Supercomputer reliability and mitigation

Reliability, Availability and Serviceability (RAS) are key to high performance computing. Because of the relatively high costs of supercomputers and the required support needed to operate these systems, a holistic approach must be taken to assure system reliability as defined: Reliability is the probability that a material, component, or system will perform its intended function under defined operating conditions for a specific period of time. Extension of the computer system lifetime will be achieved through the implementation of built-in-test-equipment (BITE) monitoring that will provide the required feedback for optimum environmental controls. Reliability improvements of MTBF of 25% have been predicted based upon redesign of the microprocessor cooling system to reduce average case temperatures from 90 °C to 75 °C