Low-Power Approximate RPR Scheme for Unsigned Integer Arithmetic Computation

Chen, Ke; Liu, Weiqiang; Louri, Ahmed; Lombardi, Fabrizio

doi:10.1109/ojnano.2022.3153329

Cited by 3 publications

(2 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, one of the major issues with the RPR scheme is its complex voter and consequently imposed overheads, which is due to the required extra hardware for comparing the results of full precision and reduced precision modules with each other, and also with a predetermined threshold for generating the final output [8], [29]. In [35], a circuit-level power gating-based RPR scheme for unsigned integer arithmetic units was proposed. In this method, the full precision module is powered off, and in case the difference between the outputs of the reduced precision modules is higher than a threshold, the full precision module is turned on to achieve higher accuracy.…”

Section: B Approximate Computing In Fault-tolerant Systemsmentioning

confidence: 99%

“…Unlike state-of-the-art techniques that either target only approximating voter (such as [6][7]) or only approximating certain redundant modules (like [8] [35]), in this paper, we aim at targeting full system approximations where both redundant compute modules and voter can be approximated to leverage the full potential of approximate computing. However, this comes with certain additional challenges, e.g., meeting the user-defined output quality constraint and removing the required complex voters in the RPR-based designs.…”

Section: B Approximate Computing In Fault-tolerant Systemsmentioning

confidence: 99%

See 1 more Smart Citation

X-Rel: Energy-Efficient and Low-Overhead Approximate Reliability Framework for Error-Tolerant Applications Deployed in Critical Systems

Vafaei,

Akbari,

Shafique

et al. 2023

IEEE Trans. VLSI Syst.

View full text Add to dashboard Cite

Triple Modular Redundancy (TMR) is one of the most common techniques in fault-tolerant systems, in which the output is determined by a majority voter. However, the design diversity of replicated modules and/or soft errors that are more likely to happen in the nanoscale era may affect the majority voting scheme. Besides, the significant overheads of the TMR scheme may limit its usage in energy consumption and areaconstrained critical systems. However, for most inherently errorresilient applications such as image processing and vision deployed in critical systems (like autonomous vehicles and robotics), achieving a given level of reliability has more priority than precise results. Therefore, these applications can benefit from the approximate computing paradigm to achieve higher energy efficiency and a lower area. This paper proposes an energyefficient approximate reliability (X-Rel) framework to overcome the aforementioned challenges of the TMR systems and get the full potential of approximate computing without sacrificing the desired reliability constraint and output quality. The X-Rel framework relies on relaxing the precision of the voter based on a systematical error bounding method that leverages user-defined quality and reliability constraints. Afterward, the size of the achieved voter is used to approximate the TMR modules such that the overall area and energy consumption are minimized. The effectiveness of employing the proposed X-Rel technique in a TMR structure, for different quality constraints as well as with various reliability bounds are evaluated in a 15-nm FinFET technology. The results of the X-Rel voter show delay, area, and energy consumption reductions of up to 86%, 87%, and 98%, respectively, when compared to those of the state-of-the-art approximate TMR voters. Also, the effectiveness of the proposed X-Rel-based TMR structure is assessed in four benchmark applications from different domains. For these benchmarks, results show 1.59×, 2.35×, and 3.39× Energy-Delay-Area-Product (EDAP) reduction for less than 1%, 5%, and 10% output quality degradations, respectively. Finally, an image processing application is benchmarked to evaluate the X-Rel framework efficacy in presence of errors, where the results show up to a 4.78× higher output image quality in comparison with the typical TMR voters.

show abstract

Section: B Approximate Computing In Fault-tolerant Systemsmentioning

confidence: 99%

Section: B Approximate Computing In Fault-tolerant Systemsmentioning

confidence: 99%

X-Rel: Energy-Efficient and Low-Overhead Approximate Reliability Framework for Error-Tolerant Applications Deployed in Critical Systems

Vafaei,

Akbari,

Shafique

et al. 2023

IEEE Trans. VLSI Syst.

View full text Add to dashboard Cite

show abstract

Reduced Precision Redundancy Systems by Approximation (RPA): Design and Analysis

Junsangsri

Lombardi

2023

2023 IEEE International Symposium on Circuits and Systems (ISCAS)

View full text Add to dashboard Cite

Approximate Computing: Hardware and Software Techniques, Tools and Their Applications

Raza,

Javed,

Kazmi

et al. 2023

J CIRCUIT SYST COMP

View full text Add to dashboard Cite

The limitations of scaling in CMOS technology pose challenges in meeting the requirements of future applications. To address these challenges, researchers are exploring various design techniques, including Approximate Computing (AC), which leverages the inherent error resilience of applications to achieve high performance and energy gains with desired quality. AC has gained popularity as a computer paradigm for error-resilient applications, and many researchers have studied AC across computing layers and developed tools for implementing these techniques. This paper provides a comprehensive survey of AC techniques at the abstraction levels of software and hardware and discusses the tools to implement AC in hardware and software, quality evaluation tools and comparison points. The paper also covers existing frameworks for AC, potential applications, future research directions, challenges and limitations. This information can guide researchers in identifying promising avenues for further advancements and innovations in this domain. Additionally, this paper compares state-of-the-art surveys of AC and highlights the unique features and contributions of this work that distinguish our work from previous surveys.

show abstract

Low-Power Approximate RPR Scheme for Unsigned Integer Arithmetic Computation

Cited by 3 publications

References 28 publications

X-Rel: Energy-Efficient and Low-Overhead Approximate Reliability Framework for Error-Tolerant Applications Deployed in Critical Systems

X-Rel: Energy-Efficient and Low-Overhead Approximate Reliability Framework for Error-Tolerant Applications Deployed in Critical Systems

Reduced Precision Redundancy Systems by Approximation (RPA): Design and Analysis

Approximate Computing: Hardware and Software Techniques, Tools and Their Applications

Contact Info

Product

Resources

About