Predicting multi-core system Fmax by data-learning methodology

The use of low-cost structural Fmax measurement as a replacement for in-system Fmax measurement for speed binning has been aided by the use of a data-learning approach that can be used to build a reliable single-core system Fmax predictor given structural Fmax. This paper uses industry test measurements to demonstrate how the data-learning approach can be applied to predict multi-core system Fmax, how to use the information of each core to achieve better prediction, and how the proposed methodology works on multiple lots.