多处理器片上系统中一种结合二阶导数的温度预测模型

摘要

区域温度预测是多处理器片上系统（MultiProcessor System-on-Chips，MPSoCs）高效散热的基础．本文以RC 热传导（Thermal Resistance and Capacitance，Thermal RC）模型为基础，结合二阶导数提出了一种温度预测模型．该模型不仅可以在较低的运算复杂度下准确预测温度，而且能在固定的预测误差率范围内拓宽预测时间长度，进而减少模型在实际运行中被调用的次数，降低额外功耗．实验结果表明，相比现有的一次导数预测模型，在相同可接受误差率范围内，该模型能将预测时长拓宽至对比模型的1．6倍．同时，当预测时长拓展至2．5s时，该模型的预测准确率比对比模型高3．84％．%The regional temperature prediction is the basis of the efficient heat dissipation in multiprocessor system-on-chips (MPSoCs).Based on the thermal resistance and capacitance (Thermal RC)model,this paper proposed a predic-tive thermal model combined with the second derivative.It predicts the temperature accurately with low complexity,and in-creases the prediction time length within a certain prediction error range to reduce the number of times that the prediction module is invoked and the extra power consumption.Experimental results show that,compared to the existing model com-bined with the first derivative,the proposed model increases the prediction length by 1. 6 times within the same acceptable prediction error range.When the prediction time length is increased to 2. 5 s,the prediction accuracy of the proposed model is 3. 84% higher than that of the contrastive model.