arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2010.11686cs.LGcs.ARcs.CV
收藏

一种基于对数计算的紧凑型嵌入式CNN处理器设计

A Very Compact Embedded CNN Processor Design Based on Logarithmic Computing

National Tsing Hua University(国立清华大学)

Tsung-Ying Lu, Hsu-Hsun Chin, Hsin-I Wu, Ren-Song Tsay

摘要:

在本文中,我们提出了一个非常紧凑的嵌入式CNN处理器设计的基础上修改的对数计算方法,使用非常低的位宽表示。我们的高品质CNN处理器可以轻松安装到边缘设备中。对于Yolov 2,我们的处理电路只需要0.15 mm 2使用台积电40 nm单元库。关键思想是将所有层的激活和权重值统一约束在[-1,1]范围内,并产生低位宽的对数表示。通过统一表示,我们设计了一个统一的,可重用的CNN计算内核,并显着减少计算资源。所提出的方法已经在许多流行的图像分类CNN模型(AlexNet,VGG 16和ResNet-18/34)和目标检测模型(Yolov 2)上进行了广泛的评估。硬件实现的结果表明,我们的设计只消耗最少的计算和存储资源,但达到了非常高的精度。该设计在FPGA上得到了充分的验证,SoC集成正在进行中,取得了可喜的成果。凭借极其高效的资源和能源使用,我们的设计非常适合边缘计算。

英文摘要:

In this paper, we propose a very compact embedded CNN processor design based on a modified logarithmic computing method using very low bit-width representation. Our high-quality CNN processor can easily fit into edge devices. For Yolov2, our processing circuit takes only 0.15 mm2 using TSMC 40 nm cell library. The key idea is to constrain the activation and weight values of all layers uniformly to be within the range [-1, 1] and produce low bit-width logarithmic representation. With the uniform representations, we devise a unified, reusable CNN computing kernel and significantly reduce computing resources. The proposed approach has been extensively evaluated on many popular image classification CNN models (AlexNet, VGG16, and ResNet-18/34) and object detection models (Yolov2). The hardware-implemented results show that our design consumes only minimal computing and storage resources, yet attains very high accuracy. The design is thoroughly verified on FPGAs, and the SoC integration is underway with promising results. With extremely efficient resource and energy usage, our design is excellent for edge computing purposes.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑