Application Optimization Of Random Walk Algorithm In Dynamic IC Power Network Analysis

As the IC manufacturing technique in sub-micron era,the traditional SPICE circuit simulator can not meet the design requirements of large-scale IC,so the random walk algorithm is widely regarded as a more effective design method.However,there are some limitations in the current acceleration strategy to solve the dynamic analysis of large-scale power network.In order to solve the shortcomings of the existing acceleration strategies,a random walk acceleration algorithm based on variable step-size is proposed.In the following part,the experiment results show that the variable step-size acceleration algorithm is more efficient than the traditional acceleration algorithm.Moreover,with the expansion of the power network,the running space of the program using random walk algorithm is also facing great challenges.This problem has not been solved.It improves the utilization ratio of the unit running space.Then the paper analyzes the running track of the random walk Algorithm and concludes that it has local characteristics.Based on this feature,this paper proposes a solution which can release the running space further.Then the power supply network of different scales is tested and the results demonstrated that the method is reasonable and effective.Then this paper focuses on noise problem of the power supply network.First of all,the power supply network switching noise analysis technology based on random walk algorithm is discussed,and the equivalent RLC power network model is embarked on,then,the similarity between RLC network model and random walk model is proved by derivation,and the random walk algorithm is used to carry out experimental analysis.Then,the experiment of applying variable step-size random walk algorithm to analyze the noise model of power network is carried on,and the result shows that the efficiency of the algorithm is improved effectively.Finally,the thermal analysis of IC chip is mainly discussed.Firstly,the thermal equivalent RC network model of IC power network is established,and the application technology of random walk algorithm for thermal equivalent model is studied,the experiments show that the random walk algorithm can analyze the thermal equivalent RC network.On the basis of this,an experiment is designed to apply variable step-size Random Walk Algorithm and run-space optimization technology to deal with large-scale thermal equivalent network model.The results show that the algorithm can improve the efficiency andsave the program running space.