While "switching times" within a photonic transistor are usually  
    calculated as the time it takes light to go from the beam combining 
    optics to the mask, the fact is, that actual switching occurs much  
    faster.  Information is actually processed AT THE SPEED OF LIGHT. 

     The actual switching occurs on a photon for photon basis.  That is, 
    photons that cause the destructive interference at the hole do so by 
    "interfering" with like photons that arrive at the same time from 
    the hologram which is always on.  Therefore, even a photonic NOT 
    that is several nanoseconds long (several feet in length) will be 
    able to NOT an input beam that is pulsating in the femtosecond 
    range. (Millionths of a nanosecond.)   

      In some applications that may not be helpful, but in digital 
    processors, for example, the fact becomes very useful. This is because 
    the signals in a processor operate on a matching basis for producing 
    such things as RAM read and write signals and the like. In the case 
    of information that must be processed at light speed, the information 
    must merely be pipelined into the photonic circuitry timed with the 
    information, with which it is being processed, so it will all show up 
    at the output on time.