Transactional Memory Execution for Parallel Multithread Programming without Lock

With the increasing popularity of shared-memory programming model, especially at the advent of multicore processors, applications need to become more concurrent to take advantage of the increased computational power provided by chip level multiprocessing. Traditionally locks are used to enforce data dependence and timing constraints between the various threads. However locks are error- prone, and often leading to unwanted race conditions, priority inversion, or deadlock. Therefore, recent waves of research projects are exploring transaction memory systems as an alternative synchronization mechanism to locks. This paper presents a software transactional memory execution model for parallel multithread programming without lock.