Dynamic Yield Analysis and Enhancement of FPGA Reconfigurable Memory Systems

Minsu Choi, Missouri University of Science and Technology
Nohpill Park

This document has been relocated to http://scholarsmine.mst.edu/ele_comeng_facwork/1569

There were 13 downloads as of 28 Jun 2016.

Abstract

This paper addresses the issues of field programmable gate arrays (FPGA) reconfigurable memory systems with faulty physical memory cells and proposes yield measurement techniques. Static yield (i.e., the yield which does not take into account the inherited redundancy utilization for repair) and dynamic yield (i.e., the yield which takes into account the inherited redundancy utilization for repair) of FPGA reconfigurable memory systems and their characteristics are extensively analyzed. Yield enhancement of conventional memory systems relies on additional redundancy, but FPGA reconfigurable memory systems have inherited redundancy and customizability. Thus, they can accommodate numerous target memory configurations, and redundant memory cells, if any, can be used as spares to enhance the dynamic yield of a target memory configuration. Three fundamental strategies are introduced and analyzed; i.e., redundant bit utilization, redundant word utilization, and a combination of both. Mathematical analysis of those techniques also has been conducted to study their effects on the yield. Selecting the most yield enhancing logical memory configuration which can accommodate a target memory requirement among the candidate configurations is referred to as optimal fitting. Optimal fitting algorithms for single configuration fitting, sequential reconfiguration system fitting, and concurrent reconfiguration system fitting are investigated based on the proposed yield analysis techniques.