A Field Failure Analysis of Microprocessors Used in Information Systems

Soft errors due to cosmic particles are a growing reliability threat to information systems. In this paper, a methodology is developed to analyze the effects of Single Event Upsets (SEU) and obtain Failure In Time (FIT) rates for commercial server microprocessors in live information systems. The methodology is based on data collected from error logs and error traces of the microprocessors collected from systems in the field. Soft errors are further localized within the microprocessor resources with the assistance of the machine check architecture.