Design And Optimization Of A Network-On-Chip Simulator On GPU Platform

Network-on-Chip (NoC) solves the problem of low throughput and augmentability within traditional bus-structured processors. At present, there are many network simulators on the market, but these tools are focused on some research directions. Researchers have had to use a variety of different tools, or to develop a new simulation tool, which affects the efficiency of the their work.This paper design and realize a NoC simulator based on GPU platform. In order to facilitate the operation of the data and reduce the possibility of using the GPU local memory, all the elements involving computing are presented by arrays. This paper also analyzed the characteristics of the GPU and the GPU programming model, and designed the simulator GNSIM on the GPU platform. GNSIM can use GPU’s powerful parallel ability to improve the performance, and shorten the simulator’s running time.In order to evaluate the performance,GNSIM was compared with a efficient simulator, MABSIM, with the same configurations. We compare their execution time,and arrived at a conclusion that running speed of our simulator is faster than that of the MABSIM. At last, the performance of GNSIM is optimized, and the running speed of the system is further.