The fast Fourier transform (f.f.t.) has found wide applications in areas requiring the processing of uniformly sampled signals. This paper investigates the characteristics of using a standard, inexpensive micro-computer system to implement an f.f.t. processor operating in either real-time or online mode. Various aspects on the maximum sampling rate in either mode of operation, such as the limitation by the system input/output time and f.f.t. computation time, as well as the aliasing-free condition to process bandlimited periodic signals, are revealed. The sampling clock is made adaptive to the real-time signal. So the processor is capable of processing signals from d.c. up to the online aliasing frequency. An example of real-time/online line-voltage fault detection is also included.