Gpu Implementation Of The Vis In Circuit Simulation Algorithm

Circuit simulation is a common and important method in Electronic Design Automation (EDA).With the increasing complexity and tight time-to-marker schedules of today's digital, the traditional sequential simulation method could not satisfy the need of IC designers any more. In order to speed the simulation, lots of people have focused their eyes on the concept of parallel circuit simulation.In recent years, as the technology of Graphics Processing Unit (GPU) in hardware and software developing, more people have attached importance on the general purpose computation of GPU. It is becoming a hot topic for research.GPU general purpose computing based on CUDA is able to provide high memory band width and strong computing ability.VIS is a verification and synthesis system for finite-state hardware systems, which is being developed at Berkeley and Boulder. It improves upon first generation tools like HSIS and SMV. In this paper I analyze the source code corresponding to circuit simulation. The main contents are as follows:I describe what is VIS , what it can do , how to write limited Verilog code for its input and its commands. And then I describe the intermediate format file BLIF-MV. The development and characteristics of GPU are described. The hardware and software patterns of Computed Unified Device Architecture (CUDA) from NVIDIA are showed. By comparison with the traditional parallel programming, I described the great advantage of CUDA parallel programming.Describing the common circuit simulation method, event driven simulation algorithm and cycle driven simulation algorithm. There is also a description about the common parallel circuit simulation algorithm. Under the analysis of the VIS source code, I summarized the algorithm of VIS simulation code. And then I raised a parallel circuit simulation algorithm based on GPU acceleration.