ENAP: An Efficient Number-Aware Pruning Framework for Design Space Exploration of Approximate Configurations

Dou, Yuqin; Wang, Chenghua; Woods, Roger; Liu, Weiqiang

doi:10.1109/tcsi.2023.3252483

Cited by 6 publications

(9 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Whilst approximate units have also been proposed for other functions, such as dividers [16], only approximate adders and approximate multipliers have been considered here. This is because as in other research [9], [17], [18], [19], they are the commonly used operations in a wide range of applications [20]. The mature development of these approximate arithmetic units presents both opportunities and challenges for exploring the design space of approximate circuits.…”

Section: Introductionmentioning

confidence: 87%

“…Utilizing learned knowledge, it avoids the redundant calculations used in previous DSE approaches, allowing rapid identification of high-quality approximate configurations with only minimal DSE effort. We show that it outperforms our recent heuristic search methodology (ENAP) [9], avoiding the multiple iterations to obtain approximate configurations of similar quality. The major contributions of the paper are summarized as follows:…”

Section: Introductionmentioning

confidence: 88%

“…Fig. 2 depicts the exploration process of the approximate configurations for three applications, Sobel, Laplace, and 2x2 matrix multiplication, using the ENAP framework [9]. It presents the relationship between N Err ac and N AL ac where each point represents an approximate configuration.…”

Section: Motivating Case Studymentioning

confidence: 99%

“…It is worth noting that FPAX can be applied to any approximate unit with area and energy consumption values. Mean Error Distance (MED) and Peak signal-to-noise ratio (PSNR) are used to evaluate the fault tolerance of approximate circuits, which are defined in [9]. MED and PSNR are selected as accuracy metrics for this experiment.…”

Section: Design Space Explorationmentioning

confidence: 99%

“…In [9], we proposed the ENAP framework which utilizes the close relationship between the number of approximate units and error constraints to design a number-aware pruning technology. This paradigm further compresses the search space, using an improved genetic algorithm to complete the DSE.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

FPAX: A Fast Prior Knowledge-Based Framework for DSE in Approximate Configurations

Dou,

Wang,

Waris

et al. 2024

IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst.

Self Cite

View full text Add to dashboard Cite

Current artificial intelligence and data science applications typically require complex computations and massive amounts of data handling, presenting unprecedented challenges for embedded platforms. Approximate computing has emerged as the most promising design technique to address this issue, by providing a potential performance increase, while sacrificing accuracy within an acceptable range. Approximate arithmetic units require the creation of design space exploration techniques that can swiftly and automatically form an approximate configuration in fault-tolerant systems. Existing methods, however, use iterative design space sampling, resulting in a large amount of redundant computation. In this work, we propose the efficient FPAX automatic search framework which can learn from prior knowledge regarding the exploration process of known applications and use it to guide design exploration. This avoids excessive redundant computation and quickly provides an impressive approximate configuration. Compared with the Jump Search algorithm known for its efficiency, FPAX can also achieve faster convergence speed and better exploration quality. Even compared to our previous ENAP framework, it exhibits an 18x faster performance while achieving almost identical exploration quality for several commonly used fault-tolerant applications.

show abstract