| Branch predictors are very critical in modern superscalar processors and are responsible for achieving high performance. As the depth of pipeline and instruction issue rate of highperformance superscalar processors increase, a branch predictor with high accuracy becomes indispensable. It has been speculated that by 2010 branch prediction will become the most limiting factor in the performance of a processor, than the memory system. Branch mispredictions have heavy penalty, causing flushing of the pipeline and re-fetching of instructions from the correct location.; In recent times, neural based branch predictors, like perceptron predictor, are found to have an edge over other popular two-level branch predictors. Branch predictors based on neural learning are the most accurate predictors in the literature as they have sophisticated learning ability to make predictions based on previous outcomes and predictions. However, they are expensive to implement. But perceptron based branch predictors are simple and are easy to implement with less hardware resources. One major advantage of perceptron predictors over the two-level schemes is that we can have longer global or local history length, and consequently the perceptron predictor is robust to aliasing, resulting in better prediction accuracy.; In this thesis, the behavior and the intricacies of the perceptron predictor are extensively studied. The perceptron predictor has outperformed the classic Gshare predictor with lesser hardware resource. For a memory size of 64KB, the perceptron branch predictor has prediction accuracy about 2-10% higher than that of Gshare. The advantage of having longer history lengths was exploited to determine the performance and the IPC values for the perceptron predictor and showed commendable results. Also, varying the training parameter and the number of perceptrons for prediction helped in analyzing the behavior of the perceptron predictor under different environments. |
