首页> 外文会议>2012 IEEE International SOC Conference. >Neuromorphic computing: A SoC scaling path for the next decades
【24h】

Neuromorphic computing: A SoC scaling path for the next decades

机译:神经形态计算:未来几十年的SoC扩展之路

获取原文
获取原文并翻译 | 示例

摘要

A human brain weighs ∼1.5kg, and is made up of 100 billion neurons connected via an incredibly dense and complex neuron network. The network contains 4km of wire in every cubic millimeter, offering >100-million MIPS computing capability and <20W power. The neuromorphic computing model inspired by brain structure obtained the increased attentions from VLSI design and EDA communities following the recent device inventions like memristor, nanotube/wire, and CMOS/molecular circuitries. The three invited papers introduce the grand challenges of neuromorphic computing algorithm, system and architecture; discuss the-state-of-the-art neuromorphic systems, design methodologies, computing modeling, and applications, as well as and the implementation opportunities on System-on-Chip platform.
机译:人脑重约1.5千克,由1000亿个神经元组成,这些神经元通过一个极其密集和复杂的神经元网络连接在一起。该网络每平方毫米包含4公里的电线,可提供1亿多MIPS的计算能力和20瓦以下的功率。在最近的器件发明(如忆阻器,纳米管/导线和CMOS /分子电路)之后,受大脑结构启发的神经形态计算模型引起了VLSI设计和EDA社区的越来越多的关注。被邀请的三篇论文介绍了神经形态计算算法,系统和体系结构的巨大挑战;讨论最新的神经形态系统,设计方法,计算模型和应用程序,以及片上系统平台上的实现机会。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号