A Noise-Aware Dynamic Voltage-Frequency Adjustment Transition Sequence Design Method

Abstract

The invention discloses a noise-perceived dynamic voltage frequency adjustment transition sequence design method, and relates to noise analysis of an on-chip power distribution network and efficient analysis and programming of dynamic voltage frequency adjustment operation. By defining related problems as a hybrid 0 / 1 linear programming model and improving a branch and bound method, solving is conducted with the combination of a heuristic algorithm; for the problem that by means of the heuristic algorithm, a programmed result is difficult to evaluate, the linear planning method can accuratelyevaluate the distance between the current solution and an optimal solution by building and solving a dual problem, on the basis, the optimizing degree of the obtained solution is accurately evaluated,and the method has the capacity of obtaining the optimal solution through search; for the problem that the heuristic algorithm is prone to getting into local optimization and search stagnation is caused, the linear programming model is utilized, solving is conducted by using the branch and bound method, a correct search direction can be ensured, and the result which is obviously better than the result of the heuristic algorithm is quickly searched, and the method has the capacity of ensuring the optimal solution.

Description

technical field [0001] The invention relates to parameter extraction and noise analysis of on-chip power distribution network, efficient analysis and planning of gating operation in dynamic voltage frequency adjustment transition sequence, and is suitable for noise reduction of chip power distribution network and low power consumption design of mobile devices. Background technique [0002] In order to ensure the battery life of mobile devices, technologies such as dynamic voltage frequency adjustment dynamically adjust and allocate chip resources and work according to the needs of tasks, and have achieved remarkable results in the design of low-power systems, becoming a modern system chip design. The key basic technology in. In the process of dynamic voltage frequency adjustment, a large amount of current concentrates on the power supply network in a short period of time with the transition of the clock frequency, causing large-scale power supply noise in the power distribut...

AI Technical Summary

Problems solved by technology

For the target problem designed by the present invention, due to the large scale of the problem and the lack of sub-problem structure, it is difficult for algorithms such as dynamic programming to achieve a good efficiency improvement in solving this problem

For the heuristic search algorithm, the relatively random search process makes it difficult for designers to make a more accurate assessment of the quality of the obtained results, and this type of algorithm is prone to fall into local optimum problems, and its search results and Efficiency is difficult to meet design requirements

[0005] The random search algorithm proposed in Non-Patent Document 1 (W.Lam, et.al. "Clock scheduling for power supply noise suppression using genetic algorithm with selective gene therapy." Proc. ISQED, 2003.), due to the overall random search direction , so that its stability is low and the search efficiency is low, and it usually occurs that a long search time is allocated but a good search result matching it cannot be obtained

[0006] The model using the heuristic algorithm proposed in Non-Patent Document 2 (Y.Kim, et.al."Power Supply Noise Reduction by ClockScheduling with Gate-Level Current Waveform Estimation."Proc.InternationalSoC Design, 2008.) exists easily Falling into local optimum and unable to effectively evaluate the current solution, etc.

[0007] Non-patent literature 3 (H.Jiang, et.al. "Power gating scheduling for power / groundnoise reduction." Proc.DAC, 2008.) The proposed method of using triangular wave to approximate voltage drop and using genetic algorithm for noise optimization has precision Insufficient, difficult to evaluate the optimization results and other issues

Images