Speeding Up MPSoC virtual platform simulation by Ultra Synchronization Checking Method

DATE(2011)

引用 3|浏览19
暂无评分
摘要
Virtual platform simulation is an essential technique for early-stage system-level design space exploration and embedded software development. In order to explore the hardware behavior and verify the embedded software, simulation speed and accuracy are the two most critical factors. However, given the increasing complexity of the Multi-Processor System-on-Chip (MPSoC) designs, even the state-of-the-art virtual platform simulation algorithms may suffer from the simulation speed issue. In this paper, we proposed an Ultra Synchronization Checking Method (USCM) for fast and robust virtual platform simulation. We devise a data dependency table (DDT) so that the memory access information by the hardware modules and software programs can be predicted and checked. By reducing the unnecessary synchronizations among simulation modules and utilizing the asynchronous discrete event simulation technique, we can significantly improve the virtual platform simulation speed. Our experimental results show that the proposed USCM can simulate a 32-processor SoC design in the speed of multimillion instructions per second. We also demonstrate that our method is less sensitive to the number of cores in the virtual platform simulation.
更多
查看译文
关键词
hardware behavior,robust virtual platform simulation,software program,ultra synchronization checking method,mpsoc virtual platform simulation,microprocessor chips,soc,data dependency table,embedded software development,system-on-chip,early-stage system-level design space exploration,multiprocessing systems,virtual platform simulation,multimillion instruction,memory access information,embedded systems,synchronization,multiprocessor system-on-chip design,asynchronous discrete event simulation,system on chip,hardware,computational modeling,computer model,discrete event simulation,embedded software,system level design,accuracy,data models,system on a chip,data model
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要