论文部分内容阅读
高密度及高性能存储器芯片的设计经常会产生一些新的可靠性问题;例如,高密度RAM和CCD芯片的α粒子问题便是一个很好的例子,α粒子引起的软错误会与已经存在的硬错误联结在一起,产生双位错,而此双位错是不能利用常规的单错纠正双错检测码来纠错的。本文绘出一个完整的系统方法,这种基于纠错码和系统维护策略的方法将减少系统级的主存失效率,其效果就如同α粒子问题未产生过一样。这种系统解决方法的设计可与大多数现存的存储器设计相兼容,因而使得实现这种方法所需附加的成本将是最小的。本文中描述的方法是利用单错纠正双错检测码的能力来检测出一个硬的和一个软的错误;然后使用一个微码和硬件算法纠正这软硬两个错误。本文也包括了这种方法的分析和模拟结果以及与其他技术比较结果。
The design of high-density and high-performance memory chips often leads to new reliability issues; for example, the problem of alpha particles in high-density RAM and CCD chips is a good example. Soft errors caused by alpha particles can be associated with existing Hard errors are linked together to create double dislocations that can not be corrected using conventional single-error correction double-error detection codes. This paper presents a complete system approach, which based on error correction code and system maintenance strategy will reduce the system-level failure of main memory, the effect just as the problem of α particles did not produce the same. The design of this system solution is compatible with most existing memory designs, thus minimizing the additional cost of implementing this approach. The approach described in this article detects a hard and a soft error using the ability of a single error correction double-error detection code, and then uses a microcode and hardware algorithm to correct both the hard and soft errors. This article also includes the analysis and simulation results of this method as well as the comparison with other techniques.