As the pipeline depth and issue rate of high-performance superscalar processors increase, the importance of an excellent branch predictor becomes more crucial to delivering the potential performance of a wide-issue, deep pipelined processor. Conventional two-level branch predictors predict the outcome of a branch either based on the local branch history information, comprising the previous outcomes of a single branch, or based on the global branch history information, comprising the previous outcomes of all branches. The authors propose a new branch prediction scheme, called LGshare, which employs both the global and local branch history information simultaneously to improve the branch prediction accuracy for superscalar processors. It is shown that LGshare can achieve higher branch prediction accuracy than conventional two-level predictors when the size of the pattern history table is fixed.
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Hardware and Architecture
- Computational Theory and Mathematics