Font Size: a A A

Research On Vector Generation And Coverage Metrics In Design Verification Via Simulation

Posted on:2007-04-03Degree:DoctorType:Dissertation
Country:ChinaCandidate:W LuFull Text:PDF
GTID:1118360185495703Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
Design verification is the task of verifying that a given design, specified at whateverlevel, has certain properties required by a specification. Usually, design teams spend morethan 50% e?ort of the development on verification nowadays. By far simulation is stillthe primary workhorse for design verification and 40%-70% time spent in verification willconsumed during this process. While there are still some problems boring the verificationengineers during design verification via simulation. On one hand the quality of patternscan't be guaranteed often, and on the other hand coverage metrics taking observabilityinto account are not very satisfactory. Aiming at these problems, this dissertation coversthe key aspects of simulation: stimuli generation and coverage analysis, and the originalcontributions of this dissertation are as follows.1. New representations to model observability information are proposed.Observability coverage metric is also a kind of behavior coverage metric, regardinganother abstract model: observability, and it enhances the traditional coverage metricswithout taking observability into account. In this dissertation, we present precise and con-cise abstract representations from HDL (Hardware Description Language) descriptionsat RTL (Register Transfer Level) to model observability information. We use EPCT(Enhanced Process Controlling Trees) to represent the activation of codes in some timeframe and the statistics information about observability up to the current time frame. Weuse COC (Controlling-Observing Chains) to represent the observability information dur-ing time frames expansion. The proposed representations could denote the observabilityinformation clearly and based on these representations the analysis about observabilitycould be done easily.2. A novel observability evaluation method is proposed.The observability analysis procedure could be divided into two phases: simulationphase and analysis phase. In the first phase, we simulate a vector and construct theCOC of each OdP (Observed Point). In the second phase, we analyze the COC of each...
Keywords/Search Tags:Design Verification, Simulation, Vector Generation, Coverage Analy-sis
PDF Full Text Request
Related items