autoAx: An Automatic Design Space Exploration and Circuit Building Methodology utilizing Libraries of Approximate Components

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F19%3APU132969" target="_blank" >RIV/00216305:26230/19:PU132969 - isvavai.cz</a>

Výsledek na webu

<a href="https://arxiv.org/abs/1902.10807" target="_blank" >https://arxiv.org/abs/1902.10807</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1145/3316781.3317781" target="_blank" >10.1145/3316781.3317781</a>

Jazyk výsledku

angličtina

Název v původním jazyce

autoAx: An Automatic Design Space Exploration and Circuit Building Methodology utilizing Libraries of Approximate Components

Popis výsledku v původním jazyce

Approximate computing is an emerging paradigm for developing highly energy-efficient computing systems such as various accelerators. In the literature, many libraries of elementary approximate circuits have already been proposed to simplify the design process of approximate accelerators. Because these libraries contain from tens to thousands of approximate implementations for a single arithmetic operation it is intractable to find an optimal combination of approximate circuits in the library even for an application consisting of a few operations. An open problem is "how to effectively combine circuits from these libraries to construct complex approximate accelerators''. This paper proposes a novel methodology for searching, selecting and combining the most suitable approximate circuits from a set of available libraries to generate an approximate accelerator for a given application. To enable fast design space generation and exploration, the methodology utilizes machine learning techniques to create  computational models estimating the overall quality of processing and hardware cost without performing full synthesis at the accelerator  level. Using the methodology, we construct hundreds of approximate accelerators (for a Sobel edge detector) showing different but relevant tradeoffs between the quality of processing and hardware cost and identify a corresponding Pareto-frontier. Furthermore, when searching for  approximate implementations of a generic Gaussian filter consisting of 17 arithmetic operations, the proposed approach allows us to identify approximately 10^3 highly relevant implementations from 10^23 possible solutions in a few hours, while the exhaustive search would take four months on a high-end processor.

Název v anglickém jazyce

autoAx: An Automatic Design Space Exploration and Circuit Building Methodology utilizing Libraries of Approximate Components

Popis výsledku anglicky

Approximate computing is an emerging paradigm for developing highly energy-efficient computing systems such as various accelerators. In the literature, many libraries of elementary approximate circuits have already been proposed to simplify the design process of approximate accelerators. Because these libraries contain from tens to thousands of approximate implementations for a single arithmetic operation it is intractable to find an optimal combination of approximate circuits in the library even for an application consisting of a few operations. An open problem is "how to effectively combine circuits from these libraries to construct complex approximate accelerators''. This paper proposes a novel methodology for searching, selecting and combining the most suitable approximate circuits from a set of available libraries to generate an approximate accelerator for a given application. To enable fast design space generation and exploration, the methodology utilizes machine learning techniques to create  computational models estimating the overall quality of processing and hardware cost without performing full synthesis at the accelerator  level. Using the methodology, we construct hundreds of approximate accelerators (for a Sobel edge detector) showing different but relevant tradeoffs between the quality of processing and hardware cost and identify a corresponding Pareto-frontier. Furthermore, when searching for  approximate implementations of a generic Gaussian filter consisting of 17 arithmetic operations, the proposed approach allows us to identify approximately 10^3 highly relevant implementations from 10^23 possible solutions in a few hours, while the exhaustive search would take four months on a high-end processor.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/GA19-10137S" target="_blank" >GA19-10137S: Navrhování a využívání knihoven aproximativních obvodů</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2019

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

The 56th Annual Design Automation Conference 2019 (DAC '19)

ISBN

978-1-4503-6725-7

ISSN

—

e-ISSN

—

Počet stran výsledku

6

Strana od-do

1-6

Název nakladatele

Association for Computing Machinery

Místo vydání

Las Vegas

Místo konání akce

Las Vegas

Datum konání akce

2. 6. 2019

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000482058200123