Improving the Performance of a Supercomputer System Using Quality Engineering and Simulation

Abstract

The purpose of this paper is to describe a new approach in the design of robust computer systems using computer simulation experiments. The proposed approach is motivated by Taguchi's method and computer system engineering. The seven most important parameters were examined to determine the best configuration of a computer system. The results show that, based on this approach, a robust computer system can be designed such that, without further capital investment, the system will not only satisfy performance criteria, but also be less sensitive to uncontrollable factors. This study clearly illustrates that the combination of simulation and quality engineering is a feasible and efficient way to improve the performance of supercomputer systems.