A NEW MODEL FOR INTEGRATED NESTED TASK AND DATA-PARALLEL PROGRAMMING

High Performance Fortran (HPF) has emerged as a standard language for data parallel computing. However, a wide variety of scientific applica tions are best programmed by a combination of task and data parallelis m Therefore, a good model of task parallelism is important for continu ed success of HPF for parallel programming. This paper presents a task parallelism model that is simple, elegant, and relatively easy to imp lement in an HPF environment. Task parallelism is exploited by mechani sms for dividing processors into subgroups and mapping computations an d data onto processor subgroups. This model of task parallelism has be en implemented in the Fx compiler at Carnegie Mellon University. The p aper addresses the main issues in compiling integrated task and data p arallel programs and reports on the use of this model for programming various hat and nested task structures. Performance results are presen ted for a set of programs spanning signal processing, image processing , computer vision and environment modeling. A variant of this task mod el is a new approved extension of HPF and this paper offers insight in to the power of expression and ease of implementation of this extensio n.