Supercomputing has to be part of an organization-wide automation concept
W. Loeve · 1990
Development and application of digital simulation methods for fluid mechanics and solid mechanics very often requires supercomputers. In development of these methods mathematics, software engineering, knowledge of the underlying physics and knowledge of the application all are involved. For application of these methods in computer aided engineering in industry it is required that the methods meet quality standards. In the present paper it is concluded that as a consequence, specialists of different kinds, representing the disciplines involved, have to cooperate in the development of digital simulation methods. In general, cooperation of specialists of different kinds requires decomposition of the total task in sub-tasks that each can be performed preferably by one specialist. Efficiency requires that each specialist can perform his task under maximal self-control. This means that scientific information that specialists need as input, has to be accessible. Large part of the scientific information becomes available as data or as software that both have to be structured and enhanced with a description of relevance and purpose to become information. For management of the information mentioned above, supercomputers are not suitable. A mainframe is required for this. In addition for development of digital simulation methods, work stations are needed for pre- and post processing. Via a network the supercomputer shall be connected to the other systems that are needed. On this network tools have to be implemented such as tools for management of scientific information (enhanced data and software) and tools for computer aided software engineering. In this way supercomputing has to become part of an organization-wide automation concept. Invited paper presented at the Singapore Supercomputing Conference, 11 and 12 December 1990.