Scholarlink

<p>Content-addressable memories (CAMs) are well-suited for certain computing tasks since they can search through a whole dataset in a single cycle, which makes them suitable for cryogenic applications like quantum computing and deep space exploration. This work presents a cryogenic ternary CAM (TCAM) based on ferroelectric superconducting quantum interference devices (FeSQUIDs). FeSQUID-based TCAM provides binary decisions (zero or non-zero voltage) for matching and mismatching conditions and achieves exceptional energy efficiency—consuming only 1.36 aJ and 26.5 aJ for 1-bit binary and ternary searches, respectively. To demonstrate its system-level potential, we integrate the TCAM into a brain-inspired hyperdimensional computing (HDC) framework, where it performs associative memory tasks during inference. For a vector size of 10,000 bits, the total energy consumption is estimated at just 89.4 fJ per vector. Compared to a 5 nm FinFET SRAM-based TCAM, the FeSQUID-based design achieves over an order of magnitude reduction in energy consumption.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Energy-efficient cryogenic ternary content addressable memory using ferroelectric SQUID

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月1日

查看全文
  • Shamiul Alam,
  • Simon Thomann,
  • Shivendra Singh Parihar,
  • Yogesh Singh Chauhan,
  • Kai Ni,
  • Hussam Amrouch,
  • Ahmedullah Aziz

摘要

Content-addressable memories (CAMs) are well-suited for certain computing tasks since they can search through a whole dataset in a single cycle, which makes them suitable for cryogenic applications like quantum computing and deep space exploration. This work presents a cryogenic ternary CAM (TCAM) based on ferroelectric superconducting quantum interference devices (FeSQUIDs). FeSQUID-based TCAM provides binary decisions (zero or non-zero voltage) for matching and mismatching conditions and achieves exceptional energy efficiency—consuming only 1.36 aJ and 26.5 aJ for 1-bit binary and ternary searches, respectively. To demonstrate its system-level potential, we integrate the TCAM into a brain-inspired hyperdimensional computing (HDC) framework, where it performs associative memory tasks during inference. For a vector size of 10,000 bits, the total energy consumption is estimated at just 89.4 fJ per vector. Compared to a 5 nm FinFET SRAM-based TCAM, the FeSQUID-based design achieves over an order of magnitude reduction in energy consumption.