Abstract SUPPORTING NESTED PARALLELISM

Many parallel applications do not completely fit into the data parallel model. Although these applications contain data parallelism, task parallelism is needed to represent the natural computation structure or enhance performance. To combine the easiness of programming of the data parallel model with the efficiency of the task parallel model allows to parallel forms to be nested, giving Nested parallelism. In this work, we examine the solutions provided to N ested parallelism in two standard parallel programming platforms, HPF and MPI. Both their expression capacity and their efficiency are compared on a Cray- 3TE, which is distributed memory machine. Finally, an additional speech about the use of the methodology proposed for MPI is done on two different architectures