autoAx: An Automatic Design Space Exploration and Circuit Building Methodology utilizing Libraries of Approximate Components

被引:48
作者
Mrazek, Vojtech [1 ,2 ]
Hand, Muhammad Abdullah [2 ]
Vasicek, Zdenek [1 ]
Sekanina, Lukas [1 ]
Shafique, Muhammad [2 ]
机构
[1] Brno Univ Technol, IT4Innovat Ctr Excellence, Fac Informat Technol, Brno, Czech Republic
[2] Vienna Univ Technol, TU Wien, Inst Comp Engn, Vienna, Austria
来源
PROCEEDINGS OF THE 2019 56TH ACM/EDAC/IEEE DESIGN AUTOMATION CONFERENCE (DAC) | 2019年
关键词
D O I
10.1145/3316781.3317781
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Approximate computing is an emerging paradigm for developing highly energy-efficient computing systems such as various accelerators. In the literature, many libraries of elementary approximate circuits have already been proposed to simplify the design process of approximate accelerators. Because these libraries contain from tens to thousands of approximate implementations for a single arithmetic operation it is intractable to find an optimal combination of approximate circuits in the library even for an application consisting of a few operations. An open problem is "how to effectively combine circuits from these libraries to construct complex approximate accelerators". This paper proposes a novel methodology for searching, selecting and combining the most suitable approximate circuits from a set of available libraries to generate an approximate accelerator for a given application. To enable fast design space generation and exploration, the methodology utilizes machine learning techniques to create computational models estimating the overall quality of processing and hardware cost without performing full synthesis at the accelerator level. Using the methodology, we construct hundreds of approximate accelerators (for a Sobel edge detector) showing different but relevant tradeoffs between the quality of processing and hardware cost and identify a corresponding Pareto-frontier. Furthermore, when searching for approximate implementations of a generic Gaussian filter consisting of 17 arithmetic operations, the proposed approach allows us to identify approximately 10(3) highly relevant implementations from 10(23) possible solutions in a few hours, while the exhaustive search would take four months on a high-end processor.
引用
收藏
页数:6
相关论文
共 16 条
[1]  
[Anonymous], P 52 ANN DES AUT C
[2]  
[Anonymous], P 52 ANN DES AUT C
[3]   Fast and Accurate Estimation of Quality of Results in High-Level Synthesis with Machine Learning [J].
Dai, Steve ;
Zhou, Yuan ;
Zhang, Hang ;
Ustun, Ecenur ;
Young, Evangeline F. Y. ;
Zhang, Zhiru .
PROCEEDINGS 26TH IEEE ANNUAL INTERNATIONAL SYMPOSIUM ON FIELD-PROGRAMMABLE CUSTOM COMPUTING MACHINES (FCCM 2018), 2018, :129-132
[4]   SPIRAL: Extreme Performance Portability [J].
Franchetti, Franz ;
Low, Tze Meng ;
Popovici, Doru Thom ;
Veras, Richard M. ;
Spampinato, Daniele G. ;
Johnson, Jeremy R. ;
Puschel, Markus ;
Hoe, James C. ;
Moura, Jose M. F. .
PROCEEDINGS OF THE IEEE, 2018, 106 (11) :1935-1968
[5]   QuAd: Design and Analysis of Quality-Area Optimal Low-Latency Approximate Adders [J].
Hanif, Muhammad Abdullah ;
Hafiz, Rehan ;
Hasan, Osman ;
Shafique, Muhammad .
PROCEEDINGS OF THE 2017 54TH ACM/EDAC/IEEE DESIGN AUTOMATION CONFERENCE (DAC), 2017,
[6]   A Review, Classification, and Comparative Evaluation of Approximate Arithmetic Circuits [J].
Jiang, Honglan ;
Liu, Cong ;
Liu, Leibo ;
Lombardi, Fabrizio ;
Han, Jie .
ACM JOURNAL ON EMERGING TECHNOLOGIES IN COMPUTING SYSTEMS, 2017, 13 (04)
[7]   Bio-Inspired Imprecise Computational Blocks for Efficient VLSI Implementation of Soft-Computing Applications [J].
Mahdiani, H. R. ;
Ahmadi, A. ;
Fakhraie, S. M. ;
Lucas, C. .
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS, 2010, 57 (04) :850-862
[8]  
Mazahir S., 2017, IEEE T COMPUT, V66, P11
[9]   Probabilistic Error Modeling for Approximate Adders [J].
Mazahir, Sana ;
Hasan, Osman ;
Hafiz, Rehan ;
Shafique, Muhammad ;
Henkel, Joerg .
IEEE TRANSACTIONS ON COMPUTERS, 2017, 66 (03) :515-530
[10]  
Mrazek V, 2017, DES AUT TEST EUROPE, P258, DOI 10.23919/DATE.2017.7926993