Traditional design methodologies of a digital system assumes the worst case operating conditions to tolerate the physical and environmental variations, which ensure the system operates correctly and conservatively. The clock frequency of the processor is generally set to operate below the maximum permissible operating frequency which achieves less than the maximum performance gains. However, in-fact higher performance gain can be achieved by dynamic overclocking mechanism, which tunes the clock rate beyond the worst case assumptions. Dynamic overclocking mechanism may eventually lead to rise in temperature if it is adopted for a longer duration of time hence this problem is circumvented by using throttling mechanism at appropriate instances. Further, the throttling technique can be used to conserve energy when the system is with low work load. The aim of the work is to exploit the techniques of overclocking and throttling to enhance the performance, also to achieve optimal utilization of processor resources and at the same time maintaining the system reliability. The goal is hence achieved by employing an Adaptive Neuro Fuzzy Inference System to self-tune its frequency based on work load characteristics and also adapting itself to environmental variations.

Overclocking Throttling Energy