片上虫洞网络的缓冲区感知最坏情况时序分析
Buffer-aware Worst Case Timing Analysis of Wormhole Network On Chip
摘要：
本文提出了一种基于缓冲区感知的虫孔片上网络最坏情况时序分析方法，综合考虑缓冲区大小对直接阻塞流和间接阻塞流之间不同依赖关系的影响，进而分析其时序性能。首先，引入了直接和间接阻塞流集的更准确定义，以考虑缓冲区大小的影响。然后，基于网络微积分理论和新定义的阻塞流集，详细分析了虫孔片上网络的建模和最坏情况下的时序。这种引入的方法已经在一个现实的NoC案例研究中进行了说明，以显示延迟和缓冲区大小之间的权衡。我们提出的缓冲区感知的时序分析与传统的方法进行了比较分析和显着的增强，在最大延迟已被证明。
英文摘要：
A buffer-aware worst-case timing analysis of wormhole NoC is proposed in this paper to integrate the impact of buffer size on the different dependencies relationship between flows, i.e. direct and indirect blocking flows, and consequently the timing performance. First, more accurate definitions of direct and indirect blocking flows sets have been introduced to take into account the buffer size impact. Then, the modeling and worst-case timing analysis of wormhole NoC have been detailed, based on Network Calculus formalism and the newly defined blocking flows sets. This introduced approach has been illustrated in the case of a realistic NoC case study to show the trade off between latency and buffer size. The comparative analysis of our proposed Buffer-aware timing analysis with conventional approaches is conducted and noticeable enhancements in terms of maximum latency have been proved.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
