Segmented Precharge Free Content Addressable Memory for Energy Efficiency

Content-addressable memory (CAM) enables massively parallel comparison of a query pattern against its entire content. Baseline CAMs are either the NOR-type for fast but power-hungry or the NAND-type for slow operation and less energy consumption. In this work, we propose a Segmented Precharge-Free Content Addressable Memory (SPF-CAM) by synergistically combining the energy-efficiency characteristics of a modified NAND-type and the high-performance operation of the NOR-type. The proposed design reduces the energy-hungry precharge cycle by segmenting the match-line (ML). The first segment incorporates a precharge-free NAND-type that drives the operation of the second segment implemented with a dynamic NOR-type. The proposed segmentation strategy reduces unnecessary switching activity due to the precharge cycle by up to 99.61%. A delay-based model was developed to determine the optimal segmentation ratios for word lengths of 8, 16, 32, and 64 bits. Simulation results on 128x8, 128x16, 128x32 and 128x64 CAM banks show that, compared to conventional and state of the art CAM architectures, the SPF-CAM achieves up to 3.1× lower energy consumption and 1.5× improvement in energy-delay product (EDP).