Gsearch/s 2Mb/mm2 TCAM Using Two-PhasePrecharge ML Sensing and Power-Grid PreConditioning to Reduce Ldi/dt Power-Supply Noise by 50%

📋 论文概要

本文提出了一种采用两阶段预充电匹配线感测和电源网格预条件技术的TCAM设计，实现了Gsearch/s的搜索速度和2Mb/mm²的密度，同时降低了Ldi/dt电源噪声。解决了传统TCAM在高速搜索时面临的功耗和噪声问题。

💡 主要创新点

• 提出两阶段预充电匹配线感测技术，通过分阶段预充电降低峰值电流和功耗。
• 采用电源网格预条件技术，有效抑制Ldi/dt电源噪声，提升电源完整性。
• 实现了高搜索速度（Gsearch/s）和高密度（2Mb/mm²）的TCAM设计。

🏷 关键词

📄 原文摘要

Van Butler1,2, Raymond Kim3, Ramon Rodriguez1, Tom Maffitt4, Joseph J. Oler1, John Goss1, Christopher Parkinson5,6, Michael A. Ziegerhofer1, Steven Burns1 Globalfoundries, Essex Junction, VT Green Mountain Semiconductor, Burlington, VT 3 Globalfoundries, Endicott, NY 4 IBM Research, Essex Junction, VT 5 Globalfoundries, Raleigh, NC 6 ASIC North, Raleigh, NC 1 2 Ternary Content Addressable Memory (TCAM) executes a fully parallel search of its entire memory contents and uses powerful wild-card pattern matching to return search results in a single clock cycle. This capability makes TCAM attractive for implementing fast hardware look-up tables in network routers, processor caches, and many pattern recognition applications. However, the push for higher performance and increased memory density coupled with parallel TCAM array activation during search operation creates large Ldi/dt power supply noise

👥 作者与机构

Igor Arsovski1, Michael Fragano1, Robert M. Houle1, Akhilesh Patil1,