Improved 'dynamic random access memory' prediction improves design decisions

A method of reliability prediction is discussed that provides a means to quantify design tradeoffs for memory systems comprised of dynamic random-access memories (DRAMs). This method, the NEW-NEW method, builds on earlier work that has predicted the reliability of narrow, specific classes of memory systems, such as systems without error detection and correction. The NEW-NEW approach provides memory reliability analysis of a wider spectrum of memory systems, and can serve as the framework for future memory evaluation as DRAMs evolve and their behaviour becomes better understood.<
>