Microprocessor Loadline Measurement Prediction Using Artificial Intelligence Regression
Sameer Shekhar, Simon Chun Kit See · 2025
Simulation to measurement validation is a vital component for computing system power supply design. This paper presents application of AI based regression for predicting power supply impedance measurement. Unlike classical regression approach, proposed probabilistic regression approaches lead to distribution computation which help to predict measurement data based on simulation. Paper presents data from several regression approaches and concludes Bayesian regression as the optimal model with low mean squared error of 8%. Additionally, paper employed considerations, hyperparameter sensitivity. Finally, discusses AI methodology, over-underfitting and prediction accuracy, model insight and guidelines based on measurement verification for 6 power supplies across 2 product generations are provided.