Hardware design verification system and method based on reinforcement learning

Abstract

The invention relates to the technical field of reinforcement learning and the technical field of chip verification, in particular to a hardware design verification system and method based on reinforcement learning. The core of the invention is in that the hardware design verification platform excitation is automatically generated by using a reinforcement learning algorithm to enable the coveragerate to reach 100%, and the method and the system are suitable for convergence of the line coverage rate, the condition coverage rate, the state machine coverage rate and the function coverage rate. Amode of generating the verification platform excitation through reinforcement learning is used, and compared with a mode of manually constructing the excitation, automatic excitation generation is advantageous in that the verification work period is greatly shortened, and the labor cost is greatly reduced; and the number and types of generated excitations are more abundant, so that the verification completeness is greatly improved. Therefore, a verification worker can invest more energy on the improvement of the reference model and on a new project, and the chip verification process is greatly accelerated.

Description

technical field [0001] The invention relates to the field of reinforcement learning technology and the field of chip verification technology, in particular to a hardware design verification system and method based on reinforcement learning. Background technique [0002] Machine learning (Machine Learning, ML) is a multi-field interdisciplinary subject, specializing in the study of how computers simulate or implement human learning behaviors to acquire new knowledge or skills, and reorganize existing knowledge structures to continuously improve their performance. [0003] Reinforcement Learning (RL), also known as Reinforcement Learning and Evaluation Learning, is an important machine learning algorithm that has many applications in the fields of intelligent control robots and analysis and prediction. AlphaGo is the reinforcement learning algorithm used. Reinforcement learning is that the agent (Agent) learns in a "trial and error" way, and obtains rewards and knowing behavio...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problem that the incentive generation of the existing hardware design verification platform needs to be completely dependent on manual generation, and proposes a hardware design verification system and method based on reinforcement learning; the present invention combines reinforcement learning and hardware design verification , the verification platform incentives that need to be manually generated are generated through reinforcement learning, which greatly reduces labor costs, shortens the time for verification convergence, and greatly improves the reliability and completeness of verification

Images