TY - JOUR
T1 - Design of an Approximate 4-2 Compressor with Error Recovery for Efficient Approximate Multiplication
AU - Hwang, Sungyoun
AU - Seok, Hyelin
AU - Kim, Yongtae
N1 - Publisher Copyright:
© 2024, Institute of Electronics Engineers of Korea. All rights reserved.
PY - 2024/8
Y1 - 2024/8
N2 - This paper introduces a novel and efficient approximate 4-2 compressor and multipliers that significantly improve overall computation accuracy with marginal hardware overhead. The proposed compressor incorporates an error recovery logic to rectify output errors under specific input conditions. As a result, the proposed multipliers, featuring this error recovery compressor, exhibit substantial improvements in normalized mean error distance (NMED) and mean relative error distance (MRED) by up to 89.8% and 97.1%, respectively, compared to existing approximate multipliers considered in this paper. Furthermore, when implemented in a 32-nm CMOS technology, the proposed designs enable noteworthy reductions of up to 25.2%, 22.9%, and 23.4% in area, power, and energy, respectively, in comparison to the alternative designs. The effectiveness of the proposed design is further validated through its application in a digital image processing algorithm.
AB - This paper introduces a novel and efficient approximate 4-2 compressor and multipliers that significantly improve overall computation accuracy with marginal hardware overhead. The proposed compressor incorporates an error recovery logic to rectify output errors under specific input conditions. As a result, the proposed multipliers, featuring this error recovery compressor, exhibit substantial improvements in normalized mean error distance (NMED) and mean relative error distance (MRED) by up to 89.8% and 97.1%, respectively, compared to existing approximate multipliers considered in this paper. Furthermore, when implemented in a 32-nm CMOS technology, the proposed designs enable noteworthy reductions of up to 25.2%, 22.9%, and 23.4% in area, power, and energy, respectively, in comparison to the alternative designs. The effectiveness of the proposed design is further validated through its application in a digital image processing algorithm.
KW - approximate compressor
KW - Approximate computing
KW - approximate multiplier
KW - energy efficiency erpedi
KW - error recovery
UR - http://www.scopus.com/inward/record.url?scp=85203492867&partnerID=8YFLogxK
U2 - 10.5573/JSTS.2024.24.4.305
DO - 10.5573/JSTS.2024.24.4.305
M3 - Article
AN - SCOPUS:85203492867
SN - 1598-1657
VL - 24
SP - 305
EP - 315
JO - Journal of Semiconductor Technology and Science
JF - Journal of Semiconductor Technology and Science
IS - 4
ER -