A Case for Using Value Prediction to Improve Performance of Transactional Memory

Transactional Memory(TM) is an optimistic speculative synchronization scheme that provides atomic execution for a region of code marked as a transaction by the programmer. TM avoids many of the problems associated with mutual-exclusion-based synchronization and can make writing parallel programs relatively easier. Programs with critical sections that are not heavily contended benefit from the optimistic nature of TM systems. However, for heavily contended critical sections, performance for TM systems can degrade due to conflicts leading to stalls and expensive rollbacks. In this paper, we look into the nature of the shared data involved in conflicts for TM systems. We find that for a certain class of applications, shared-conflicting data is often updated in a predictable manner by different transactions. We propose using a value predictor to capture this predictability for such data structures and increase overall concurrency by satisfying loads from conflicting transactions with predicted values, instead of stalling. In this paper, we present a possible design and implementation of TM system with a value predictor to avoid conflicts between concurrent transactions. Our benchmark results show us that the value predictor can capture this predictable behavior for most benchmarks and can improve performance of TM programs by minimizing stalls and rollbacks due to conflicts.