Using the Fast Fourier Transform algorithm or derivative, on a two-dimensional image or data set, incurs a process time of the order of 0(2N²log₂(N)). This can be reduced significantly by employing various reduction techniques, but the general form of the plot below remains. By contrast, the process time for the Optical Fourier Transform remains static.

The chart shows the process times for 2-D FFT calculation using single processors, based on the state of the art processor speeds of the Xilinx Virtex-4 series (500Hz), the Texas Instruments TMS320C67x series of DSP processors (1GHz). A theoretical plot based on the current maximum Intel processor speed (3.74GHz) is also included. Memory and logic requirements have been disregarded for the comparison.

As can be seen from the chart, as the resolution increases, process times increase rapidly for the electronic methods. To achieve workable, real-time operation would require several processors with huge amounts of logic elements (known as “butterflies”) operating in parallel (it should be noted that both TI and Xilinx do not currently offer 2-D FFT cores).