Providing an infrastructure for assertion-based test generation and GPU accelerated mutation testing

Functional verification of modern digital designs is a never ending challenge in the Integrated Circuit (IC) industry. Fuelled by the continuous demand of more integration, the increased effort in verification does not always entail error-free circuits after first production. Emerging technologies such as Assertion-based verification, can help in verifying the functional correctness of digital designs and can be easily integrated into existing design verification methodologies. Simulation-based verification is still the most predominant method in industry because of its ability to scale with largedesigns. Assertions can be inserted into the design and they can be treated as coverage points, where the input tests are responsible for exerting the design's conditions in evaluating those assertions. The effectiveness of this approach relies on the quality of the tests, where poor test quality can prevent the design from being thoroughly verified.This thesis presents novel techniques and algorithms for generating tests from assertions. Assertions serve as an invaluable source of information, where one can leverage the defined behaviours for generating the appropriate functional tests that can be used in simulation. A proposed set of coverage metrics helps in generating tests that thoroughly evaluate assertions during simulation. Verification engineers can make use of these tests in performing effective simulation in order to detect and then correct any design errors. The tool developed for generating tests from assertions was evaluated using nearly 300 assertions that were written for verifying the correctness of several industry-based designs. As a result, the proposed test generation approach was able to provide additional tests which led to an improvement in coverage compared to assertion-based test generator developed by another research team. This thesis also developed novel algorithms for Graphics Processing Units and used for accelerating mutation-based simulations, which is a computationally intensive application. It was empirically shown for a set of 10 industry-based designs, that efficiently using the GPU's resources can drastically improve the simulation performance on the GPU, when compared to a commercial tool. The additional performance is a necessity, where maximal acceleration is needed for rigorously assessing test quality when simulating large quantities of mutations. This can have a positive impact in the quest for improving assertion quality, ultimately leading to an effective dynamic verification of digital designs.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.