Aligning netbook microprocessor reliability to market demands

Abstract

In 2008 Intel released a low-power and low-cost microprocessor that opened up a new market for smaller mobile computers, commonly known as netbooks. During the subsequent two years, netbooks have grown to be a substantial component of the mobile computer industry, capturing a major share of mobile computer sales in 2009. As Intel considers its growth opportunities, nothing attracts more attention than the processor for netbook computers. However, low-cost processors are a new undertaking for Intel, in contrast to an historically strong presence in the mainstream and high-end personal computer and server industries. In order to compete in the low-cost sector, the design of the microprocessor cannot include any unneeded capability. This is certainly true for aspects based heavily on customer usage characteristics, such as design for reliability. The package-level CPU design includes solder joints which mount the CPU to the computer motherboard. Most of these solder joints serve the dual function of an electronic bridge between the silicon microprocessor and the computer as well as providing structural stability for the connection between the two. In recent years, Intel has added additional solder joints for certain products that are solely meant to provide additional stability. This is required to meet stringent reliability standards based upon estimated customer use patterns. In the case of the netbook processor, the usage patterns were originally assumed to be the same as those used for notebooks, influencing package-level design and requiring additional solder joints. However, after being in the consumer market for only a short period, evidence points towards different usage patterns for netbooks, which in turn affect the reliability targets and overall package-level design. Through recent studies and analysis, it has become evident that consumer usage of netbooks is indeed different from notebooks and that Intel's existing set of assumptions may need to be revised. Specifically, the device lifetime, percentage of active time, power cycle frequency, and application use vary significantly between notebook and netbook devices. By leveraging Intel's extensive experience in relating usage patterns to reliability targets, it is possible to estimate the effect that such differences may have and conclude that additional solder joints are not required. These analyses estimate that by using more realistic usage assumptions, it is possible to avoid an additional $24 million in production costs over a two-year period for the upcoming microprocessor design to be released in 2011.