Résultats exploitables
The majority of scientific and engineering programmes are now using parallel technology. A typical platform is the workstation cluster, which allows many users to share the total processing resource of the cluster. This offers financial advantages as work can be completed more quickly. When each machine has its own private memory (distributed memory), processors communicate by passing messages. Programming then becomes much more difficult as, in principle, the programmer has to write a different code for each machine and the data has to be coded to pass from processor to processor. This is particularly complicated for scientific and engineering codes and it would be much better if the process could be automated High Performance Fortran (HPF) makes it possible for programmers to write codes in standard Fortran which can then be compiled to run on a wide range of parallel hardware. The efficiency of the resulting code relative to a hand-parallelized version is matched or exceeded.
Four industrial codes have been parallelized using the HPF compiler. These are an electromagnetic modelling code, a code used for seismic analysis of dams, an elastostatics code used in the automotive industry, and a compressible air flow code to model flow around complex objects. All of the codes were developed in Fortran 77 over many years. These were updated to Fortran 90 and then parallelized using the HPF compiler. The benefits of using the compiler are now better understood and the HPF versions allow further code development for both serial and parallel platforms. Codes in many other industries can be converted economically and the power of parallel processing can be harnessed.
Recherche de données OpenAIRE...
Une erreur s’est produite lors de la recherche de données OpenAIRE
Aucun résultat disponible