FINE GRAIN PARALLELISM AND MULTITHREADED COMPUTING

In this paper, fine grain parallelism is emphasized in the design of massively parallel computers to acquire higher performance, but the increasing communication overhead should be solved first. Multithreads can not only implement fine grain parallelism effectively, but also be helpful to hide communication latency with reasonable scheduling strategies. So multithreaded computing is also proposed as a method for fine grained parallel processing. Multithreaded processors are expected to reduce the threads’ switching overhead.