神經形態計算：人腦啟發的計算技術分析

Neuromorphic Computing: Technology, Chipsets, and Applications - Technology Analysis of Computing Technologies Inspired by the Brain

出版日期: 2020年09月18日 | 出版商:

OMDIA | 英文 14 Pages; 3 Tables, Charts & Figures | 商品交期: 最快1-2個工作天內

價格

如有價格方面的疑問請按下「詢問」鍵來信查詢

全貌
簡介
目錄

簡介

隨著摩爾定律接近其極限，神經形態計算正在成為未來硬件發展的催化劑。這項技術使用脈衝神經網路（SNN）和憶阻器來模仿人的大腦，可提供實時，低功耗的計算，從而實現節能型自學習設備的夢想。

諸如人工智能（AI），物聯網（IoT）和邊緣設備之類的新技術需要一種新的硬件體系結構，該體系結構可以執行低功耗的高速計算。神經型態運算芯片仍處於起步階段，但會對從機器人技術到邊緣物聯網的下一代計算和通信解決方案產生重大影響。

儘管在不久的將來似乎將成為主流芯片技術，但這些芯片在實驗案例研究中已被證明是很有前途的。目前市場上只有少量實驗套件，但研發工作正在穩步進行，預計商業產品將在兩年內推出。

本報告探討了支持神經形態計算的新技術，提供有關技術概述，市場動態，主要市場以及其他重要建議的信息。

本報告解決的主要問題

什麼是神經形態計算？
為什麼神經形態計算吸引了人們的注意？
此技術在硬件級別如何工作？
技術的成熟程度如何？
哪些公司正在這一領域進行創新？

As Moore's law starts to reach its limits, neuromorphic computing is becoming a catalyst for future hardware developments. Neuromorphic computing mimics the human brain using spiking neural networks (SNNs) and memristors. It offers real-time, low power computing that enables users to realize the dream of self-learning, energy efficient devices. Emerging technologies like artificial intelligence (AI), Internet of Things (IoT), and edge devices demand a new hardware architecture that can scale and perform accelerated computation with small power consumption.

Although still in their infancy, neuromorphic chips have the potential to shape next-generation computing and communications solutions ranging from robotics to edge IoT. To be accepted as a mainstream chip technology might look like a distant possibility, but these chips have shown promise during experimental case studies. Only a few experimental kits are available on the market today, but steady progress is being made in neuromorphic research and commercial products are expected in two years.

This Omdia report covers the new technologies enabling neuromorphic computing. It analyzes the potential for exponential growth in the commercial use of neuromorphic technology within the next five years. The report also explores the significant impact of neuromorphic computing on various markets, particularly edge and IoT devices, and compares the key players developing neuromorphic technology.