ML-Based Static Verification for Hardware Design Elements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional static verification tools for hardware designs fail to effectively reuse learning and settings across derivative design versions, leading to inefficient resource usage and potential design reliability issues due to manual management and underutilization of constraints, settings, and waivers.
Innovation Solution
A machine learning-based system that extracts feature sets from hardware descriptions, evaluates similarity indices between design elements, and updates parameters for static verification, allowing for the propagation of constraints, settings, and waivers across a design hierarchy, thereby enhancing the reuse of learning and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual management of verification settings is used for each derivative design version, then flexibility in addressing specific violations is maintained, but productivity decreases due to repetition of verification steps and loss of learned settings across versions
Solution Approach 1:
The verification system automatically identifies similar hardware elements across design versions and applies previously learned settings, constraints, and waivers without requiring manual intervention. The system serves itself by reusing verification knowledge across derivative designs, eliminating the need for users to manually manage each version's verification parameters.
Solution Approach 2:
The system creates a knowledge base that copies and stores verification settings, constraints, and waivers from verified design versions. When analyzing derivative designs, the system retrieves and applies identical settings from the knowledge base, effectively copying successful verification configurations across multiple design versions.
2Loss of time
If conventional static verification tools start from scratch for each derivative version, then complete verification coverage is ensured, but loss of time increases due to redundant analysis of unchanged components
Solution Approach 1:
The verification process is segmented into identifying similar hardware elements versus analyzing unique modifications. The system divides the verification task to focus only on changed components while reusing results for unchanged components, rather than treating each derivative design as entirely new.
Solution Approach 2:
The system performs preliminary identification of similar hardware elements before conducting full verification. By pre-classifying which components can reuse previous verification results and which require fresh analysis, the system avoids redundant verification of unchanged components while ensuring complete coverage where needed.
3Loss of information
If human resources manually manage verification settings for each design version, then adaptability to specific design requirements is maintained, but loss of information occurs when settings are not properly migrated between versions
Solution Approach 1:
The system continuously learns from verification results across different design versions and feeds this knowledge back into the verification process. Successful settings and constraints identified in one version are automatically fed forward to subsequent derivative versions, creating a cumulative knowledge base that prevents information loss.
Solution Approach 2:
The verification system serves multiple design versions simultaneously by maintaining a universal knowledge base that stores settings, constraints, and waivers applicable across all derivative designs. This universal repository eliminates the need for separate manual management of each version's settings.
Data Source
AI summary
Disclosed herein are system, computer-readable storage medium, and method embodiments of machine-learning (ML)-based static verification for derived hardware-design elements. A system including at least one processor may be configured to extract a feature set from a hardware description and evaluate a similarity index of a first hardware element with respect to a second hardware element, using an ML process based on the feature set, wherein the first hardware element is described in the hardware description. The at least one processor may be further configured to update one or more parameters corresponding to a static verification of the hardware description while the static verification is being performed, by providing at least one test attribute, corresponding to the second hardware element, applicable to the first hardware element, in response to determining that the similarity index is within a specified range, and additionally output a first result of the static verification.