arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2406.05282cs.AR
收藏

基于查找表的神经网络硬件

Look-Up Table based Neural Network Hardware

Ovishake Sen, Chukwufumnanya Ogbogu, Peyman Dehghanzadeh, Janardhan Rao Doppa, Swarup Bhunia, Partha Pratim Pande, Baibhab Chatterjee

AI总结:

本文提出基于CMOS查找表的神经加速器LUT-NA,通过分而治之拆分高精度MAC运算解决扩展性问题,在面积和能耗上显著优于传统LUT及数字MAC方案,并支持混合精度分析。

摘要:

传统数字实现的神经加速器受到高功耗和高面积开销的限制,而模拟和非CMOS实现则存在噪声、器件失配和可靠性问题。本文提出了一种基于CMOS查找表(Look-Up Table,LUT)的神经加速器(LUT-NA)框架,通过预先计算、更快的查找来降低传统数字加速器的功耗、延迟和面积消耗,同时避免模拟电路的噪声和失配问题。为了解决传统基于LUT计算的扩展性问题,我们采用分而治之的方法,将高精度乘加(multiply and accumulate,MAC)运算拆分为较低精度的MAC运算。我们表明,与传统的基于LUT的技术相比,LUT-NA在每个推理任务上实现了高达29.54倍更低的面积和3.34倍更低的能耗;与传统的基于数字MAC的技术(Wallace Tree/Array乘法器)相比,实现了高达1.23倍更低的面积和1.80倍更低的能耗,且无需重新训练,也不影响精度,即使在彩票剪枝(lottery ticket pruned,LTP)模型上也是如此,这些模型已经将所需的MAC运算数量减少了高达98%。最后,我们在LUT-NA框架中针对各种LTP模型(VGG11、VGG19、Resnet18、Resnet34、GoogleNet)引入了混合精度分析,与传统的基于LUT的技术相比,各模型实现了高达32.22倍至50.95倍更低的面积和每个推理任务3.68倍至6.25倍更低的能耗;与传统的基于数字MAC的技术相比,各模型实现了高达1.35倍至2.14倍更低的面积需求和1.99倍至3.38倍更低的每个推理任务能耗,精度损失约为1%。

英文摘要:

Traditional digital implementations of neural accelerators are limited by high power and area overheads, while analog and non-CMOS implementations suffer from noise, device mismatch, and reliability issues. This paper introduces a CMOS Look-Up Table (LUT)-based Neural Accelerator (LUT-NA) framework that reduces the power, latency, and area consumption of traditional digital accelerators through pre-computed, faster look-ups while avoiding noise and mismatch of analog circuits. To solve the scalability issues of conventional LUT-based computation, we split the high-precision multiply and accumulate (MAC) operations into lower-precision MACs using a divide-and-conquer-based approach. We show that LUT-NA achieves up to $29.54\times$ lower area with $3.34\times$ lower energy per inference task than traditional LUT-based techniques and up to $1.23\times$ lower area with $1.80\times$ lower energy per inference task than conventional digital MAC-based techniques (Wallace Tree/Array Multipliers) without retraining and without affecting accuracy, even on lottery ticket pruned (LTP) models that already reduce the number of required MAC operations by up to 98%. Finally, we introduce mixed precision analysis in LUT-NA framework for various LTP models (VGG11, VGG19, Resnet18, Resnet34, GoogleNet) that achieved up to $32.22\times$-$50.95\times$ lower area across models with $3.68\times$-$6.25\times$ lower energy per inference than traditional LUT-based techniques, and up to $1.35\times$-$2.14\times$ lower area requirement with $1.99\times$-$3.38\times$ lower energy per inference across models as compared to conventional digital MAC-based techniques with $\sim$1% accuracy loss.

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。

↑