Compositional Oil Reservoir Simulation in Fortran D: a Feasibility Study On Intel iPsc/860

This paper describes a study of the use of data-parallel languages such as Fortran D or High Performance Fortran (HPF) and their compilation systems for existing large scienti c applications. The central question addressed in this study is: Do such languages allow the expression of the parallelism available in the applications so that the underlying compilation systems are able to generate e cient code for the speci ed parallelism? Our results are based on Fortran D and the current prototype implementation of the Fortran D compilation sytem. This compilation system represents the state-of-the-art in compiler technology with respect to compile time optimizations, which is the main point of interest in this study. We test the feasibility of Fortran D on a small section of a reservoir simulation code of wide use in the oil industry. The code is written in a style to take advantage of the machine characteristics of a vector supercomputer such as a Cray. However, this machine dependent programming style inhibits many state-of-the-art compile time optimizations. This paper provides an insight into a data-parallel programming style that allows state-ofthe-art compilers to generate e cient code across a variety of machines. Our experiments on Intel's iPSC/860 distributed-memory multiprocessor indicate that the Fortran D language and its compilation system can exploit the parallelism that exists in computations related to reservoir modeling, provided that the code is written in data-parallel programming style. Additional experiments show that other machines and their compilation systems can also exploit the compile time information available for programs written in a data-parallel programming style. This line of work will help to design enhancements to current parallel languages and compilation techniques, allowing the parallelization of large existing codes not written in data-parallel programming style. 2