Indexed, global objects for distributed memory parallel architectures

Distributed memory, message passing (DMMP) parallel architectures, such as the Intel iPSC® computer system lintel 86], provide the processing power and memory to solve very large parallel programming problems. However, their large grain process model and distributed memory complicate the programmeCs view of the system. In contrast, the object oriented approach is well suited to distributed memory, message passing architectures. In addition to the well known advantages for software engineering, the partitioning of application data into objects provides a convenient basis for their distribution among the processing nodes. The transmittal of objects and their procedural invocations make efficient use of the message passing mechanism. The large number of objects relative to the number of nodes simplifies the balancing of the processing load and provides a basis for dynamic load management.