High-performance computing and networking for climate research

The numerical simulation of Earth's climate stresses advances in computing and networking. This paper discusses the parallelization and distribution of the UCLA general circulation model of the atmosphere-ocean system both in a single multiprocessor computer or a distributed computer environment. It is argued that a 'metacomputer' consisting of a GRAY C90 connected to a massively parallel processor by a gigabit per second network can result in a substantial speedup in the execution of the code, which can eventually be superlinear. Such a speedup is realized in spite of overheads due to latency and communication delays.