CAJ | 학술논문

针对片上网络有限的缓冲资源,提出了一种缓冲分配算法.该算法首先通过分析模型估算出路由器每个输入通道的负载大小,随后根据输入通道的负载分布情况采用遗传算法来实现缓冲资源的分配. 实验结果表明,在均匀随机流量下,与均匀分配算法和贪婪分配算法相比,新算法能获得更小的网络数据包延时,并可节省约33.3%的缓冲资源.
침대편상망락유한적완충자원,제출료일충완충분배산법.해산법수선통과분석모형고산출로유기매개수입통도적부재대소,수후근거수입통도적부재분포정황채용유전산법래실현완충자원적분배. 실험결과표명,재균균수궤류량하,여균균분배산법화탐람분배산법상비,신산법능획득경소적망락수거포연시,병가절성약33.3%적완충자원.
For the finite buffer resources in networks-on-chip, a buffer allocation algorithm is proposed. The proposed algorithm first estimates the load on each input channel in different routers across the chip by the performance analytical model. Then, it uses genetic algorithm to allocate the resources according to the distribution of the load on all the channels. Simulation results indicate that, under uniform traffic, this algorithm can get less average packet latency than those by uniform allocation and greedy allocation strategy. At the same time, about 33.3% buffer can be saved.