Abstract
Approximate computing is one of the methods to improve performance in various error-resilient applications such as image and video processing. Multipliers are part of their computing unit that often requires many resources. In this paper, an approximate (8; 2) compressor is designed and compared with other available approximate compressors for quality, power consumption, delay and area in the circuit. The proposed approximate compressor is used in and multipliers. To show the quality of the proposed compressor, the proposed approximate multiplier of used to multiply two images in MATLAB tools, then qualitative parameters of SSIM and PSNR were measured and acceptable results were obtained. Also, the MED and NED accuracy parameters indicate the acceptable error rate in the output of the proposed multiplier circuit. Finally, we synthesized the proposed approximate compressor and multiplier designs by a Synopsys Design Compiler. The proposed multiplier can improve delay, area and Power Delay Products by 5%, 17%, and 8%, respectively, compared to other similar existing approximate multipliers.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Additional information
Notes on contributors
Maryam Banisharif Dehkordi
Maryam Banisharif Dehkordi received her BS degree in software engineering from Malayer University, Malayer, Iran, in 2018, and her master's degree in computer system architecture from the University of Guilan, Rasht, in 2021. She is currently a software development engineer with Hamkaran Systems. Her main research interests include approximate computing and low power circuits. Email: [email protected]
HamidReza Ahmadifar
HamidReza Ahmadifar received his PhD and a BS degrees in computer engineering from Shahid Beheshti University, Tehran, Iran in 2013 and 1998, respectively. He received his MSc degree in computer system architecture from Amir Kabir University of Technology, Tehran, Iran in 2001. He is currently an assistant professor in the Computer Engineering Department, the University of Guilan, Rasht, Iran. His research interests are residue number system, computer architecture and cloud computing.