Neural branch prediction: from the first ideas, to implementations in advanced microprocessors and medical applications

摘要

This paper presents the idea of dynamic neural branch prediction, from its early theoretical foundations, firstly developed by the author of this paper, up to its implementations in present-day commercial advanced microprocessors. Also, the paper analyses a scientific work that proposes the use of neural branch predictors in predicting the brain’s neuronal activity, with useful medical applications. Machine Learning domain significantly fertilized Computer Architecture research during the last years. Dynamic Neural Branch Prediction was one of the earliest successful examples in this sense, being now an interesting fertile story, already used by industry in developing commercial advanced superscalar and simultaneous multithreading microprocessors.