Abstract
Advances in System-on-Chip (SoC) technology rely on manufacturing and assembling high-performance system cores for many critical applications. Among these cores, memory occupies the largest portion of the SoC area; this trend much likely will continue in the future as it is widely anticipated that it will approach the 94% level by the year 2014. As memory cells are more prone to defects and faults than logic cells, redundancy has been extensively used for enhancing defect and fault tolerance through repair by spare (row and column) replacement. Unlike legacy PCB (printed circuit board) or MCM (multichip module) based systems, embedded cores cannot be physically replaced once they are fabricated onto a SoC. To realize both enhanced manufacturing yield and .eld reliability, ATE (automated test equipment) and BISR (built-in-self-repair) are utilized to allocate redundancy for the embedded memory cores. As ATEs (for the repair of manufacturing defects) and BISR (for repairing field faults) rely on the provided redundancy (rows and columns), spare partition and utilization techniques are proposed in this paper to achieve an optimal combination of yield and reliability for embedded memory cores. Parametric simulation results for the single dimensional (i.e., spare columns) and two-dimensional (i.e., both spare columns and rows) cases are provided.