Energy- and Quality-Efficient Approximate Multipliers for Neural Network and Image Processing Applications

被引:52
作者
Ahmadinejad, Mohammad [1 ]
Moaiyeri, Mohammad Hosein [1 ]
机构
[1] Shahid Beheshti Univ, Fac Elect Engn, Tehran 1983969411, Iran
关键词
Compressors; Logic gates; Measurement; Complexity theory; Adders; Transistors; Neural networks; Approximate computing; approximate multiplier; compressor; energy efficiency; neural network; image processing; 4-2; COMPRESSORS; DESIGN; POWER; FINFETS; CMOS;
D O I
10.1109/TETC.2021.3072666
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Approximate computing is a new trend that trades off computational accuracy for lower energy dissipation and design complexity in various applications, where high precision is not a critical need. In this paper, energy- and quality- efficient approximate multipliers based on new approximate compressors are proposed. We use NAND gates for generating the complemented partial products, which reduces the number of transistors. Furthermore, new approximate compressors with different accuracy and performance characteristics are designed. Accordingly, three hybrid approximate multipliers offering different trade-offs between accuracy and hardware efficiency are proposed. The proposed designs are simulated using HSPICE with the 7nm FinFET model as a modern technology. Furthermore, the efficacies of the approximate multipliers in the neural network and image processing applications are evaluated using MATLAB. According to the results, the proposed designs provide far better compromises between the quality and energy metrics in comparison with the previous designs and can be considered as efficient alternatives for the exact multipliers in neural network and image processing applications.
引用
收藏
页码:1105 / 1116
页数:12
相关论文
共 43 条
[1]   Low power multipliers based on new hybrid full adders [J].
Abid, Z. ;
El-Razouk, H. ;
El-Dib, D. A. .
MICROELECTRONICS JOURNAL, 2008, 39 (12) :1509-1515
[2]   Design Exploration of Energy-Efficient Accuracy-Configurable Dadda Multipliers With Improved Lifetime Based on Voltage Overscaling [J].
Afzali-Kusha, Hassan ;
Vaeztourshizi, Marzieh ;
Kamal, Mehdi ;
Pedram, Massoud .
IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2020, 28 (05) :1207-1220
[3]   Energy and area efficient imprecise compressors for approximate multiplication at nanoscale [J].
Ahmadinejad, Mohammad ;
Moaiyeri, Mohammad Hossein ;
Sabetzadeh, Farnaz .
AEU-INTERNATIONAL JOURNAL OF ELECTRONICS AND COMMUNICATIONS, 2019, 110
[4]   Dual-Quality 4:2 Compressors for Utilizing in Dynamic Accuracy Configurable Multipliers [J].
Akbari, Omid ;
Kamal, Mehdi ;
Afzali-Kusha, Ali ;
Pedram, Massoud .
IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2017, 25 (04) :1352-1361
[5]  
[Anonymous], 2015, PROC 25 GREAT LAKES
[6]   Improving the Accuracy and Hardware Efficiency of Neural Networks Using Approximate Multipliers [J].
Ansari, Mohammad Saeed ;
Mrazek, Vojtech ;
Cockburn, Bruce F. ;
Sekanina, Lukas ;
Vasicek, Zdenek ;
Han, Jie .
IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS, 2020, 28 (02) :317-328
[7]   Low-Power Approximate Multipliers Using Encoded Partial Products and Approximate Compressors [J].
Ansari, Mohammad Saeed ;
Jiang, Honglan ;
Cockburn, Bruce F. ;
Han, Jie .
IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS, 2018, 8 (03) :404-416
[8]   An energy and area efficient 4:2 compressor based on FinFETs [J].
Arasteh, Armineh ;
Moaiyeri, Mohammad Hossein ;
Taheri, MohammadReza ;
Navi, Keivan ;
Bagherzadeh, Nader .
INTEGRATION-THE VLSI JOURNAL, 2018, 60 :224-231
[9]  
Baran Dursun, 2010, Proceedings of the 16th ACM/IEEE International Symposium on Low Power Electronics and Design (ISLPED 2010), P147, DOI 10.1145/1840845.1840876
[10]  
Chang CH, 2004, IEEE T CIRCUITS-I, V51, P1985, DOI [10.1109/TCSI.2004.835683, 10.1109/tcsi.2004.835683]