Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

An Autodidact Verification System

IP.com Disclosure Number: IPCOM000247195D
Publication Date: 2016-Aug-15
Document File: 4 page(s) / 76K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is an autodidact verification system that utilizes the information in the Regression Data Base (RDB), generated over multiple regression cycles and/or over multiple design releases, to bridge the gap between the state-of-the-art and a truly efficient system.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 4

Title

An Autodidact Verification System

Abstract

Disclosed is an autodidact verification system that utilizes the information in the Regression Data Base (RDB), generated over multiple regression cycles and/or over multiple design releases, to bridge the gap between the state-of-the-art and a truly efficient system.

Problem

A verification system is meant to provide a high quality release in the least possible time. Verification of a Design Under Test (DUT) is done at various development stages. A DUT has a large number of configurations and allowable stimuli (practically
infinite). Verification environments include test cases, running infrastructure, and reports (regression system). Log files capture DUT configuration, the stimuli, and events triggered during the simulation.

A regression generates a huge volume of data. The data, generally, is used only to qualify the pass/fail and to debug failures. Otherwise, the data remains archived, unused, or destroyed forever. The data contains a wealth of hidden information such as the failure signatures, the associated connections with the configurations and the design versions, etc.

Time-to-market conditions allows the run of a limited number of test scenarios. The current state-of-the-art allows for constraint randomization (by machine) and directed testing (defined and limited by the imagination of verification and design engineer). The quality of the design improves with more constraint randomized regressions; however, becomes saturated. The approach is non-exhaustive, which leads to potential escapes.

A truly efficient verification system should derive the best possible combinations of stimuli and DUT configuration, that stress the design well enough to help identify any hidden issues in the least possible amount of time.

Solution/Novel Contribution

The proposed solution is an autodidact verification system that utilizes the information in the Regression Data Base (RDB), generated over multiple regression cycles and/or over multiple design releases, to bridge the gap between the state-of-the-art and a truly efficient system.

This solution includes:


• Regression Database (RDB)

• Data Analytics Engine (DAE)

• Design partitioning and configuration definitions

• Mapping of the partitions to the database

• Ability to map, re-map based on fail/pass signatures over regressions
• System usage to debug field failures in an efficient way


Page 02 of 4


• Efficient sign-off regression test suite generation

Method/Process

Terms used to describe the method and process follow.

Signature: A unique combination of test case configuration, stimuli and DUT response. A test case passing will result in a pass signature and one failing, a fail signature.

Regression Database (RDB): A database which contains simulation signatures accumulated over multiple regression...