The potential of using dynamic information flow analysis in data value prediction

  • Authors:

  • Walid J. Ghandour;Haitham Akkary;Wes Masri

  • Affiliations:

  • American University of Beirut, Beirut, Lebanon;American University of Beirut, Beirut, Lebanon;American University of Beirut, Beirut, Lebanon

  • Venue:
  • Proceedings of the 19th international conference on Parallel architectures and compilation techniques
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Value prediction is a technique to increase parallelism by attempting to overcome serialization constraints caused by true data dependences. By predicting the outcome of an instruction before it executes, value prediction allows data dependent instructions to issue and execute speculatively, hence increasing parallelism when the prediction is correct. In case of a misprediction, the execution is redone with the corrected value. If the benefit from increased parallelism outweighs the misprediction recovery penalty, overall performance could be improved. Enhancing performance with value prediction therefore requires highly accurate prediction methods. Most existing general value prediction techniques are local and future outputs of an instruction are predicted based on outputs from previous executions of the same instruction. In this paper, we explore the possibility of introducing highly accurate general correlating value predictor using dynamic information flow analysis. We use information theory to mathematically prove the validity and benefits of correlating value predictors. We also introduce the concept of linear value predictors, a new technique that predicts a new value from another one using a linear relation. We then conduct empirical analysis using programs from SPECjvm2008 and Siemens benchmarks. Our empirical measurements support our mathematical theory and allow us to make important observations on the relation between predictability of data values and information flow. Furthermore, we provide a scheme to select highly predictable variables, and explain when a specific value predictor can perform well and even outperform other predictors. Using a dynamic information flow analysis tool called DynFlow, we show that the values of a set of selected variables can be predicted with very high accuracy, up to 100%, from previous history of the same variables or other variables that have strong information flow into the predicted variable.