An improved PAp branch prediction method

Abstract

An improved PAp branch prediction method comprises a predictor training time reducing method and a branch prediction accuracy improving method. The predictor training time reducing method comprises the steps of using a new finite-state machine to express the jump direction of a jump instruction, improving a jump direction training method of first training, and improving PT table counter counting rules. The branch prediction accuracy improving method comprises the steps of taking one more set of prediction information from the same fetch packet when encountering a jump instruction failing to jump in fetch, and resetting the PT table counter accurately after jump goes wrong. The method can reduce prediction training time, increase prediction efficiency and prediction accuracy and further improve processor performance.

Description

technical field [0001] The invention relates to the field of electronic technology, in particular to an improved PAp branch prediction method. Background technique [0002] In modern processors, pipeline technology is widely used, but for jump instructions, there are problems in its application in pipeline technology. For processors that do not perform branch prediction, when a jump instruction is encountered, the pipeline must be cleared and re-executed from the jump target address. Another way to treat the jump instruction is to add empty instructions such as bubbles after the jump instruction in the compilation stage. The purpose of this is to wait for the jump instruction to decode the jump target address before executing the next instruction. These practices greatly waste the execution cycle and seriously reduce the execution efficiency of the processor. For deeper series pipelines, the jump penalty is even greater. It can be seen that branch prediction technology is...

AI Technical Summary

Problems solved by technology

For larger programs, the PAp predictor has a clear advantage in prediction accuracy, but for small and medium-sized programs, due to the long training time, the prediction accuracy during training is low

Images