Brain-inspired computing exploiting carbon nanotube FETs and resistive RAM: Hyperdimensional computing case study
Brain-inspired computing exploiting carbon nanotube FETs and resistive RAM: Hyperdimensional computing case study
复制标题
DOI：
10.1109/isscc.2018.8310399
复制
发表时间：
2018-02
期刊：
影响因子：
--
通讯作者：
Tony F. Wu;Haitong Li;Ping-Chen Huang;Abbas Rahimi;J. Rabaey;H. Wong;M. Shulaker;S. Mitra
中科院分区：
文献类型：
--
作者：
Tony F. Wu;Haitong Li;Ping-Chen Huang;Abbas Rahimi;J. Rabaey;H. Wong;M. Shulaker;S. Mitra
我们展示了一个端到端的大脑启发的超维（HD）计算纳米系统，有效的认知任务，如语言识别，使用多种新兴纳米技术的异构集成。它使用碳纳米管场效应晶体管（CNFET，一种新兴的逻辑技术，与硅CMOS相比具有显著的能量延迟积（EDP）优势[1]）和电阻式RAM（RRAM，一种新兴的存储器，有望实现密集的非易失性和模拟存储[2]）的单片3D集成。由于它们的低制造温度（每对语言20，000个句子（640万个字符））。2.一次性学习（即，从几个例子中学习），每种语言使用一个文本样本（大约100，000个字符）。3.尽管有78%的硬件错误（电路输出停留在0或1），仍能弹性运行（98%的准确度）。我们的HD纳米系统由1，952个CNFET和224个RRAM单元组成。
We demonstrate an end-to-end brain-inspired hyperdimensional (HD) computing nanosystem, effective for cognitive tasks such as language recognition, using heterogeneous integration of multiple emerging nanotechnologies. It uses monolithic 3D integration of carbon nanotube field-effect transistors (CNFETs, an emerging logic technology with significant energy-delay product (EDP) benefit vs. silicon CMOS [1]) and Resistive RAM (RRAM, an emerging memory that promises dense non-volatile and analog storage [2]). Due to their low fabrication temperature (20,000 sentences (6.4 million characters) per language pair. 2. One-shot learning (i.e., learning from few examples) using one text sample (∼100,000 characters) per language. 3. Resilient operation (98% accuracy) despite 78% hardware errors (circuit outputs stuck at 0 or 1). Our HD nanosystem consists of 1,952 CNFETs integrated with 224 RRAM cells.