The proposed perceptron branch predictor combines local history XOR hashing with that of global history. The local branch history (or self-history) helps to predict the branches exhibiting repetitive behaviors.
Perceptron is the simplest model of an artificial neural network used for pattern classification. As depicted in figure 1, each perceptron keeps track of a weight vector w. The w0 is a bias weight to which the corresponding input unit is always set to +1. The perceptron receives an input vector x. Each input unit xi is associated with a weight wi. Perceptron
The output y, is the dot product of input vector and weight vector. The sign of output y serves to make prediction, that non-negative output indicates “taken” and negative output indicates “not taken”.
wenbenz- py-pbp-sim
https://ieeexplore.ieee.org/abstract/document/4492487
Aneelkumar Ammanagi
Vinod Kumar R