Distributed shared memory(DSM) machines provide shared memory paradigm and achieve high performance by the caching of shared data. However, they suffer from cache miss and remote access latency with coarse-grain patterns. In this paper, we suggest the combination of bulk transfer and prefetching as a new latency hiding technique in DSM machines. The purpose of bulk transfer is to replicate re-mote data into local memory and thus reduce remote ac-cesses. Adaptive Granularity was used for bulk transfer, Prefetching is added to fetch those replicated data to the cache at the right teme. We could apply a simple prefetch scheduling as in uniprocessor since bulk transfer converts remote access into local ones. Simulation results show the reduced latency and the potential of AG as a preferable architecture for the prefetching in DSM machines.
展开▼
