An iteration method is proposed for sub-optimal output feedback computer control of decentralized singularly-perturbed systems. Computer control, decentralized output feedback control, and order-reduced control are three primary goals for this control scheme. The analog to digital transformation technique is used to convert continuous-time state models to discrete-time domain for the controllers which can be implemented by computer and be able to deal with today's complicated large scale system. Singular perturbation methods are applied to reduce the order of the system. After order reduction, it will help us to analysis the system; reduce the manufacturing process and cost. The contribution of this research provides a sub-optimal computer control algorithm for large-scale and high-dimension systems with strong potential for practical applications, such as power distribution and communication networks.