A DATA-PARALLEL SCIENTIFIC MODELING LANGUAGE

The data parallel meta language (DPML) and its associated Fortran sour ce code rewriter (DP77) support architecture independent, high perform ance climate and weather prediction models. The language allows the da ta domains over which a program operates, the communication patterns r equired between elements of those data domains, and some or all of the calculations of a program to be expressed at a very high level. DPML uses explicit data parallelism to express the inherent parallelism of the models, with the result that programs are easily compilable into t arget machine code. DP77 uses information from the DPML program to tra nslate Fortran routines into the host specific Fortran form required f or their parallel execution within the model. This paper describes the general strategy behind the development of DPML, discusses its langua ge features using examples drawn from climate modelling, and provides details of the mechanism it uses for incorporating Fortran into data p arallel programs. Encouraging results are reported for DPML versions o f the standard weather benchmark models executing on vector, SIMD, and MIMD (shared memory) machines. While the paper is set within the fram ework of climate modelling, the technique has obvious wider implicatio ns. (C) 1994 Academic Press, Inc.