Speed Up Functional Coverage Closure of CORDIC Designs Using Machine Learning Models

Abstract

Accurate Machine Learning ML models used for speeding up coverage closure are presented in this paper. Different ML models: Artificial Neural Network ANN, Deep Neural Network DNN, Support Vector Regression SVR and Decision Trees DT are trained to constrain the randomization of a Coordinate Rotation Digital Computer CORDIC design input values to hit the planned coverage items. Used ML models are compared in terms of evaluation metrics such as: Mean Squared Error MSE and R2 score. Training time overhead for each model is also considered. Tested ML models demonstrate an improvement of 55% in the number of transactions required to reach complete coverage closure when compared to traditional open-loop randomization method. Comparative analysis shows that DT is the most effective ML model to be incorporated in a CORDIC functional verification environment, due to its low training time overhead and high prediction accuracy.