摘要 : The predictability of data values is studied at a fundamental level. Two basic predictor models are defined: Computational predictors perform an operation on previous values to yield predicted next value values. Examples we study ... 展开 The predictability of data values is studied at a fundamental level. Two basic predictor models are defined: Computational predictors perform an operation on previous values to yield predicted next value values. Examples we study are stride value prediction and last value prediction; Contest-Based predictors match recent value history (context) with previous value history and predict values based entirely on previously observed patterns. To understand the potential of value prediction we perform simulations with unbounded prediction tables that are immediately updated using correct data values. 收起