论文部分内容阅读
Most transactional memory (TM) research focused on multi-core processors, and others investigated at the clusters, leaving the area of non-uniform memory access (NUMA) system unexplored. The existing TM implementations made significant performance degrada