arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2309.06019cs.ARcs.AIcs.PF
收藏

DSLOT-NN:数字串行从左到右神经网络加速器

DSLOT-NN: Digit-Serial Left-to-Right Neural Network Accelerator

Chosun University(朝鲜大学)

Muhammad Sohail Ibrahim, Muhammad Usman, Malik Zohaib Nisar, Jeong-A Lee

AI总结:

本文提出 DSLOT-NN 加速器,利用数字串行从左到右的在线运算器加速 DNN 卷积推理,通过提前终止无效卷积并在 FPGA 上实现,相比 Stripes 显著节省功耗并提升约 50% 的每瓦特性能。

摘要:

我们提出了一种基于数字串行从左到右(Digit-Serial Left-tO-righT,DSLOT)算术的处理技术,称为 DSLOT-NN,旨在加速深度神经网络(DNNs)中卷积运算的推理。该工作能够评估并终止无效卷积,从而带来显著的功耗和能量节省。处理引擎由低延迟的最高有效位优先(most-significant-digit-first,MSDF)(也称为在线)乘法器和加法器组成,这些运算器按从左到右的顺序处理数据,从而允许后续操作以数字流水线方式执行。使用在线运算器无需开发复杂的负激活识别机制,因为具有最高权重值的输出会首先生成,并且一旦生成第一个非零数字即可识别结果的符号。在线运算器的精度可以在运行时调节,这使得它们在精度可以为功耗和能量节省而妥协的场景中极为有用。所提出的设计已在 Xilinx Virtex-7 FPGA 上实现,并在各种性能指标上与最先进的 Stripes 进行了比较。结果表明,所提出的设计具有功耗节省、更短的周期时间,并且每瓦特 OPS 大约提高 50%。

英文摘要:

We propose a Digit-Serial Left-tO-righT (DSLOT) arithmetic based processing technique called DSLOT-NN with aim to accelerate inference of the convolution operation in the deep neural networks (DNNs). The proposed work has the ability to assess and terminate the ineffective convolutions which results in massive power and energy savings. The processing engine is comprised of low-latency most-significant-digit-first (MSDF) (also called online) multipliers and adders that processes data from left-to-right, allowing the execution of subsequent operations in digit-pipelined manner. Use of online operators eliminates the need for the development of complex mechanism of identifying the negative activation, as the output with highest weight value is generated first, and the sign of the result can be identified as soon as first non-zero digit is generated. The precision of the online operators can be tuned at run-time, making them extremely useful in situations where accuracy can be compromised for power and energy savings. The proposed design has been implemented on Xilinx Virtex-7 FPGA and is compared with state-of-the-art Stripes on various performance metrics. The results show the proposed design presents power savings, has shorter cycle time, and approximately 50% higher OPS per watt.

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。

↑