APPLICATIONS ON HIGH PERFORMANCE CLUSTER COMPUTERS Production of Mars Panoramic Mosaic Images’

The development, application and commercialization of cluster computer systems have escalated dramatically over the last several years. Driven by a range of applications that need relatively low-cost access to high performance computing systems, cluster computers have reached worldwide acceptance and use. A cluster system consists of commercial-off-the-shelf hardware coupled to (generally) open source software. These commodity personal computers are interconnected through commodity network switches and protocols to produce scalable computing systems usable in a wide range of applications. First developed by NASA Goddard Space Flight Center in the mid 1990s, the initial Caltech/JPL development resulted in the Gordon Bell Prize for price-per-performance using the 16-node machine HygZac in 1997 [I]. Currently the JPL High Performance Computing Group uses and maintains three generations of clusters including Hyglac. The available hardware resources include over 100 CPUs, over 80Gbytes of RAM, and over 600Gbytes of disk space. The individual machines are connected via 100Mbit.h and 2.0Gbit/s networks.