits hard to predict problem sizes, on the up to 9 cores that the multi-core industry delivers nowadays (9 in IBM’s CellBE).