Fetch Reordering and Partitioning of Execution Resources for Atomic Instruction Blocks with Control Flow Assertions by Yen - Ting Tony

The rePLay framework provides a mechanism upon which a variety of code optimizations can be deployed as an application executes. In this thesis, two optimizations are explored. First, the order in which instructions are fetched is optimized. Performance with an optimized schedule is shown to improve by 2.63%. The second optimization is to partition instructions for a clustered microarchitecture. This thesis demonstrates that when such a microarchitecture is implemented along with an optimized fetch schedule, performance still improves by 0.61%.