EDA Testbench Coverage Convergence via Random Variable Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for achieving high coverage point coverage in electronic circuit verification are time-consuming and require complex modifications to existing verification tools, as well as computationally expensive processes and extensive training data.
Innovation Solution
An improved testbench for EDA software tools that uses a data analysis engine to identify initial random variables corresponding to sampled coverage point solutions through a time-based association process, generating revised constraint parameters to produce focused random variables that systematically tune stimulus data to achieve higher coverage without requiring fundamental changes to existing configurations or extensive memory resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional methods are used to achieve high coverage point coverage, then coverage completeness is improved, but verification time and complexity increase significantly
Solution Approach 1:
The patent implements feedback by analyzing simulation results to identify uncovered coverage points, then automatically generating new random variables and stimulus data targeted at those specific gaps. This closed-loop approach uses coverage metrics to guide subsequent simulation iterations, efficiently directing verification efforts toward remaining uncovered areas rather than random exploration.
Solution Approach 2:
The patent changes parameters by dynamically adjusting random variable distributions and constraint parameters based on coverage analysis. When coverage gaps are identified, the system modifies the statistical parameters of random variable generation to bias toward stimulus patterns more likely to exercise uncovered functionality, thereby accelerating coverage convergence.
2Manufacturing precision
If manual analysis and test generation are used to achieve coverage convergence, then coverage completeness is improved, but device complexity and operational complexity increase
Solution Approach 1:
The verification system performs self-service by automatically analyzing its own simulation results, identifying coverage gaps, and generating targeted test stimuli without external intervention. The data analysis engine autonomously processes coverage metrics and directs subsequent verification efforts, eliminating the need for manual analysis and reducing operational complexity.
Solution Approach 2:
The patent introduces a data analysis engine as an intermediary between the simulator and testbench. This intermediary automatically processes simulation results, correlates random variables with coverage points, and generates refined constraint parameters, thereby automating the complex analysis task and reducing the complexity burden on users.
3Manufacturing precision
If extensive simulation runs are performed to achieve coverage goals, then coverage completeness is improved, but computational cost and memory resources increase
Solution Approach 1:
Instead of performing exhaustive simulation runs to achieve complete coverage, the patent applies partial action by targeting only the specific coverage gaps identified through analysis. The system generates stimulus data focused on uncovered areas rather than uniformly exploring all possible scenarios, reducing computational effort while maintaining coverage effectiveness.
Solution Approach 2:
The patent performs preliminary analysis of simulation results to identify coverage gaps before launching subsequent simulation runs. This preliminary action allows the system to pre-target specific uncovered functionality, avoiding wasteful computational expenditure on already-covered areas and reducing overall computational cost.
Data Source
AI summary
A data analysis engine is implemented in a testbench to improve coverage convergence during simulation of a device-under-validation (DUV). During a first simulation phase initial stimulus data is generated according to initial random variables based on user-provided constraint parameters. The data analysis engine then uses a time-based technique to match coverage variables sampled from simulation response data with corresponding initial random variables, determines a functional dependency (relationship) between the sampled coverage variables and corresponding initial random variables, then automatically generates revised constraint parameters based on the functional dependency. The revised constraint parameters are then used during a second simulation phase to generate focused random variables used to stimulate the DUV to reach additional coverage variables. In one embodiment, the functional dependency is determined by cross-correlating sampled coverage variables and corresponding initial random variables.