论文部分内容阅读
实时光谱图像技术的应用是数据和通信的密集型算法的综合。因为片上系统(SoC)具有较大的灵活性及良好的性能,并可以直接采用IP核,基于片上网络(NoC)的SoC架构是实现这种实时系统最合适的解决方案。针对不同的应用算法,调整基于SoC的NoC架构是较为复杂的工作,因为设计空间探索和实验需要大量的时间。针对这些问题,提出了一种探索和评估SoC架构性能的仿真平台。该仿真平台基于NoC和参数化的评估模块,在FPGA上对SoC的通信结构进行仿真,并将时间性能和模拟结果进行对比。基于该仿真平台的系统评估主要用来考察用于油画鉴别应用的MPSoC的性能。
The use of real-time spectral image technology is a combination of intensive algorithms for data and communication. Because of the flexibility and good performance of system-on-chip (SoC), and the direct adoption of IP cores, SoC architectures based on a system-on-chip (NoC) are the most suitable solution for implementing this real-time system. Tuning SoC-based NoC architectures is a more complex task for different application algorithms, as designing space exploration and experimentation requires a significant amount of time. In response to these problems, a simulation platform to explore and evaluate the performance of SoC architecture is proposed. Based on NoC and parameterized evaluation module, the simulation platform simulates the communication structure of SoC on FPGA, and compares the time performance with simulation results. The system evaluation based on this simulation platform is mainly used to examine the performance of MPSoC used in oil painting identification applications.