Many applications in different fields need to access large amounts of data. The difficulty of handling out-of-core data limits the performance of supercomputers. We first describe an out-of-core computation model on cluster of workstations. Then an object-oriented method is offered, which makes the out-of-core programs easier to write. Some optimal strategies, such as distribution of local array files, data sieving and prefetching, also make an improvement in the performance of out-of-core parallel computations.
Date of Conference: 27-29 Aug. 2003