Software development is an integral part of the science support we provide to our customers. During the past 15 years, scientific computing has evolved from entirely mainframe- based to today's configuration of distributed processing where a significant and growing portion of the work is done on workstations. These workstations either function alone or are connected by local networks with other workstations, individual systems forming clusters, and with supercomputers. This trend has accelerated as the computational power of workstations has advanced. At the same time, better-resolved modeling requires significantly greater computational resources and many modeling problems are extremely resource demanding on even the most powerful super computers.
The S/W challenge faced by SGT requires developing codes for two increasingly divergent environments, and also necessitates the need to avoid platform specificity. This trend mandates flexibility in design within the S/W development environment. There are also great opportunities for enhancements and advances in visualization that we are undertaking which are made possible by these developments. Furthermore, it has been necessary to identify and address issues regarding new trends, new computational environments, exploitation of multiple processor architectures, and most importantly, efficiency in coding and operations.