The E ects of Context Switching on Branch Predictor

This paper shows that context switching is not a sig-niicant factor to be considered when performing general branch prediction studies. Branch prediction allows for speculative execution by increasing available instruction level parallelism (ILP) and hiding the time required to resolve branch conditions. Accurate simulation of branch prediction is important because branch prediction strongly innuences the behavior of processor structures. For this study, a timesharing framework was developed by modifying SimpleScalar's branch predictor simulator. A thorough characterization of the eeects of branch predictor conng-uration, branch predictor area, and time slice length is provided. As further veriication, branch predictor performance with and without ushing the predictor structures is compared. Experiments show that operating system context switches have little eeect on branch prediction rate when using time slices representative of today's operating systems. Our ndings show that this results from the fact that time slices are much larger than the training time required by the branch predictor structures. For all predictor conngurations tested, the predictors train in under 128K instructions with or without ushing the branch predictor structures.