在卷积神经网络中,用于内存计算硬件的2D记忆器阵列和选择器的异质集成
Samarth Jain1, Sifan Li1, Haofei Zheng1
1Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, Singapore, 117583, Singapore.
Nature communications
|March 20, 2025
概括
这项研究介绍了一种使用二维哈二化记忆器和选择器的新型内存计算系统. 综合系统实现了AI任务的高精度,同时提高了能源效率.
科学领域:
- 材料科学 材料科学 材料科学
- 计算机工程 计算机工程
- 人工智能的人工智能
背景情况:
- 传统的·诺伊曼架构在能源消耗和延迟方面面临限制.
- 二维 (2D) memristor交叉阵列 (CBA) 提供了一个潜在的解决方案,但面临着诸如有限的阵列大小和潜入路径电流等挑战.
- 与外围电路的集成对于有效的硬件内存计算 (CIM) 系统至关重要.
研究的目的:
- 展示一个硬件CIM系统,使用2D HfSe2记忆元和Si选择器的异质集成.
- 为了应对2D记忆器CBA的挑战,包括潜入路径电流和外围电路集成.
- 为AI硬件加速开发一种与兼容的方法.
主要方法:
- 可扩展的2D hafnium diselenide (HfSe2) 记忆器和 (Si) 选择器的异质集成.
- 构建一个32x32一个选择器-一个记忆器 (1S1R) 阵列,以减轻潜入电流.
- 集成CBA与外围控制传感电路,包括时间域传感电路.
- 实现一个完整的硬件二元卷积神经网络 (CNN),内置激活功能.
主要成果:
- 1S1R阵列成功地减轻了潜伏电流,实现了89%的收益率.
- 集成的CBA证明了能效和响应时间的改进,与最先进的2D记忆器相美.
- 实现的硬件CNN在模式识别任务中实现了97.5%的准确性.
- 时间域传感电路提高了能源效率,超过了模拟数字转换器 (ADC) 的2.5倍.
结论:
- 2D HfSe2记忆元和Si选择器的异质集成为硬件CIM系统提供了可行的解决方案.
- 开发的系统为人工智能应用提供了能源效率和响应时间的显著改进.
- 这种与兼容的方法为先进的AI硬件解决方案铺平了道路.
相关概念视频
01:22Semiconductors
520
There is variation in the electrical conductivity of materials - metals, semiconductors, and insulators that are showcased with the help of the energy band diagrams.
Metals such as copper (Cu), zinc (Zn), or lead (Pb) have low resistivity and feature conduction bands that are either not fully occupied or overlap with the valence band, making a bandgap non-existent. This allows electrons in the highest energy levels of the valence band to easily transition to the conduction band upon gaining...
Metals such as copper (Cu), zinc (Zn), or lead (Pb) have low resistivity and feature conduction bands that are either not fully occupied or overlap with the valence band, making a bandgap non-existent. This allows electrons in the highest energy levels of the valence band to easily transition to the conduction band upon gaining...
520
01:25Neural Circuits
957
Neural circuits and neuronal pools are two of the main structures found in the nervous system. Neural circuits are networks of neurons that work together to carry out a specific task or process. They consist of interconnected neurons and glial cells, which provide structural and metabolic support.
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
Neuronal pools are collections of nerve cells with similar functions and interact through chemical and electrical signals. These pools include both interneurons (the central neural circuit nodes that...
957