A Real-Time Error Detection (RTD) Architecture and Its Use for Reliability and Post-Silicon Validation for F/F Based Memory Arrays

This work proposes in-situ Real-Time Error Detection (RTD): embedding hardware in a memory array for detecting a fault in the array when it occurs, rather than when it is read. RTD breaks the serialization between data access and error-detection and, thus, it can speed-up the access-time of arrays that use in-line error-correction. The approach can also reduce the time needed to root-cause array related bugs during post-silicon validation and product testing. The paper introduces a two-dimensional error-correction scheme based on RTD and, also, presents a proactive error-correction method that combines RTD with demand-scrubbing. The work describes how to build RTD into a memory array with flip-flops to track in real-time the column-parity. A comparison of the proposed two-dimensional ECC scheme, as compared to single-error-correction-double-error-detection, shows that the RTD design has comparable error-detection-and-correction strength and, depending on the array dimensions and configuration, RTD reduces access time by 4% to 26% at an area and power overhead (negative is a reduction) between -7% to 33% and -42% to 86% respectively.