The PMESC programming library for distributed-memory MIMD computers

Efficient programming of task-parallel problems, where the number and execu tion times of the computational tasks can vary unpredictably, demands an as ynchronous and adaptive approach. In this sort of approach, however, such f undamental programming issues as load sharing, data sharing, and terminatio n detection can present difficult programming problems. This paper presents the PMESC library for managing task-parallel problems on distributed-memor y MIMD computers within the context of the SPMD (single program, multiple d ata) programming model. PMESC offers support for all of the application-ind ependent programming issues involved in SPMD task-parallel computation in a portable and efficient way while still allowing users to customize their c odes. Because different problems may require different strategies to achiev e good performance, PMESC is based on a straightforward model in which diff erent building blocks can be easily put together and changed to accommodate the particular needs of the different applications. The library provides a n interface that allows users to program a virtual machine and thereby igno re the details associated with message passing and machine architecture. Th ese features make PMESC accessible to a wide variety of users. (C) 1999 Aca demic Press, Inc.