Hardware implementation of genetic algorithm processor (GAP) is important for proven effectiveness as optimization engines for real-time solutions. To implement the robust GAP, it is significant to maintain the population diversity for sustaining the convergence capacity and preventing local optimum problem. In this reason, we propose a deterministic mutation method for providing the high population diversity to GAP. Experimental results with mathematical problems and pattern recognition show that the proposed method enhances the convergence capacity up to 34.5% and reduces computation power about 40% compared with the conventional method.