Due to the growing number of disasters caused by bad weather the Federal Office of Meteorology and Climatology (MeteoSwiss) has developed a weather forecast of higher precision and frequency. This new forecasting suite will work on the basis of a 2-kilometre grid as opposed to the 7-kilometre grid that is currently in use. In order to deal with the leap in compute power triggered by the new forecasting suite, CSCS, the Swiss National Supercomputer Centre, autonomous unit of ETH Zurich that hosts the forecasting suite, has given MeteoSwiss access to the Cray XT massively parallel computer architectures. This change in compute platform is a complex process that requires the migration of data, the porting and optimization of the compute codes and of course the validation and verification of the results obtained, which is crucial for weather forecasting.

As of January 2008 this new high-resolution forecast will go into production with 8 daily simulation runs lasting 25 minutes each, up from the former 2 daily low-resolution simulation runs of 60 minutes. This increase in number of runs and resolution along with the shortening of the run-time leads to a 10-fold increase in required compute power.

This challenge was met by CSCS with the newly installed Cray XT4 that was inaugurated on September 17th. All large computers at CSCS are given the name of a Swiss mountain. The Cray XT4 used by MeteoSwiss was named "Buin". With its 448 dual-core processors it provides a compute power of 5 trillion (10 to 12) computations per second.

During the validation period, CSCS experts demonstrated the ability of the Cray XT systems to provide the stability and reliability required for the new high-resolution forecasting suite. During the entire validation period lasting several months the machine delivered its services with the reliability of a Swiss clock.

The new supercomputer "Buin", with its elegant Swiss cross design, will also be made available to researchers from the Swiss universities and federal institutes of technology for scientific simulation projects in fields such as chemistry, physics, nano-science, material sciences, fluid dynamics and, of course, climate research.