PIXEL: Photonic Neural Network Accelerator
PIXEL: Photonic Neural Network Accelerator
复制标题
DOI：
10.1109/hpca47549.2020.00046
复制
发表时间：
2020-02
期刊：
影响因子：
--
通讯作者：
Kyle Shiflett;Dylan Wright;Avinash Karanth;A. Louri
中科院分区：
文献类型：
--
作者：
Kyle Shiflett;Dylan Wright;Avinash Karanth;A. Louri
深度神经网络（DNN）等机器学习（ML）架构在图像分类和语音识别等现代应用中实现了前所未有的准确性。随着功耗成为ML架构的主要关注点，计算机架构师专注于设计节能硬件平台以及优化ML算法。为了大幅降低功耗并提高神经网络加速器的并行性，已经提出了诸如硅光子学的破坏性技术，与电气实现相比，该技术可以提高每瓦性能。在本文中，我们提出了PIXEL -光子神经网络加速器，它有效地实现了神经计算中的基本操作，即使用光子器件（如微谐振器（MRR）和马赫-曾德尔干涉仪（MZI））的乘法和累加（MAC）功能。我们设计了两个版本的PIXEL -一个混合版本，光学倍增和累积电和一个完全光学版本，光学倍增和累积。我们对不同卷积神经网络（AlexNet，VGG 16等）的不同版本的光子和电子加速器进行了详细的功率，面积和时序分析。我们的研究结果表明，与传统电气设计相比，两种PIXEL设计的能量延迟积都有显着改善（OE为48.4%，OO为73.9%），同时最大限度地减少了延迟，代价是增加了电气设计的面积。
Machine learning (ML) architectures such as Deep Neural Networks (DNNs) have achieved unprecedented accuracy on modern applications such as image classification and speech recognition. With power dissipation becoming a major concern in ML architectures, computer architects have focused on designing both energy-efficient hardware platforms as well as optimizing ML algorithms. To dramatically reduce power consumption and increase parallelism in neural network accelerators, disruptive technology such as silicon photonics has been proposed which can improve the performance-per-Watt when compared to electrical implementation. In this paper, we propose PIXEL - Photonic Neural Network Accelerator that efficiently implements the fundamental operation in neural computation, namely the multiply and accumulate (MAC) functionality using photonic components such as microring resonators (MRRs) and Mach-Zehnder interferometer (MZI). We design two versions of PIXEL - a hybrid version that multiplies optically and accumulates electrically and a fully optical version that multiplies and accumulates optically. We perform a detailed power, area and timing analysis of the different versions of photonic and electronic accelerators for different convolution neural networks (AlexNet, VGG16, and others). Our results indicate a significant improvement in the energy-delay product for both PIXEL designs over traditional electrical designs (48.4% for OE and 73.9% for OO) while minimizing latency, at the cost of increased area over electrical designs.