Microarchitecture to Exploit Repetitive Computations and Values: Redundant Stores, Non Redundant Data Cache, Instruction Level Reuse, Trace Level Reuse and Trace Level Speculation

This book proposes several microarchitectural techniques that can be applied to various parts of current microprocessor designs to improve the memory system and to boost the execution of instructions. Some techniques attempt to ease the gap between processor and memory speeds, while the others attempt to alleviate the serialization caused by data dependences. The underlying aim behind all the proposed microarchitectural techniques is to exploit the repetitive behaviour in conventional programs. Instructions executed by real-world programs tend to be repetitious, in the sense that most of the data consumed and produced by several dynamic instructions are often the same. It is refereed the repetition of any source or result value as Value Repetition and the repetition of source values and operation as Computation Repetition. In particular, the techniques proposed for improving the memory system are based on exploiting the value repetition produced by store instructions, while the techniques proposed for boosting the execution of instructions are based on exploiting the computation repetition produced by all the instructions.