Fast Synchronization on Shared-Memory Multiprocessors: An Architectural Approach
In today world, synchronization is most important operation in many of parallel applications. Traditional synchronization mechanisms are not able to keep up with the increasing demand for efficient synchronization operations as systems grow larger, the network size increases. The performance of synchronization of operation is controlled by two factors, i.e. the number of remote accesses that required for a synchronization operation and the latency of each remote access. The impact of performance of synchronization on the overall performance of parallel applications is increasing due to the speeding gap between processors and memory.