An Energy-Efficient and High Throughput in-Memory Computing Bit-Cell With Excellent Robustness Under Process Variations for Binary Neural Network

被引:6
|
作者
Saha, Gobinda [1 ,3 ]
Jiang, Zhewei [1 ,4 ]
Parihar, Sanjay [2 ]
Cao, Xi [1 ]
Higman, Jack [2 ]
Ul Karim, Muhammed Ahosan [1 ]
机构
[1] GLOBALFOUNDRIES, Santa Clara, CA 95054 USA
[2] GLOBALFOUNDRIES, Austin, TX 78735 USA
[3] Purdue Univ, W Lafayette, IN 47907 USA
[4] Columbia Univ, New York, NY 10027 USA
来源
IEEE ACCESS | 2020年 / 8卷
关键词
Discharges (electric); Convolution; Method of moments; Capacitors; Random access memory; Throughput; Neural networks; In-memory computing; SRAM; binary neural network; nonideality; process variation; MACRO;
D O I
10.1109/ACCESS.2020.2993989
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
In-memory computing (IMC) is a promising approach for energy cost reduction due to data movement between memory and processor for running data-intensive deep learning applications on the computing systems. Together with Binary Neural Network (BNN), IMC provides a viable solution for running deep neural networks at the edge devices with stringent memory and energy constraints. In this paper, we propose a novel 10T bit-cell with a back-end-of-line (BEOL) metal-oxide-metal (MOM) capacitor laid on pitch for in-memory computing. Our IMC bit-cell, when arranged in a memory array, performs binary convolution (XNOR followed by Bit-count operations) and binary activation generation operations. We show, when binary layers of BNN are mapped into our IMC arrays for MNIST digit classification, 98.75 accuracy with energy efficiency of 2193 TOPS/W and throughput of 22857 GOPS can be obtained. We determine the memory array size considering the word-line and bit-line nonidealities and show how these impact classification accuracy. We analyze the impact of process variations on classification accuracy and show how word-line pulse tunability provided by our design can be used to improve the robustness of classification under process variations.
引用
收藏
页码:91405 / 91414
页数:10
相关论文
共 31 条
  • [1] An Energy-Efficient Time Domain Based Compute In-Memory Architecture for Binary Neural Network
    Chakraborty, Subhradip
    Kushwaha, Dinesh
    Goel, Abhishek
    Singla, Anmol
    Bulusu, Anand
    Dasgupta, Sudeb
    [J]. 2024 25TH INTERNATIONAL SYMPOSIUM ON QUALITY ELECTRONIC DESIGN, ISQED 2024, 2024,
  • [2] IMCE: Energy-Efficient Bit-Wise In-Memory Convolution Engine for Deep Neural Network
    Angizi, Shaahin
    He, Zhezhi
    Parveen, Farhana
    Fan, Deliang
    [J]. 2018 23RD ASIA AND SOUTH PACIFIC DESIGN AUTOMATION CONFERENCE (ASP-DAC), 2018, : 111 - 116
  • [3] Vesti: Energy-Efficient In-Memory Computing Accelerator for Deep Neural Networks
    Yin, Shihui
    Jiang, Zhewei
    Kim, Minkyu
    Gupta, Tushar
    Seok, Mingoo
    Seo, Jae-Sun
    [J]. IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2020, 28 (01) : 48 - 61
  • [4] A Bit-Precision Reconfigurable Digital In-Memory Computing Macro for Energy-Efficient Processing of Artificial Neural Networks
    Kim, Hyunjoon
    Chen, Qian
    Yoo, Taegeun
    Kim, Tony Tae-Hyoung
    Kim, Bongjin
    [J]. 2019 INTERNATIONAL SOC DESIGN CONFERENCE (ISOCC), 2019, : 166 - 167
  • [5] An Energy-efficient Matrix Multiplication Accelerator by Distributed In-memory Computing on Binary RRAM Crossbar
    Ni, Leibin
    Wang, Yuhao
    Yu, Hao
    Yang, Wei
    Weng, Chuliang
    Zhao, Junfeng
    [J]. 2016 21ST ASIA AND SOUTH PACIFIC DESIGN AUTOMATION CONFERENCE (ASP-DAC), 2016, : 280 - 285
  • [6] Energy-Efficient In-Memory Binary Neural Network Accelerator Design Based on 8T2C SRAM Cell
    Oh, Hyunmyung
    Kim, Hyungjun
    Ahn, Daehyun
    Park, Jihoon
    Kim, Yulhwa
    Lee, Inhwan
    Kim, Jae-Joon
    [J]. IEEE SOLID-STATE CIRCUITS LETTERS, 2022, 5 : 70 - 73
  • [7] IMC: Energy-Efficient In-Memory Convolver for Accelerating Binarized Deep Neural Network
    Angizi, Shaahin
    Fan, Deliang
    [J]. PROCEEDINGS OF NEUROMORPHIC COMPUTING SYMPOSIUM (NCS 2017), 2017,
  • [8] High Performance and Energy-Efficient In-Memory Computing Architecture based on SOT-MRAM
    He, Zhezhi
    Angizi, Shaahin
    Parveen, Farhana
    Fan, Deliang
    [J]. PROCEEDINGS OF THE IEEE/ACM INTERNATIONAL SYMPOSIUM ON NANOSCALE ARCHITECTURES (NANOARCH 2017), 2017, : 97 - 102
  • [9] 2-Bit-Per-Cell RRAM-Based In-Memory Computing for Area-/Energy-Efficient Deep Learning
    He, Wangxin
    Yin, Shihui
    Kim, Yulhwa
    Sun, Xiaoyu
    Kim, Jae-Joon
    Yu, Shimeng
    Seo, Jae-Sun
    [J]. IEEE SOLID-STATE CIRCUITS LETTERS, 2020, 3 : 194 - 197
  • [10] Energy-Efficient Neural Network Acceleration in the Presence of Bit-Level Memory Errors
    Kim, Sung
    Howe, Patrick
    Moreau, Thierry
    Alaghi, Armin
    Ceze, Luis
    Sathe, Visvesh S.
    [J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS, 2018, 65 (12) : 4285 - 4298