A study by Reinhart and Herkel showed term predictions of daylight availability in architectural spaces should take the conditions of all individual time steps into account. However, most contemporary simulation tool algorithms require such long computation times that it is impractical or even unrealistic to simulation of the daylight penetration for each time step. This paper discusses two adaptations of a known algorithm, radiosity, that bring down the required computation time for an annual prediction from the order of days to the order of seconds.