A major breakthrough in applying Massively Parallel Processing (MPP), computer technology to scientific problems is expected to be achieved by an initiative launched at CERN1 on 8 March by a consortium of leading European supercomputer technology users and suppliers. Massively Parallel Processing is a new direction in Supercomputing technology, in which many (from 32 to over 1000) processing elements, often of the same technology as personal work-stations, are tightly coupled together to provide computing speeds and capabilities equal and even superior to traditional larger mainframe Supercomputers.
Contributors to the initiative, whose members include: CERN, CERFACS (Centre Européen de Recherche et Formation Avancée en Calcul Scientifique), major Meteorological and Climate Centres in France, Germany and the United Kingdom, and three leading European industrial supercomputer suppliers (Meiko, Parsys and Telmat), met at CERN on 8 and 9 March for the official start-up of the project.
This is the first time that CERN is the prime co-ordinating partner of a multimillion ECU Esprit (European Strategic Programme for Research and Development in Information Technology) project and illustrates the stronger links which are developing between the European Community and CERN. This new Massively Parallel Processing initiative is a direct result of the "Report of the High Performance Computing and Networking Advisory Committee" (chaired by Prof. Carlo Rubbia, Director-General of CERN), recently published by the Commission of the European Communities (CEC). The start-up of this initiative is a step forward in the development of High Performance Computing culture and know-how in the European scientific environment and will contribute to build up user confidence in the emerging and potentially strategic European MPP computing market.
An important part of the project will be the installation of two scalable parallel computing machines in both the CERFACS and CERN computing centres. These machines will be integrated into the respective computing environments and services. Complex physics applications and large climate and meteorological simulation programs will be developed and installed on these computing platforms with the goal of demonstrating the effectiveness of the MPP approach.
More than 50 scientists will work over a period of three years in the project; some of these will be recruited on a fixed term basis with CEC funding support and contribute to the permanent staff teams at the collaboration sites. First preliminary results are expected at the end of 1994. The project will be completed by 1996.