Monolithic 3D IC designs for low-power deep neural networks targeting speech recognition

In recent years, deep learning has become widespread for various real-world recognition tasks. In addition to recognition accuracy, energy efficiency is another grand challenge to enable local intelligence in edge devices. In this paper, we investigate the adoption of monolithic 3D IC (M3D) technology for deep learning hardware design, using speech recognition as a test vehicle. M3D has recently proven to be one of the leading contenders to address the power, performance and area (PPA) scaling challenges in advanced technology nodes. Our study encompasses the influence of key parameters in DNN hardware implementations towards energy efficiency, including DNN architectural choices, underlying workloads, and tier partitioning choices in M3D. Our post-layout M3D designs, together with hardware-efficient sparse algorithms, produce power savings beyond what can be achieved using conventional 2D ICs. Experimental results show that M3D offers 22.3% iso-performance power saving, convincingly demonstrating its entitlement as a solution for DNN ASICs. We further present architectural guidelines for M3D DNNs to maximize the power saving.

[1]  A. Toffoli,et al.  Advances in 3D CMOS sequential integration , 2009, 2009 IEEE International Electron Devices Meeting (IEDM).

[2]  Geoffrey Zweig,et al.  The microsoft 2016 conversational speech recognition system , 2016, 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[3]  Daniel Povey,et al.  The Kaldi Speech Recognition Toolkit , 2011 .

[4]  Song Han,et al.  Deep Compression: Compressing Deep Neural Network with Pruning, Trained Quantization and Huffman Coding , 2015, ICLR.

[5]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[6]  Sung Kyu Lim,et al.  Design and CAD methodologies for low power gate-level monolithic 3D ICs , 2014, 2014 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED).

[7]  Brian Kingsbury,et al.  New types of deep neural network learning for speech recognition and related applications: an overview , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[8]  Chaitali Chakrabarti,et al.  Efficient memory compression in deep neural networks using coarse-grain sparsification for speech applications , 2016, 2016 IEEE/ACM International Conference on Computer-Aided Design (ICCAD).

[9]  W. Gardner.  Learning characteristics of stochastic-gradient-descent algorithms: A general study, analysis, and critique , 1984 .

[10]  Sung Kyu Lim,et al.  Match-making for Monolithic 3D IC: Finding the right technology node , 2016, 2016 53nd ACM/EDAC/IEEE Design Automation Conference (DAC).

[11]  Vivienne Sze,et al.  Hardware for machine learning: Challenges and opportunities , 2017, 2017 IEEE Custom Integrated Circuits Conference (CICC).

[12]  Xihong Wu,et al.  GMM-HMM acoustic model training by a two level procedure with Gaussian components determined by automatic model selection , 2010, 2010 IEEE International Conference on Acoustics, Speech and Signal Processing.