基於基因演算法應用於異質性網路單晶片之任務排程方法

Abstract

網路單晶片是為了應付未來極為複雜的系統單晶片的通訊需求所提出的一種新的設計方式，在這篇論文中，我們提出一個基於基因演算法的任務排程方法把應用排程至一個使用異質性網路單晶片，這個任務排程方法試著去為每一個任務找到最適合的處理器，使得系統的資料處理率提升至最大。在基因演算法中，我們考慮到應用中的特性，而提出了一個新的交配運算元，藉此提升基因演算法的效能，實驗結果顯示了我們所提出的交配運算元的效能較傳統的還要好上平均10%，而在基因演算法的運算時間方面也較使用傳統交配運算元還要快。

Network-on-Chip is a new design paradigm to meet the communication requirement of future billion-transistor System-on-Chip. In this thesis, we propose a genetic algorithm (GA) based task scheduling technique to schedule the applications to the heterogeneous Network-on-Chip. The task scheduling process attempts to arrange the allocation of processor for each task such that the system throughput is maximized. As well, a new mating operator of GA is also proposed to improve the performance of traditional GA by considering the characteristics of application. The experimental results show that proposed mating operator not only outperforms traditional ones by 10% averagely, but also requires less computation time.