DVCLUB Europe | Using ML and AI in EDA Tools

Predicting Bugs

Conference:DVCLUB Europe | Using ML and AI in EDA Tools
Speaker:Daniel Hansson, Founder, Verifyter
Abstract:It is possible to predict bugs. More specifically, it is possible to look at a code change and predict whether any test will fail due to this change. This prediction can be made before running any tests, at time zero, and consequently allows you to concentrate on high-risk code changes first, resulting in better performance. It’s all based on historical bug data and ML. We use it in PinDown, our automatic debugger to debug faster.

3 Key Points:

  • Predicting bugs is possible
  • You need a mechanism to validate predictions (true for all ML/AI)
  • Bug prediction allows you to find and debug more bugs is less time
Speaker Biography:Daniel Hansson has 25 years’ experience as ASIC designer, EDA software engineer, project manager from both small and large companies within the ASIC and EDA industry. The last 3 years have been focused on using machine learning for automatic debug.

Sponsors

DVCLUB Europe is made possible through the generosity of our sponsors.