This paper studies how the concept of global optimization can be modeled in granular computing. The basic mathematical theory is filled function method, one of the effective deterministic global optimization methods. Such filled function method is viewed as a new computing architecture that is focused on processing information granulation. Based on the modified concept of filled function, we propose a more practicable one-parameter filled function, then present a novel filled-function-based granulating algorithm with satisfactory numerical results. In the proposed algorithm, we attain a local minimizer by implementing a local search procedure, and use filled function to escape the current local minimizer to a lower local minimizer. By repeating these steps, we obtain a global minimizer. In the end, extension conceivable applications are given in order to evaluate the merits of this method.