Restart Optimization for Transactional Memory with Lazy Conflict Detection

This paper presents an optimization algorithm for transactional memory with lazy conflict detection. The proposed optimization attempts to minimize the execution time of restarted transactions. Minimizing happens during restart, by avoiding the re-execution of a section of a transaction that is unaffected by the restart. The proposed optimization builds on previous research and differs in that it eliminates the need for the prediction of conflicting accesses and introduces incremental context saving. Moreover, the paper introduces analytical models for estimating the execution time of transactions, with and without the restart optimization, that are developed using the continuous-time model. A critical evaluation comparing analytical models with the simulation results is discussed in the paper.