| Summary: | 碩士 === 國立中山大學 === 資訊工程學系研究所 === 94 === The state assignment is one of the most important problems in hardware implementation of controllers (finite state machines, FSMs). Traditional heuristics include simulated annealing (SA), greedy approach, and recursive Min-Cut partitioning. Since these methods can’t reduce both area and power, thus we propose a new approach which using integer linear programming (ILP) to solve the state assignment. The proposed of ILP approach can set the weight and reach best solution between less area and low power. The approach can find out the best state assignment for both low-area and low power consumption. In addition, we also use ILP to solve the output encoding of controller in order to reduce the power consumption of datapath.
Finally, to verify the effectiveness of our proposed approach, we do some experiments on several MCNC FSM benchmarks and controller-datapath systems. The experimental results show that a significant power and area savings can be achieved.
|
|---|
