Paper 2.2 – Tutorial – Debug Automation with AI

September 7, 2023 from 2:00 pm to 2:30 pm

  • Track 2
  • 階梯教室
Speaker: Craig Yang - Synopsys

Session: AI

Session Chair: YJ Chen, Realtek

Title: Debug Automation with AI

Language: English

Abstract—Regression analysis is a critical step in chip design verification, but it can be resource-intensive and time-consuming, particularly when dealing with numerous failures. To address these challenges, this paper presents a set of artificial intelligence (AI) driven verification technologies that automate and optimize the regression analysis process. The proposed technologies include regression binning, bug prediction, correlative root cause analysis, and message analysis. By combining these tools, the paper demonstrates the transformative power of AI in chip design verification, enabling faster root cause analysis and improved efficiency. The regression debug flow is illustrated, showcasing how each technology contributes to streamlining the debugging process. Overall, the AI-driven approach offers significant benefits regarding reduced triage time, improved risk assessment, and accurate root cause identification.

Craig Yang, Synopsys, Hsinchu, Taiwan (tsyang@synopsys.com)

Jaw Lee, Synopsys, Hsinchu, Taiwan (jawlee@synopsys.com)

Sherwin Lai, Synopsys, Hsinchu, Taiwan (sherwinl@synopsys.com)

Speaker: Craig Yang

Craig Yang is a staff engineer at Synopsys, specializing in front-end verification with a focus on root cause analysis and regression debug automation. He joined Synopsys in 2015. Craig has a proven track record of leading projects that involve different teams, successfully delivering cross-product features for RDA and ExecMan. With a keen interest in machine learning, Craig manages projects related to failure triage and bug prediction, utilizing unsupervised and supervised machine learning algorithms. Throughout his career, Craig has deployed new products and features in more than 8 customer’s regression environments, bridging the gap between the outside customers and inside developers. Craig holds a master’s degree from the Department of Computer and Information Science at National Chiao-Tung University.