READ: A fixed restoring array based accuracy-configurable approximate divider for energy efficiency

Arya, Neelam; Soni, Teena; Pattanaik, Manisha; Sharma, Govind

doi:10.1016/j.vlsi.2020.08.002

Cited by 6 publications

(2 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Researchers present a new method [27] for inverse and iterative approximation of integer data using the Newton-Raphson method, designed to achieve results close to exact values and enable efficient performance. This paper [28] presents the reconfigurable energy-efficient approximate divider (READ), which implements several configurable modes of energy quality using a fixed recovery array divider architecture. The READ achieves energy efficiency while meeting the dynamically changing accuracy requirements of the target application.…”

Section: Previous Workmentioning

confidence: 99%

QIAD: A quadratic interpolation approximate divider

Liu

Wang

Liu

2023

IEICE Electron. Express

View full text Add to dashboard Cite

The balance between computational performance and power consumption is a key constraint in computing systems, with integrated circuit technology posing a bottleneck. Approximate computation can trade off accuracy for performance or power improvement in error-tolerant scenarios. Division, with high computational demands and latency, is a bottleneck for computational efficiency. We propose a quadratic interpolation approximate divider (QIAD) based on multiplicative division, which has superior statistical performance. The design is simulated and synthesized at TSMC 65 nm process and tested on image color quantization, showing the best quantization effect using evaluation indicators such as PSNR, MSE, and SSIM.

show abstract

Section: Previous Workmentioning

confidence: 99%

QIAD: A quadratic interpolation approximate divider

Liu

Wang

Liu

2023

IEICE Electron. Express

View full text Add to dashboard Cite

show abstract

“…Novel approach of combining restoring and logarithmic divider leads to excellent tradeoff between accuracy and hardware performance (7). Fixed restoring array based accuracy-configurable approximate divider architecture (READ) introduces technique to change accuracy dynamically by selecting exact and approximate divider cells during computation (8). SEERAD, a high speed yet anergy efficient rounding based approximate divider rounds the divisor to nearest value such that division can be implemented as multiplication and bit shifts (9).…”

Section: Introductionmentioning

confidence: 99%

Design of Approximate Restoring Dividers for Error Resilient Applications

Gowtham¹,

Sasipriya²,

Angeline³

2022

ECS Trans.

View full text Add to dashboard Cite

In this paper, a low-power and high-speed approximate divider using restoring array architecture has been proposed. Approximation is realized by replacing the subtractor/divider cells with the approximate subtractor/divider cells through the use of reduced gate level complexity. Four approximate divider architectures, namely, AD1, AD2, AD3, and AD4, have been proposed. The amount of approximation can be scaled by introducing the approximation factor and the proposed dividers have been analyzed for different values of approximation factor. The simulation results show that the proposed dividers AD1, AD2, AD3, and AD4, with approximation factor of 10 have achieved power reduction of 17%, 30%, 23%, and 37% respectively, when compared to the exact restoring divider. The proposed dividers have also been compared with the existing approximate restoring divider and shows signification reduction in power and delay. All the simulations are carried out using 180nm CMOS technology. The image processing applications, such as change detection and background removal, have been implemented using the proposed divider to show the feasibility of employing the approximate divider for real time applications. The simulation results prove that the approximate dividers have realized a feasible PSNR for the resultant images.

show abstract

Non-iterative division circuit design with accuracy and performance trade-off based on mixed integer linear programming approach

Jong

2022

Microelectronics Journal

View full text Add to dashboard Cite

READ: A fixed restoring array based accuracy-configurable approximate divider for energy efficiency

Cited by 6 publications

References 13 publications

QIAD: A quadratic interpolation approximate divider

QIAD: A quadratic interpolation approximate divider

Design of Approximate Restoring Dividers for Error Resilient Applications

Non-iterative division circuit design with accuracy and performance trade-off based on mixed integer linear programming approach

Contact Info

Product

Resources

About